Multiwavelength Studies of Ultra-soft X-ray AGN

Elizabeth Maria Puchnarewicz

Milliard Space Science Laboratory Department of Physics and Astronomy University College London

A thesis submitted to the University of London for the degree of Doctor of Philosophy May 1992 ProQuest Number: 10608856

All rights reserved

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a com plete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest

ProQuest 10608856

Published by ProQuest LLC(2017). Copyright of the Dissertation is held by the Author.

All rights reserved. This work is protected against unauthorized copying under Title 17, United States C ode Microform Edition © ProQuest LLC.

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106- 1346

A bstract

I investigate the X-ray, optical and infra-red properties of the 53 AGN identified as part of the Einstein Ultra-Soft Survey (USS). Of all sources in the IPC database, these have the strongest, distinct soft X-ray excesses at 0.25 keV.

The IPC data are modelled using two separate components, the ‘soft’ and ‘hard’ X- ray components. Continuum and line parameters are measured from optical spec­ tra while CCD images provide accurate magnitudes and information on the AGN morphology and environment. Infra-red fluxes of the nucleus and the underlying are measured from UKIRT images. Multiwavelength spectra, combining IR, optical, UV and X-ray data, are presented for thirteen AGN.

I find that approximately one third of USS AGN are hard X-ray quiet ; the re­ maining hard X-ray fluxes are typical of other X-ray selected AGN. All optical and IR continuum luminosities are ‘normal’. A striking characteristic of the sam­ ple is the high proportion of narrow-line objects it contains; I also find that the permitted lines are weak. Optical Fell emission is strong where hard X-rays are weak, contradicting hard X-ray dependent models for their production.

All results are considered in the context of two models for the production of a strong soft X-ray flux, accretion disks and the cool clouds model of Guilbert and Rees. The narrow lines suggest that we may be seeing a flattened BLR face-on - alternatively, the BLR may lie further from the black hole than for other AGN. An association of a face-on BLR with an would strongly favour geometrically thick disks over thin. The overall spectral distribution can also be well reproduced by the cool clouds model. 4 0

Contents

List of Figures 10 List o f Tables 13

1. INTRODUCTION 15

1.1. Observational properties of AGN ...... 15 1.1.1 But first; AGN categorization ...... 16 1.1.2 X-ray observations of AGN ...... 18 1.1.3 Ultraviolet ...... 18 1.1.4 Optical ...... 19

1.1.5 Infra-red ...... 2 0

1 .1 .6 Variability ...... 2 1

1.2. The AGN picture ...... 23 1.2.1 A central, ...... 23

1 .2 .2 The primary energy source ...... 23 1.2.3 Beyond the central source ...... 27 1.2.4 The interaction between the central source and cold matter ...... 30 1.2.5 Into the Broad Line Region ...... 32

1 .2 .6 Beyond the BLR ...... 35 1.2.7 A short note about blazars ...... 36

2. THE ULTRA-SOFT SURVEY 37

2.1. Selection criteria ...... 38

2.2. The optical identifications ...... 38

2 .2 .1 Spectroscopic identifications ...... 43 6

2.2.2 Active ...... 43

3. GALACTIC COLUMN DENSITIES 51

3.1. Soft X-ray absorption ...... 52

3.2. Calculating the absorption ...... 53 3.2.1 Elemental abundances ...... 53

3.2.2 Hydrogen column density, N h ...... 53

3.2.3 Stark et al. Nh i distribution of the USS AGN ...... 55

3.3. High resolution N h i measurements ...... 56

3.4. IRAS Infrared cirrus maps ...... 56 3.4.1 IRAS maps of the USS AGN fields ...... 59 3.4.2 Stark versus IRAS ...... 60

3.5. Jodrell Bank Nh i ...... 51

3.5.1 Results from another high resolution 2 1 cm survey ...... 62

3.6. Summary ...... 63

4. ANALYSIS OF THE X-RAY DATA 67

4.1. X-ray spectral models ...... 70 4.1.1 Two-component modelling ...... 70 4.1.2 Model parameter ranges ...... 70

4.2. Finding the best-fit model ...... 73 4.2.1 Fitting the hard X-ray component ...... 73 4.2.2 ...and the soft X-ray component ...... 73 4.2.3 A comparison of the models ...... 73 4.3. X-ray luminosities ...... 76

4.3.1 Soft-to-hard X-ray ratio - defining a ax ...... 77

4.4. Summary ...... 80

5. OPTICAL DATA 81

5.1. Spectroscopy ...... 81 5.1.1 D ata reduction ...... 83 5.1.2 Spectral analysis ...... 83

5.2. C C D Im aging ...... 99 5.2.1 Magnitudes ...... 99 5.2.2 Extended sources ...... 99

5.3. Optical Luminosities...... 100 5.3.1 Optical to X-ray luminosity ratio ...... 104

6. INFRA-RED IMAGES OF USS AGN 105

6.1. IR data reduction and analysis ...... 108

6.2. Multiwavelength spectra ...... 109 6.2.1 Multiwavelength fits ...... 109

6 .2 .2 Results from the multiwavelength fits ...... 115

6.3. Luminosities ...... 118

7. NOTES ON INDIVIDUAL OBJECTS 119

7.1. A Brief Introduction ...... 119 8

7.2. E 0132—411 119

7.2.1 Optical spectrum ...... 1 2 0 7.2.2 UV spectrum ...... 120 7.2.3 Multiwavelength spectrum ...... 121

7.3. The remaining USS AGN ...... 123

8. THE MULTIWAVELENGTH ANALYSIS 139

8.1. Comparison samples ...... 140

8.2. ...... 140

8.3. Luminosities ...... 142 8.3.1 Soft X-ray luminosities ...... 142 8.3.2 Hard X-ray luminosities ...... 145 8.3.3 Optical luminosities ...... 147 8.3.4 Infra-red luminosities ...... 149

8.4. Optical line and continuum properties ...... 155 8.4.1 Line widths ...... 155 8.4.2 Other Line Parameters ...... 161 8.4.3 Optical Fell emission line parameters ...... 165 8.4.4 Optical and IR continuum slopes ...... 168

9. DISCUSSION AND CONCLUSIONS 173

9.1. X-ray and optical continua ...... 174 9.1.1 X-ray and optical continua in an accretion disk model ...... 174 9.1.2 X-ray and optical continua in the cool clouds model ...... 175 C7

9.2. The predominance of narrow Balmer lines ...... 176 9.2.1 A face-on accretion disk ...... 176 9.2.2 ...or a more distant BLR? ...... 177

9.3. The strength of the permitted lines ...... 177 9.3.1 Weak permitted lines in a thick disk model ...... 177 9.3.2 .. and the cool clouds model ...... 178 9.3.3 ...or a more distant BLR? ...... 178 9.3.4 Anisotropic broad line emission ...... 178

9.4. Optical Fell emission ...... 179 9.4.1 A dependence on H (3 FWHM ...... 179 9.4.2 The relationship with the hard X-ray flux ...... 179 9.4.3 Optical Fell and the soft X-ray excess ...... 180

9.5. Infra-red properties ...... 182 9.5.1 Dust emission ...... 182 9.5.2 Soft excesses, narrow lines and warm IRAS Seyferts ...... 182

9.6. The picture of a USS AGN ...... 183 9.6.1 A face-on thick accretion disk ...... 183 9.6.2 Cool clouds around the central source '...... 183

9.7. Conclusion ...... 184

References 185 Acknowledgements 195 10

List of Figures

C hapter 1 Figure 1.1 Multiwavelength spectrum of Mkn 841 ...... 17 Figure 1.2 An artist’s impression of an AGN ...... 26 Figure 1.3 Thick and thin accretion disks ...... 28 Figure 1.4 Ferland and Rees ‘cool clouds’ spectra ...... 30

C hapter 2

Figure 2 .1 USS C1/C2 softness ratio parameter space ...... 39 Figure 2.2 USS selection criteria ...... 42 Figure 2.3 X-ray-to-optical source separations ...... 44 Figure 2.4 Identification confidence of the USS AGN ...... 45 Figure 2.5 Finding charts ...... 47

C hapter 3 Figure 3.1 All-sky N# map with USS AGN positions ...... 51 Figure 3.2 Effects of absorption on soft and hard X-ray luminosity ...... 52 Figure 3.3 Absorption cross-sections ...... 54 Figure 3.4 Unabsorbed X-ray fraction ...... 54 Figure 3.5 USS AGN Stark et al. Nh d istrib u tio n ...... 55 Figure 3.6 Maps of the IRAS cirrus ...... 57

Figure 3.7 A comparison of Stark and IRAS Nh measurements...... 60

Figure 3.8 Radio N h measurements compared with IRAS ...... 62

C hapter 4 Figure 4.1 Two component model reduced x 2 grids ...... 72 Figure 4.2 A comparison of the three X-ray models ...... 75 Figure 4.3 Soft component ratio dependence on N h ...... 76

C hapter 5 Figure 5.1 Optical spectra ...... 85 11

C hapter 6 Figure 6.1 UKIRT K image of E0129-066 ...... 105 Figure 6.2 Mosaic image of E1423+201 ...... 106 Figure 6.3 Mosaic image of E0337-267 ...... 108 Figure 6.4 Multiwavelength spectra ...... I l l Figure 6.5 Nuclear IR/optical spectra ...... 117

C hapter 7 Figure 7.1 Optical spectrum of E0132^411 ...... 120 Figure 7.2 UV spectrum of E0132—411 ...... 121 Figure 7.3 Multiwavelength spectrum of E0132^411 ...... 122 Figure 7.4 Optical spectrum of cooling flow candidate E0436—433 ...... 125 Figure 7.5 Optical spectrum of the double quasar E0957+561 ...... 127 Figure 7.6 Optical spectrum of the HII region galaxy E1218+693 ...... 130 Figure 7.7 CCD image of E1218+693 ...... 130 Figure 7.8 Optical spectrum of the variable AGN E1227+140 ...... 131 Figure 7.9 A ring structure around E1228+123 ...... 132 Figure 7.10 CCD image of E1640+537 - a cluster member? ...... 134 Figure 7.11 A bar or jet in E l640+537 ...... 135 Figure 7.12 Another cooling flow candidate: E1657+326 ...... 136 Figure 7.13 A cluster around E1805+700? ...... 137 Figure 7.14 The close group surrounding El805+700 ...... 138

C hapter 8

Figure 8 .1 USS AGN redshift distribution ...... 141

Figure 8 .2 Soft X-ray dependence on redshift ...... 143 Figure 8.3 Hard X-ray luminosities ...... 145 Figure 8.4 Optical luminosities ...... 146 Figure 8.5 Optical-to-hard X-ray ratio ...... 147

Figure 8 .6 a ox distribution ...... 148 Figure 8.7 IR properties compared with Worrall’s sample ...... 150

Figure 8 .8 IR properties compared with Kriss’ sample ...... 151 Figure 8.9 X-ray-to-IR ratio of the USS AGN ...... 152

Figure 8 .1 0 IR-to-optical ratio of the USS AGN ...... 153

Figure 8 .1 1 FWHM distribution ...... 156 12

Figure 8.12 Dependence of Ha FWHM on a ox ...... 159 Figure 8.13 Dependence of [OIIIJA5007/H/? ratio on H/? FW HM ...... 163 Figure 8.14 Equivalent widths of the USS A G N ...... 164 Figure 8.15 Fell equivalent widths and H/? FWHM ...... 167 Figure 8.16 Optical power-law index ...... 169

C hapter 9 Figure 9.1 Spectrum of E0132-411 compared with cool clouds models 175 List of Tables

C h a p te r 2

Table 2 .1 Einstein Ultra-Soft Survey of AGN ...... 40

C h a p te r 3 Table 3.1 Galactic column densities ...... 64

C h a p te r 4

Table 4.1 X-ray d a ta ...... 6 8 Table 4.2 X-ray luminosities ...... 78

C h a p te r 5 Table 5.1 Optical spectroscopic observations ...... 82 Table 5.2 Optical line parameters ...... 91 Table 5.3 Optical Fell blend parameters ...... 95 Table 5.4 Spectral classifications ...... 97 Table 5.5 Optical CCD observations ...... 100 Table 5.6 Optical luminosities ...... 102

C h a p te r 6 Table 6.1 Infra-red observations ...... 107

Table 6 .2 Results from multiwavelength fits ...... 110 Table 6.3 Optical and IR power-law indices ...... 116 Table 6.4 Infra-red luminosities ...... 118

C h a p te r 8

Table 8 .1 X-ray and optical continuum correlations ...... 144

Table 8 .2 IR correlations ...... 154 Table 8.3 Optical line and continuum correlations ...... 160 14 CHAPTER 1

Introduction

Active Galactic Nuclei are an elusive phenomenon. From the most distant and energetic quasars and blazars to the nearby low-luminosity HII region galaxies; AGN have many guises and forms, confounding attempts to understand these enigmatic objects. Every new insight into their character and behaviour is essential to unravel their mysteries. In this thesis, I have used the sample of Cordova et al. (1992) to examine the multiwavelength properties of AGN from a new perspective; through the presence of a strong ultra-soft (~0.25 keV) excess. A description of the sample itself is

given in Chapter 2 ; this chapter is dedicated to a review and discussion of known properties of AGN and models relevant to the work presented here.

1 . 1 OBSERVATIONAL PROPERTIES OF AGN The multiwavelength spectra of active galaxies exhibit a number of distinctive characteristics in their shape and variability. The spectrum of Mkn 841 from

the infrared to X-rays (shown in Figure 1 .1 and taken from Arnaud et al. 1985) illustrates the spectral features observed in a typical AGN. This section discusses the structure and variability of AGN spectra from the near-infrared to medium X-rays; the range covered in this thesis.

“Some vegetables Ifm fond of; peas I ’m neutral about.”

— JOHN MAJOR 16 Chapter 1

1.1.1 But first; AGN categorization Although attempts are made to explain all AGN with one ‘grand unified model’ (eg. Blandford 1984; Barthel 1989, Vagnetti, Gillongo and Cavaliere 1991), the sub-division of AGN into many categories according to the nature of their ob­ served spectra, is unavoidable. Similarly unavoidable therefore, is the necessity to describe general AGN properties within each class. So I begin this section with a list of types of AGN most relevant to this work and a summary of their properties. quasars axe high luminosity AGN (Mb <-23; Schmidt and Green 1983), originally defined to be stellar in appearance al­ though imaging of the host galaxies of quasars and QSOs is now common. The majority of quasars axe radio-quiet; radio-quiet quasars are also known as ‘quasi-stellar ob­ jects’ (QSOs). They have broad permitted lines with

FWHM typically thousands of km s - 1 and narrow for­ bidden lines (normal galaxies have FWHM of ~ 100 — 200 km s-1 ). Seyfert Is have a stellar-like nucleus and are usually found in spiral galaxies. Like quasars, they have broad permitted lines

with FWHM between ~ 1 0 3- 1 0 4 km s-1, and relatively narrow forbidden lines (~ 300 — 1000 km s-1 ).

S ey fert 2 s are similar to Seyfert Is but the permitted lines have the same widths as the forbidden lines (about 500 km s-1). The flux ratio [OIIIJA5007/H/? > 3. They are believed to be examples of Seyferts where the region of broad line emission (broad line region or BLR) is obscured so that only the narrow line region (NLR) is observed (eg. An- tonucci and Miller 1985).

Narrow-line (NL) Seyfert Is have H/? FWHM less than 2 ,0 0 0 km s- 1 (Os- terbrock and Pogge 1985). The permitted lines, although relatively narrow within the class of Seyfert Is, are still significantly broader then the forbidden lines. Goodrich (1989) has verified that for his sample of NL Seyfert Is, the permitted fines are indeed emitted from the ‘broad’ fine region and distinct from fines which are formed in the NLR. Confusion with Seyfert 2s must be avoided for the remainder of this thesis! BL Lac objects are highly variable, highly polarized radio-loud quasars which show weak or no line emission. Timescales for vari­ ability in the IR and optical are typically the order of days or weeks, shorter in the UV and shorter still in X-rays. Together with optically violent variable quasars (OVVs) these belong to a class of AGN known as ‘blazars’. H II region galaxies have spectra which resemble those of galactic HII regions and are believed to contain large numbers of hot, lumi­ nous OB in their nuclei which ionize the surrounding interstellar gas. The brightest HII region galaxies are also known as ‘starburst galaxies’.

IR optical UV EUV X-ray -10.5 thermal emission? accretion disk or cool clouds?

- 11.0

dust component soft X-ray -11.5 Fell plus Balmer blue bump excess continuum

little blue bump

-12.5 hard X-ray power-law -13.0 12 14 1816 20 Log v

Figure 1.1 : The multiwavelength spectrum of the Seyfert 1 galaxy Mkn 841 (from Arnaud et al. 1985) from the IR to X-rays, illustrating typical spectral features found in non-blazar AGN. The first soft X-ray excess in any AGN was discovered in this object (Arnaud et al. 1985). 18 Chapter 1

1.1.2 X-ray observations of AGN 1.1.2.1 The hard X-ray power-law ... The hard X-ray spectra of AGN can be well fitted by a power-law with an av­ erage ‘canonical’ energy index, a=0.7 (eg. Mushotzky 1984; Turner and Pounds 1989) where F„ = v ~ a (the convention throughout this thesis). Consequently, the success of models for the production of hard X-rays has been largely judged on their ability to reproduce this spectrum. This value for the canonical index

has been measured for objects mostly selected in hard ( > 2 keV) X-rays. However, for a sample of AGN selected on the basis of their soft X-ray flux ( 0 .2-3.5 keV), Comastri et al. (1992) find a steeper average index, a=0.9, a wide distribution of indices (0.4 < a < 1.3) and most objects distributed around o=1.0.

1.1.2.2 ... and the soft X-ray excess The first clear indication of a nuclear soft X-ray component was found by Arnaud et al. (1985) in the EXOSAT spectrum of Mkn 841, a low-redshift (z=0.037) Seyfert

1 galaxy (see Figure 1 .1 ). Since then, soft X-ray excesses have become a familiar feature of many AGN. Turner and Pounds (1989) have reported that, in their sample of 48 hard X-ray selected Seyfert galaxies detected with EXOSAT, 50% of the unobscured objects have soft X-ray ‘excesses’ over a hard X-ray power-law. In many cases, fits to Einstein IPC data of AGN also require an additional ultra-soft emission component (eg. Wilkes and Elvis 1987; Urry et al. 1989; Kruper, Urry and Canizares 1990). Similar results are inferred from the brightness distribution of EXOSAT serendipitous sources (Branduardi-Raymont et al. 1985; Giommi et al. 1991). These ultra-soft excesses have been fitted with steep power-laws of a soft —2 (eg. Branduardi-Raymont et al. 1985) and with blackbody models; fits to combined IPC/MPC data of AGN by Urry et al. (1989) required a blackbody kTeff of less than 1 0 eV or between 200-300 eV. Active galaxies identified as part of the Einstein Ultra-Soft Survey (Cordova et al. 1992), the subject of this thesis, have the steepest ultra-soft X-ray spectra known so far.

1.1.3 Ultraviolet The most dominant feature in the UV spectra of AGN is the ‘big blue bump’ (eg. Edelson and Malkan 1986). The bump rises from A >3000 A and flattens out at ~1500 A. It is well described by a single blackbody with a temperature of Introduction 19

~ 2 —3 x 10 4 K (Malkan and Sargent 1982) and provides the bulk of the bolometric emission from the AGN (see Figure 1 .1 ). The big blue bump turns over somewhere in the unobservable EUV/soft X- ray region and the soft X-ray excess may be its high energy tail. From a study of the UV continua and CIV emission in low- and medium-luminosity QSOs, Zheng (1991) predicts a turnover between 30 and 50 eV. Cheng, Gaskell and Koratkar (1991) investigated the dependence of the UV spectral index, auv ? on redshift and luminosity for a large sample of quasars with 0.09

1.1.4 Optical 1.1.4.1 Continuum For blazars, the optical continuum is smooth and convex (when plotted in logv¥v vs. logv space), but the optical continuum of quasars and Seyferts may have many separate components (see Figure 1 .1 ). In addition to the big blue bump, the ‘3000 A ’ or ‘little blue’ bump is seen in many quasar continua (eg. Neugebauer et al. 1979; Richstone and Schmidt 1980; Neugebauer et al. 1987). Wills, Netzer and Wills (1985) have suggested that this feature is a combination of strong optical Fell blends plus Baimer continuum emission; Neugebauer (1987) confirmed that the little blue bumps of PG quasars are consistent with this. In power-law fits to the optical continua of 718 bright quasars, Francis et al. (1991) find a median index of 0.32 with most lying in the range 1.5 to -1.0, consistent with the results for 114 quasars by Neugebauer et al. (1987; a median of 0 .2 and range of 1 .2 to -0.6). There is also evidence that the nuclear optical continuum is anisotropic and is emitted preferentially along the radio axis (Browne and Wright 1985; Jackson et al. 1989).

1.1.4.2 Optical line emission

HI emission lines in Seyfert Is and quasars are generally broad, ranging from

~500 km s - 1 in Mkn 359 to ~7000 km s - 1 in Mkns 279, 876 and 926, while the forbidden lines are considerably narrower. The FWHM of HI lines in Seyfert 2s have similar widths to the forbidden lines. These are the defining characteristics of these classes of AGN (see Section 1 .1 .1 ). The H(1 line is generally broader than Ha, while HeIA5876 and HeIIA4686 are broader still (Osterbrock and Shuder 1982).

In Seyfert 2 galaxies, the [OIII]AA5007,4959 FWHM range from 2 0 0 - 1 2 0 0 km s- 1 and axe centred at ~400 km s_1; in Seyfert Is the FWHM are generally broader. In both types, the [OIII] lines are asymmetric with a long blue tail and an abrupt redward cut-off. A comprehensive review of the emission line properties of AGN may be found in Osterbrock and Mathews (1986).

Optical Fell emission

By fax the strongest line emitter in the UV/optical, is permitted Fell, which in many AGN contributes between 30-50% of the total emission line flux (Wills, Netzer and Wills 1985). Fell multiplets are usually blended in low-resolution spectra (but see the spectrum of E1511+671 in Figure 5.1 where some Fell lines axe resolved). They may form a ‘quasi-continuum’ if blending is severe and, together with Balmer continuum emission, make up the little blue bump (Wills, Netzer and Wills 1985). Radio-quiet objects have stronger optical Fell than radio-loud objects (Osterbrock 1977; Phillips 1977; Peterson, Foltz and Byard 1981), suggesting a link between Fell emission and X-rays (since radio-loud AGN are usually X-ray loud).

1.1.5 Infra-red

The IR spectra of AGN have many forms, some resemble power-laws while others have broad bumps and lumps (for plots of far-IR-to-UV spectra, see eg. Neugebauer et al. 1987; Ward et al. 1987; Kriss 1988; Barvainis 1990). There are two common characteristics however, a ‘dip’ at lp m and a low-energy cut-off at 100/im. Also, low-luminosity AGN have flatter slopes in the near-IR (Kriss 1988; Barvainis 1990). miroauciion

Originally, power-laws were often adequate to describe mid- and near-IR spec­ tra, thus a non-thermal IR source was assumed. With the advent of IRAS and the extension of IR spectra to the far-IR ( 1 0 0 /xm), it became apparent that a simple power-law is not appropriate for the majority of radio-quiet objects. Recent work has shown that radiation from dust (ie. re-processed UV emission) is likely to be the major contributor to the IR in non-blazar AGN (eg. Ward et al. 1987; Barvai­ nis 1990; Berriman 1990). For example, Barvainis (1990) has suggested that the IR continuum is made up of two, thermal components,!.

1. a convex IR continuum produced by warm dust which is heated directly by quasar UV, and 2. a host galaxy consisting of starlight and cool dust emission. Further evidence in support of the dust model in the IR has been reported by Clavel, Wamsteker and Glass (1989) and Sitko (1989) who found that the near-IR responds to changes in the optical/UV continuum with delays of about one year; the expected distance of near-IR emitting dust grains from the central source is one lt-year (Barvainis 1987). In BL Lac objects, the infra-red to UV continuum is much smoother and the

1 /im dip characteristic of Seyferts and quasars is not seen.

1.1.6 Variability

Variability in AGN is observed on both short timescales ( 1 0 3 -1 0 5 seconds) in X-rays and long timescales (months-years) in the optical/UV. Continued X-ray monitor­ ing of NGC 6814 has shown that a period of 1 2 ,2 0 0 seconds has been stable for six years, although phase coherence is only maintained on timescales of ~ 1 year (Done et al. 1992). Few objects show this kind of variability however, the power spectra of most X-ray variable AGN are featureless with no characteristic timescale. There is variability in the soft and hard X-ray components; they have been seen to vary together (eg. Arnaud et al. 1985; Pounds, Turner and Warwick 1986) and independently (eg. Piro et al. 1988). An anticorrelation in soft and hard X-ray bands, with a pivot at ~ 2 keV, has also been observed (Maraschi et al. 1991). Virtually all Seyfert Is show variability in their optical/UV lines and continua (eg. Peterson et al. 1982, Peterson et al. 1984), in the form of high amplitude flaring

t Note that Barvainis also required only a steep power-law for the optical/UV, ie. there was no additional blackbody component (see Section 1 .3.4.2) 22 Chapter 1 or sinusoidal variation. Using UBVRIJHK photometry, Cutri et al. (1985) found that for most quasars, the amplitude of variability increases towards the blue, and is negligible at 2.2/j.m. Indeed, variability in the IR is hardly seen in non-blazars; for quasars and Seyfert Is, Edelson and Malkan (1987) found no variability in IRAS data (12.5-100/im) on timescales of a few months, while Neugebauer et al. (1989) found no evidence for variability at 10.1/mi in PG quasars on timescales of years.

1.1.6.1 Line reverberation studies

As changes in the central ionizing source occur, they send out a continuum pulse to the line-emitting regions which in turn, produces changes in the emission line profiles. Therefore, by closely monitoring the continuum and flux variations in an AGN and cross-correlating their separate light curves, we are able to ‘map out’ the physical and kinematic structure of the BLR.

I highlight results from a recent campaign to study the Seyfert 1 galaxy, NGC 5548. Line and continuum emission was closely monitored in the UV (Clavel et al. 1991) and the optical (Peterson et al. 1991) over a period of 8 months. These observations show that changes in the UV and optical continua were simultaneous from ~1300 A to 5000 A. The amplitude of the continuum variation decreased as the wavelength increased, ie. the spectrum was ‘bluer’ when it was brighter. The time taken for the lines to respond to changes in the continuum depended on the degree of ionization of the line; it was shortest for the high-ionization lines (HIL) and longer for the low-ionization lines (LIL). The time lag, At, between the continuum and Lya was between 8 and 16 days (Clavel et al. 1991) and 20±3 days between the continuum and R/3 (Peterson et al. 1991). The shortest continuum- line lag was for HeAl640 and NVA1240 (A t=4 - 1 0 days) and the longest (and most loosely constrained) for MgIIA2798 (At=34-72 days). The amplitude of the variation in the lines also depended on the degree of ionization, being high for the HIL and low for the LIL.

1.1.6.2 Variability in blazars

In blazars, variability is faster and stronger, eg. the B magnitude of A0235+164 increased by 2.3 in 1 2 days and AP Lib changed by 0.7 mag in 5 hours (Urry 1988). The shortest reported timescale in the IR/optical is 50 seconds for a change of ±lm ag at 1.25/zm in OJ287 (Wolstencroft, Gilmore and Williams 1982). Edelson Introduction and Malkan (1987) found variations in IRAS data of blazars by up to a factor of two on timescales of a few months. In X-rays, variability timescales are shorter still; the fastest is for a BL Lac object, H0323+022, whose X-ray intensity decreased by a factor >10 in 30 seconds (Doxsey et al. 1983, Feigelson et al. 1986).

1.2 THE AGN PICTURE In this section, I describe the picture of AGN, based on widely accepted models, which is most appropriate to this thesis. It is intended as a source of information relevant to the analysis and interpretation of the USS AGN. It is by no means complete! The illustration begins from the centre of an AGN. An artist’s impression (not drawn to scale) of the inner region of the quasar 3C 273, showing the black hole, accretion disk, broad line region and radio jets, is shown in Figure 1 .2 .

1.2.1 A central, supermassive black hole It is widely believed that at the heart of an lies a super- massive black hole. There is strong evidence for this assumption: • the rapid UV and X-ray variability observed in some AGN suggests a compact central engine which is smaller than the BLR • the mass of a source of this size, derived eg. by comparing the source lumi­

nosity with the Eddington luminosity, is between 1 0 7 M® and 1 0 n M®

• X-ray variability has timescales between 1 0 3 and 105 seconds, corresponding

to the orbital period of a supermassive black hole with a mass of ~ 1 0 6 —

1 0 8 M© • the big blue bump resembles a quasi-thermal spectrum emitted by gas with

a temperature of ~ 1 0 5 K; this gas covers an area roughly the size of the gap between a black hole and the BLR

• radio emission moving out at superluminal (> 8 c) velocities requires a rela- tivistically deep potential well

1.2.2 The primary energy source There are two primary sources of energy in AGN; 1. the release of gravitational binding energy from matter accreting onto the black hole. Angular momentum may be removed by viscous forces (which generate heat and ultimately radiation; Pringle 1981), or in a hydromagnetic 24 Chapter 1

wind due to the interaction with a surrounding magnetic field (Blandford and Payne 1982). 2. by tapping the rotational kinetic energy of the black hole itself (Phinney 1981). Energy is released through the interaction of a strong magnetic field which is generated by currents in the plasma surrounding the black hole (Blandford and Znajek 1977; MacDonald and Thorne 1982).

1.2.2.1 Processing of the primary energy by secondary pairs Next, we move into the region immediately surrounding the black hole, ie. within

a radius of 1 0 15 cm. Assume that there are a large number of localized ‘sites’ within this region where relativistic electrons are accelerated. They cool quickly (via synchrotron emission if there is a strong magnetic field, or via inverse Compton

scattering), generating a non-thermal power-law spectrum which extends to 7 -rays (Guilbert, Fabian and Rees 1983). Electron-positron pairs will be produced from photon-photon collisions and these pairs will then radiate, perhaps producing a cascade of further pairs.

Thus the region becomes optically thick to 7 -rays and the photons we actually observe are due to only mildly relativistic particles (Guilbert, Fabian and Rees 1983). The pairs may be transported away from the central region, giving rise to winds, beams and jets. This forms the basis for non-thermal models of central engines of AGN with electron-positron pair production. Thermal models of pair production have also been proposed (for recent reviews of thermal and non-thermal models see Svens- son 1990; Zdziarski 1992), and models with pairs in which the primary power is supplied to protons (a current review may be found in Zdziarski 1991). Thus it is assumed that reprocessing of the primary energy source by pair- production produces the ‘standard’ non-thermal spectrum that we observe.

26 Chapter 1

Inferred Structure of a Quasar

RADIATION

BLACK 1 ACCRETION DISK h o l e '\

LINE-EMITTING/ CLOUD

Figure 1 .2 : An artist’s impression of the inner regions of an active galactic nucleus (reproduced from Courvoisier and Robson 1991). A black hole lies at the centre and is surrounded (in this example) by an accretion disk. Powerful radio jets extend above and below the black hole. The broad lines are emitted from individual clouds which are scattered around the central engine. Introduction

1.2.3 Beyond the central source Cold, optically thick matter may surround the central compact object and it is widely believed that this would settle into an accretion disk (see Figure 1 .2 ). The cold matter interacts with the non-thermal spectrum further to produce features in the optical-to-X-ray continuum such as the big blue bump. I discuss two models for this component; accretion disks and the ‘cool clouds’ model of Guilbert and Rees (1988).

1.2.3.1 Accretion disks The apparently thermal nature of the big blue bump prompted the suggestion that the black hole is surrounded by an accretion disk (Shields 1978; Malkan and Sargent 1982). Accretion from a disk, rather than eg. spherical accretion, is also preferred because it is a more efficient way for the gas to lose angular momentum before reaching the black hole. If the accreting gas radiates below the Eddington luminosity, L# (ie. the characteristic luminosity at which radiation pressure on free electrons balances gravity), a geometrically thin accretion disk will form. However, if the internal pressure builds up so that the internal sound speed, c8 ~ y/GM/r, then the disk will become geometrically thick (Rees 1984). There are two types of thick tori; 1. radiation supported tori, where radiation pressure competes with gravity, and 2. ion supported tori where the energy dissipated by viscous friction cannot be radiated away by the disk material, so this energy remains as internal energy (R ees 1984 and references therein). Radiation tori form when the Eddington accretion rate, m > a few (where m = M/M# = Me2/L e and M is the mass accretion rate) and are strong emitters in the UV and soft X-rays. Thin disks exist when m < 1 and are also luminous in the UV/soft X-ray region. Thus radiation-supported thick disks and thin disks are both possible sources for the big blue bump component and the soft X-ray excess. (Ion tori need only radiate a small fraction of the gravitational energy released by radiation and are not considered further).

Thin disks

Sun and Malkan (1989) have fitted the IR-to-UV spectra of 60 quasars and AGN using a geometrically thin and optically thick accretion disk model (eg. Shakura and Sunyaev 1973). This disk surrounds a Kerr (rotating) black hole and relativis- The thick disk model The thin disk model 28 I 05 p* 3 o 43 co VhH oo ^ O C O O C oo CL> O C 2 4 T 8 o| <0 0 • g> 'So .22 •.noo m+ •m s s s £ 3 S • C "o «*-C/D no M CQ C M • . H ■a ^ Js •“ § u £ g e .5 “ js iii ii ^ o ,o«n O ON •O a) “ g H £ O®\ 03 c co 3 « t3 O C c,-i c c Q C S O a>£ u B u U 2 3 11

« o® 6

Figure 1.3: Thick and thin accretion disks. Introduction 29 tic effects on the emitted spectrum as a function of inclination, including Doppler boosting, gravitational focussing and gravitational redshift, are taken into account. Their results show that the emitted disk spectrum is effectively shifted to higher energies as the angle between the observer’s line of sight and the axis of the disk (the ‘viewing angle’) increases (see Figure 1.3). Low-redshift Seyfert galaxies are found to accrete at low rates (a few percent of the Eddington luminosity) while quasars accrete close to their Eddington limits.

Accretion rates range from 0 .0 1 to 1 0 0 M q yr -1 and black hole masses from 10 8 to

1010 M q (applicable to face-on disks; masses may increase by up to 1 0 1,2 depending on inclination). These masses agree with the results of Wan del and Petrosian (1988) who assumed a thin disk surrounding a Schwarzschild (non-rotating) black hole for Seyfert Is and quasars. A strong soft X-ray excess of the kind seen in Mkn 335 and PG 1211+143 is produced by this model for edge-on disks with slightly sub-Eddington accretion rate and mass.

Thick disks

In radiation supported tori, a major fraction of the luminosity is emitted from the funnel shaped, inner region of the thick disk, forming a ‘cone’ of radiation (see Figure 1.3). The surface brightness of the funnel is greatly enhanced by multiple scatterings off the funnel walls (Sikora 1981). A soft X-ray excess of the kind seen in AGN will be observed when these systems are viewed face-on (ie. at small viewing angles), and we are looking directly into the radiation cone (Madau 1988). Strong optical/UV excesses (the big blue bump) will be seen in edge-on disks; since low-luminosity quasars have weak bumps (Kriss 1988), this would imply that they are seen preferentially face-on.

1.2.3.2 The ‘cooi clouds’ model

Relatively cool clouds, confined by a magnetic field or hot, intercloud plasma, might exist close to the centre of the active nucleus and perhaps within the central continuum source itself. Reprocessing of the primary non-thermal radiation by these small, dense clouds would result in the photoelectric absorption of X-rays which are then reradiated as thermal UV continuum emission and line features (Guilbert and Rees 1988; Ferland and Rees 1988). This has the effect of lowering the medium X-ray continuum relative to the optical and making the hard X-ray spectrum harder. For instance, for an incident spectrum with an index of 1 over ou

the 0.124-100 keV range, the slope of the emitted spectrum at ~ 2 keV lies between 1 (for no extinction) and -0.5 (in the X-ray opaque limit). The canonical value (0.7) is midway between the upper and lower predicted limits of the slope. This model reproduces both the optical-UV bump, the soft X-ray excess (under certain circumstances) and the slope of the X-ray continuum (Ferland and Rees 1988; see Figure 1.4).

Key

£= -3 -10 £= -5 Mkn 841

-14

-16 14 15 16 17 18 19 Log v

Figure 1.4 : Predicted spectra for the ‘cool clouds’ model (Ferland and Rees 1988) for a density N = 1014 cm-3 and filling factors, e = -5 (dashed line) and -3 (dotted line). These have been normalized to match roughly with the model spectrum of Mkn 841, which is plotted as a solid line (Arnaud et a1. 1985).

1.2.4 The interaction between the central source and cold matter Zdziarski and Coppi (1991) considered the effect of electron-positron pairs in the vicinity of cold matter which is in the form of a disk or cool clouds. They used the non-thermal pair model of Svensson (1987) and Lightman and Zdziarski (1987), with the addition of cold matter (Zdziarski et al. 1990). The compact non-thermal source, which may be located around the axis of the black hole or form a corona above the surface of a disk, processes the primary energy source

(see Section 1 .2 .2 .1 ), while the surrounding cold matter reprocesses the radiation still further. Introduction 31

UV photons emitted by the cold matter are (Compton) upscattered to X- and 7 -rays by electron-positron pairs. Some of the X- and 7 -radiation reaches the observer, some (~50%) is intercepted by cold matter. Photons with energies in the GeV range and higher are reprocessed in the non-thermal compact source to produce secondary pairs. These pairs upscatter UV photons to X/ 7 -ray energies and so the process continues. Most of the hard radiation which reaches the cold matter (~85-90%), is bound-free absorbed and re-radiated in the UV in the form of a blackbody-type spectrum. This would be the big blue bump and/or the soft X-ray excess observed in AGN. A hard X-ray power-law with the canonical 0.7 index and the observed variability in the soft and hard components (see Section 1 .1 .6 ) are also reproduced by this model (Zdziarski 1992).

1.2.4.1 Other models

Other mechanisms suggested for the origin of the hard X-ray spectrum include syn­ chrotron self-Compton models, where photons emitted by the relativistic electrons as part of the synchrotron process are upscattered to X-ray energies (eg. Zdziarski 1986) and inverse Compton scattering of soft X-ray photons by thermal electrons in the two-temperature inner region of an accretion disc. (eg. Shapiro, Lightman and Eardley 1976).

1.2.4.2 Evidence for an optically thin medium

Q2237+0305 is a gravitationally lensed system of which four separate images are seen. Rapid variability has been reported in two of the four images (see Rauch and Blandford 1991 and references therein) implying a source radius < 2 x 1 0 15cm (Wambsganss, Paczyriski and Schneider 1990, Rauch and Blandford 1991). A con­ sideration of the geometrically thin accretion disk model shows that the required disk radius is over three times larger than the derived source size; alternatively, the disk brightness temperature must be ~ 1 0 times the equivalent brightness tem­ perature of a thermal disk (Rauch and Blandford 1991). Therefore, in the case of Q2237+0305, the optical emission must be non-thermal or optically thin. Barvainis (1990) has suggested that optically thin free-free emission dominates in the optical/UV so that there is no separate big blue bump component. The soft X-ray excess in this case would be due to free-free emission that is optically thin to absorption (Antonucci and Barvainis 1988). 32 Chapter 1

Ferland, Korista and Peterson (1990) discuss a geometry whereby the central power-law source is surrounded by ionized partially transparent gas (the ‘very broad line region’ or ‘VBLR’). This VBLR removes 50-500 eV photons and re- radiates them in the the optical/UV, thus producing the big blue bump. However, note that this model is not appropriate for the USS AGN since USS sources emit strongly between 150-560 eV.

1.2.5 Into the Broad Line Region

We now move out to a distance of between 0 .1 and 1 ( ~ 1 0 18 cm) where the broad line-emitting region (the ‘BLR’) lies. Photoionization by the observed continuum is widely believed to be responsible for much of the line emission seen in AGN. Therefore results from fitting observed line ratios to those predicted by photoionization models have drawn the ‘standard’ picture of the BLR. However, recent line reverberation studies (see Section 1 .1 .6 .1 ) may change this picture dramatically.

1.2.5.1 The ‘standard’ model of the BLR

To reproduce the observed broad and narrow line strengths, photoionization mod­ els require small filling factors of ~ 10 ~ 5 for the BLR and ~ 10 “ 4 in the NLR. This led to the idea that the line-emitting regions are formed in clouds or filaments, and are surrounded by either an outflowing wind or perhaps part of the accretion flow. The BLR clouds have densities ~ 1 0 9 — 1 0 11 cm - 3 and the lowest density clouds at the largest radii dominate the emission line spectrum (Rees, Netzer and Ferland 1989). The ‘standard’ model for the BLR (eg. Davidson and Netzer 1979; Kwan and Krolik 1981), is a two-phase medium where the cold (T~ 1 0 4 K) line emit­ ting clouds with densities ~ 1 0 10 cm - 3 are pressure-confined by a hot inter-cloud medium (ICM; T~ 1 0 8 K; although there are problems with this manner of con­ finement - see the following paragraph). Each cloud is divided into two zones. Facing the continuum source is a highly ionized zone which draws its energy from the UV and soft X-ray ionizing continuum in the range 13.6-500 eV and emits high ionization lines (HIL: which include Lya, CIII], CIV Hel, Hell and NV). In the back of the cloud lies the extended ionized zone which draws its energy from the medium X-ray range (0.5-5.0 keV for column densities between ~ 10 22 and 1024 cm-2) and emits low ionization lines (LIL: including most of the Balmer lines, Introduction 33

Mgll, CII and the Fell lines).

Cloud confinement

Recent work (Fabian et ad. 1986; Mathews and Ferland 1987) has shown that in the presence of the type of X-ray, UV and infrared fluxes which are actually observed, the Compton temperature of the confining gas would be lower than required and the BLR clouds would not form. Also, in the Rees, Netzer and Ferland (1989) model of the BLR, the ratio of radiation pressure to total pressure exceeds ~50%

when the BLR cloud density falls below ~ 1 0 9’5 cm-3. Under these conditions, constant pressure clouds which are supported by an external medium become unstable. Clouds may be supported by other means which would not be disrupted by radiation pressure; magnetic fields may confine the clouds as filaments (Rees 1987) or they may be winds or coronae surrounding stars (Penston 1988).

Angular dependence in the standard model

Netzer (1987) considered the angular dependence of the ionizing flux on the emit­ ted spectrum. The shape of the incident spectrum was assumed to be an isotropic, hard X-ray power-law plus an angular-dependent UV continuum which is brightest at small angles of inclination (ie. face-on). The results showed that standard BLR clouds at small angles produce strong Balmer continuum emission and HeIA5876 whereas low-excitation lines such as MgIIA2798 and Fell, are strongest at large angles. Notably, strong Balmer emission and Fell emission did not originate from the same region.

1.2.5.2 A two-component model Collin-Souffrin et al. (1988) discuss a two-component broad emission line model for AGN. Instead of being produced within the same cloud, the HIL are produced in a region distinct from that which produces the LIL. The HIL are produced in clouds of cooling gas behind shocks in a hypersonic flow of interstellar matter (Perry and Dyson 1985) whereas the LIL are formed in the outer parts of the accretion disc by hard X-rays which are scattered back into the disc from gas behind the shocks in the flow (see also Dumont and Collin-Souffrin 1990).

1.2.5.3 The results of line reverberation studies Important implications for the structure of the BLR (at least in variable AGN) have been made by recent line reverberation studies (see Section 1 .1 .6 .1 ). On the strength of UV (Clavel et ai. 1991), EXOSAT (Turner and Pounds 1989) and opti­ cal (Peterson et ai. 1991) observations of NGC 5548, Krolik et ai. (1991) proposed that the BLR of NGC 5548 has two zones with different physical conditions.

HIZ Stretching from ~4-14 lt-day from the centre lies the inner, high pressure

(n T ~ 1 0 15 K cm-3), high ionization zone (HIZ) which has a roughly spherical distribution. There is no evidence for strong Balmer or MgIIA2798 emission from this zone so there is no need for the standard model-type, optically thick clouds (eg. Kwan and Krolik 1981) in this region. The bulk of the line luminosity (excluding Fell) is emitted from the HIZ. LIZ Beyond this and at least ~20-30 lt-day from the central source is the low ionization zone (LIZ) which is probably flattened. A possible structure for

the LIZ is an annulus with a radius of ~ 1 0 0 lt-day (which, in the case of NGC 5548, is approximately edge-on to the line of sight). The LIZ is ~500 times more massive than the HIZ but has a smaller covering factor (~0.3 for the HIZ compared to ~ 0.05 for the LIZ). By making relatively direct measurements of the distance from the central source to the BLR, this work reveals that the BLR is considerably closer to the ionizing source than indicated by photoionization models. The effects of the more intense radiation field on the BLR has been investigated by Ferland et ai. 1992. A signifi­ cant finding was that nearly all emission lines are optically thick and line emission is beamed back towards the central source. They also found that the BLR clouds may well be optically thick to electron scattering; a line-broadening mechanism previously believed to be unimportant. o^vt/st Krolik et al. (1991) presented evidence of J correlation between FWHM of the broad lines and the time taken for the line to respond to UV and optical continuum variations, indicating that lines with higher velocity widths are produced closer to the ionizing source. The distribution of the line widths is consistent with *circular motion or randomly directed orbits in the gravitational field of a black hole (this implies a mass of ~ 2.7 x 1 0 7 M q for circular motion and a mass three times greater for random motions).

FWHM - a measure of viewing angle or BLR distance?

Assuming that the BLR has a flattened geometry and that the FWHM of the broad lines describes the velocity of the line-emitting material along the line of Introduction 35 sight, then broad line FWHM may be used as a measurement of the viewing angle to the BLR (and, if it is co-planar, the accretion disk). However, the results of Krolik et ai., ie. that line widths in a single object are a measure of the distance to that line-emitting region, suggests that for AGN in general the FWHM of the permitted lines may be a measurement of the distance from the central source to the BLR.

1.2.5.4 The optical Fell emission region

Finally, we consider the production of the optical Fell blends. • In the standard BLR model, hard X-rays (>800 eV; Krolik and Kallman 1988) penetrate deep into the broad line clouds and create a warm, partially ionized zone at high optical depths where low ionization lines, including Fell and the Balmer lines, are produced (Kwan and Krolik 1981). • In the two-component model of Collin-Souffrin et al. (1988) these lines are formed in the outer parts of the accretion disk by hard X-rays which are scattered back into the disk. • In the angular-dependent model of Netzer (1987), strong Fell and Balmer emission do not originate in the same region: Fell lines are produced in the plane of the accretion disk and the Balmer continuum above the poles.

All of these models (described in Section 1 .2.5.2) predict that the strength of the optical Fell emission increases with the hard X-ray flux. In an alternative scheme proposed by Joly (1991) and based on the Norman and Miley (1984) model, strong optical Fell (relative to H/3) is emitted in the interaction layer between the jets and the region through which they propagate. This region, which is in collisional equilibrium, demands a very weak or no external radiation field; therefore it must be shielded from a power-law continuum (Joly 1987).

1.2.6 Beyond the BLR

Between tens and hundreds of from the central source, lines are emitted from gas with a relatively small velocity dispersion - this is the Narrow Line Re­ gion (NLR). Antonucci and Miller (1985), in polarization studies of NGC 1068, have shown that the gap between the broad and narrow line regions is filled with a geometrically and optically thick torus. It has been suggested that this is made up of a large number of very optically thick, dusty molecular clouds (Krolik and 36 Chapter 1

Begelman 1986, 1988) while the torus’ hole is filled with a warm, ionized plasma which reflects the nuclear photons (Antonucci and Miller 1985). The torus colli­ mates ionizing photons so that the only narrow-line clouds which are photoionized are those which lie along the axis of the torus.

Seyfert 2 galaxies are thus believed to be Seyfert Is viewed along the plane of this torus, so that the BLR is obscured and we only see lines emitted from the

NLR. We would not expect to see a Seyfert 2 galaxy with a strong soft X-ray excess therefore, since the absorbing column would be prohibitively high along the line of sight (but see the section on E0141+020 in Chapter 7). Another explanation for

Seyfert 2 galaxies is that they are ‘normal’ Seyfert 1 nuclei in which the ionizing and X-ray continua are in a low state. Indeed, the BLRs of some AGN have been observed to vary sufficiently that their emission line spectra have changed from a

Seyfert 1 to a Seyfert 2 (Penston and Perez 1984). The dust, which reprocesses the optical/UV radiation from the inner regions and re-radiates it in the IR, may also lie in this region. For moderately luminous quasars, the dust grains will evaporate within a radius of ~ 1 parsec, thus dust grains we observe in the IR probably lie beyond the BLR (Barvainis 1987). The geometry of the dust is not known, but eg. it may lie in the obscuring torus or it may be spherically distributed with the NLR (at least for the far-IR emitting grains; Barvainis 1987).

1.2.7 A short note about blazars I end this chapter with a brief description of blazars. Their rapid variability and high polarization has led to the suggestion that we see radiation relativistically beamed towards us, perhaps in a jet analogous to the radio jets seen in radio galaxies and quasars. The subject of blazars and their rather extreme behaviour is fascinating in itself, but unfortunately, a detailed account is beyond the scope of this thesis. However, a good description of blazars may be found in Urry (1988). CHAPTER 2

The Ultra-Soft Survey

The Einstein U ltra-Soft Survey (USS) was originally designed to select hot, iso­ lated neutron stars from the Einstein Observatory database. All sources detected with the Imaging Proportional Counter (IPC), including the targets of observation, were searched for evidence of a distinct soft X-ray component with an ‘effective temperature’ of the order of 1 0 eV. Such X-ray spectra are similar to those of hot white dwarfs, polar cataclysmic variables and some nearby stars. However, due to the poor spectral resolution of the IPC and the poor statisti­ cal quality of most of the data, it was not appropriate to search for spectra of this kind using the full range of PHA channels. Instead, counts were binned up into three X-ray colours, Cl=0.16-0.56 keV, C2=0.56-1.08 keV and C3=1.08-3.5 keV, and selection for the USS was based on the relative distribution of counts in these bands. More details of the USS may be found in Cordova et ai. (1992; hereafter C92) and references therein. The USS AGN are listed in Table 2.1.

“One word sums up probably the responsibility of any vice pres­ ident, and that one word is (to be prepared’.”

— DAN QUAYLE 38 Chapter 2

2.1 SELECTION CRITERIA A soft X-ray component which has, eg. a blackbody effective temperature (kTeff) of ~ 1 0 eV, only falls in the Cl and C2 ranges, therefore selection was based on the ratio of counts in these two lower energy bands. Figure 2 .1 shows how the C1/C2 ratio changes with the soft X-ray component model (ie. blackbody kTeff or power- law index, a - see Section 4.2.2 for definitions of these) and neutral hydrogen absorbing column, Nj j. These grids represent ratios for a spectrum which has a soft component only, ie. no contribution from any hard component has been included in Cl or C2.

For inclusion in the USS, a source must have a ratio C1/C2 > 2 .8 (this ratio is indicated by the solid line in Figure 2 .1 ). There is no restriction on the counts in C3. The observed ‘shape’ of a USS spectrum is represented in Figure 2.2. In addition to the ratio criterion, all USS objects must have S/N > 3 in the standard

IPC ‘soft band’ ( 0 .2 to 0.8 keV) and the standard ‘broad band’ (0.2 to 4 keV). USS sources are divided into two categories, ‘secure’ and ‘non-secure’, ac­ cording to their statistical significance. ‘Secure’ USS sources are those for which

(Rl+lcr) < 0 .6 , and the region in which such a source is detected must be free of shadowing by the detector support ribs (which can potentially distort the apparent energy distribution of source counts - see C92). Remaining objects are classified as ‘non-secure’.

There are approximately 7700 sources in the IPC database with a S/N > 3. 230 meet the USS criteria. 136 of these are classified as secure.

2.2 THE OPTICAL IDENTIFICATIONS To date, 165 USS sources (72%) have been identified in catalogue searches and the optical spectroscopic identification programme. Ultra-soft X-ray emitting objects include the white dwarf Sirius B, Procyon, U Gem and AG Dra in outburst, a Cen, , the polar cataclysmic variables DP Leo and QQ Vul and the Cygnus Loop. Amongst the extragalactic objects are clusters of galaxies, normal galaxies and AGN (Thompson 1991; private communication). 6

5

oB © 4 o <4-1o 3

2

1

10 20 30 40 50 60 70 80 90 100

Blackbody kT(e ff

'c?

-o 10 987654321 Power law energy index

Figure 2.1 : The C1/C2 softness ratio as a function of absorbing column (N/f) and (top) effective temperature of a blackbody (Teff) and (bottom) power-law energy index. The solid line represents a ratio of 2.8; the boundary for USS selection. 40 Chapter 2

Table 2.1 : Einstein Ultra-Soft Survey of AGN

0) (h) (Hi) (iv) (Vi) J2000 Separation V magnitude Catalogue Object RA Dec £(RA) 6(Dec) m y Note z cross-ref

SECURE

E0111-015 01 14 27.4 -01 16 34 -15 7 17.70 (4) 0.12 HB, EMSS E0114-016 01 17 10.1 -01 24 39 30 -2 19.29 (4) 0.34 EMSS E0131-408 01 33 51.8 -40 39 06 23 -53 19.8 (3) 2.36 HB E0132-411* 01 34 57.6 -40 56 23 0 -19 17.47 (5) 0.27 E0136-250 01 38 43.1 -24 50 33 41 -3 17.8 (4) 0.31 EMSS E0310-557 03 11 48.3 -55 32 37 9 -28 17.0: (4) 0.23 EMSS E0331-365 03 33 12.8 -36 19 48 0 16 18.0 (4) 0.31 EMSS E0845+378 08 48 19.3 +37 40 05 -12 -40 18.03 (4) 0.31 HB, EMSS E0944+464* 09 47 18.1 +46 15 09 -10 3 18.6 (2) 0.35 E0957+561* 10 01 27.8 +55 53 28 67 -27 16.7 (3) 1.41 HB E1028+310* 10 31 38.9 +30 46 46 26 -32 19.50 (1) 0.25 E1040+123 10 42 44.7 + 12 03 31 0 -6 17.29 (3) 1.03 HB E1146+558* 11 48 52.6 +55 36 02 25 -24 19.61 (1) 0.44 E1208+322 12 10 37.7 +31 57 05 -13 -14 16.71 (6) 0.39 HB E1227+140** 12 29 34.2 +13 46 29 29 -13 17.3 (2) 0.10 HB E1251-005 12 53 37.3 -00 48 11 30 -30 18.45 (4) 0.43 EMSS E1254+221 12 56 59.0 +21 53 40 14 4 16.98 (4) 0.19 EMSS E1255+220 12 57 49.1 +21 44 40 0 -1 18.50 (4) 0.26: EMSS E1255+017 12 58 21.5 +01 31 57 30 16 18.43 (4) 0.16 EMSS E1334+038 13 37 09.8 +03 35 55 15 -29 17.72 (4) 0.14 EMSS E1346+266* 13 48 34.5 +26 22 07 13 -1 18.86 (10) 0.92 E1401+098 14 04 10.7 +09 37 45 -30 18 16.6 (8) 0.43 HB E1423+201* 14 26 13.4 + 19 55 24 14 -4 16.25 (10) 0.21 E1425+169* 14 27 44.9 +16 41 11 0 -15 17.79 (10) 0.22 E1519+279 15 21 30.5 +27 44 19 0 12 18.2 (7) 0.23 HB E1614+308 16 16 53.4 +30 45 27 13 1 18.01 (4) 0.27 EMSS E1704+608 17 04 41.5 +60 44 30 29 -23 15.28 (3) 0.37 HB E2034-228 20 37 27.4 -22 42 48 -42 -2 17.8 (4) 0.26 EMSS E2318-423 23 21 00.5 -42 03 31 22 5 18.2 (4) 0.21 HB, EMSS Table 2.1 : Einstein Ultra-Soft Survey of AGN

(0 (n) (in) (iv) M (vi) J2000 Separation V magnitude Catalogue Object RA Dec S{RA) 8{Dec) m y Note z cross-ref

NON-SECURE E0007-357 00 10 20.8 -35 26 58 24 -20 16.0 (4) 0.05 EMSS E0039-019* 00 42 20.0 -01 41 52 0 17 16.1 (2) 0.35 EO114-002* 01 17 03.8 +00 00 27 60 -15 15.9 (2) 0.04 E0129-066* 01 32 16.7 -06 25 21 30 -28 17.1 (2) 0.22 E0141+020 01 43 57.8 +02 20 59 15 11 14.16 (9) 0.02 E0150-102 01 53 24.1 -10 00 35 15 -7 18.1 (3) 0.36 HB E0200-089 02 03 26.1 -08 43 49 30 8 16.52 (4) 0.77 EMSS E0337-267* 03 39 13.5 -26 36 48 13 -33 15.5 (2) 0.11 E0436-433* 04 38 00.8 -43 15 47 44 -13 16.0 (2) 0.07 E0844+377* 08 47 16.0 + 37 32 18 12 4 17.3 (7) 0.45 HB E0952+442 09 55 29.2 +43 57 34 -11 -16 17.28 (4) 0.47 HB, EMSS E1008+348 10 11 50.0 + 34 37 56 12 4 17.62 (8) 0.14 E1011+496 10 15 04.3 +49 25 59 29 -41 16.15 (3) 0.20: HB E1059+730 11 02 37.3 +72 46 38 31 0 16.32 (4) 0.09 HB, EMSS E1215+692* 12 18 14.0 +68 59 59 -5 6 19.78 (1) 1.12 E1217+695* 12 19 25.2 +69 13 43 11 -29 17.5 (7) 0.63 HB E1218+693* 12 20 38.2 +69 05 04 -11 -26 17.69 (1) 0.11 E1352+183 13 54 35.8 +18 05 17 0 -6 16.41 (10) 0.15 HB E1423+242 14 25 50.8 + 24 04 02 -27 -20 17.2 (3) 0.65 HB E1511+671* 15 12 24.8 +66 56 42 -12 -2 17.7 (2) 0.31 E1640+537* 16 42 00.7 +53 39 51 0 -10 18.1 (1) 0.14 E l644-029 16 46 48.4 -03 04 14 0 -23 17.91 (4) 0.26 EMSS E1657+326* 16 59 00.7 + 32 37 18 25 1 17.50 (1) 0.09 E1805+700* 18 05 18.1 +70 06 21 -15 -2 18.49 (1) 0.19

NON-USS

E0906+111* 09 09 00.5 +10 59 34 -15 6 16.93 (4) 0.16 EMSS E1213+378* 12 15 29.9 +37 32 28 0 10 18.62 (1) 0.82 E1228+123* 12 31 13.2 +12 03 07 15 -10 17.10 (3) -■ 0.12 HB E1304+342* 13 06 25.1 + 34 01 35 0 -18 17.97 (4) 0.28 HB, EMSS E1654+352* 16 56 14.1 +35 10 14 25 21 17.58 (1) 0.80 42 Chapter 2

Notes to Table 2.1 (i) Object name. (ii) Optical J2000 co-ordinates. (ni) X-ray to optical separation in arcseconds (see Table 4.1 for X-ray co-ordinates). (iv) V band magnitude and its derivation (this thesis) or reference. (v) Redshift. (vi) Catalogue source for the identification.

: Value uncertain. * Optical spectrum obtained (see Chapter 5) ** Two USS spectra for this object, one secure and one non-USS. Optical spectrum obtained. Notes to optical magnitudes. (1) Calculated from optical spectrum and scaled to CCD magnitudes to compensate for light lost in the slit (see Chapter 5). (2) Calculated from optical spectrum and flux increased by a factor of 1.82 to compensate for light lost through the slit (see Chapter 5). (3) Hewitt and Burbidge (1987) . (4) Stocke et a1. (1991). (5) Calculated from optical spectrum: as this spectrum was taken with a wide slit, no correction to the flux was made (see Section 6). (6) Moles et ai. (1985). (7) Margon, Downes and Chanan (1985). (8) Kriss and Canizares (1982). (9) de Ruiter and Lub (1986). (10) Howell (private communication) from CCD photometry taken 1991 June 27. Errors are ±0.03.

4

3

2

1 \ 7

C2 0 C3 0.16 0.56 1.08 3.50 Energy (kev)

Figure 2.2 : USS selection requirements for the relative dis­ tribution of counts in the Cl, C2 and C3 bands. C1/C2 must be greater than 2.8 while C3 can take any value. lh e Ultra.-bolt burvey

2.2.1 Spectroscopic identifications As the first step in the optical identification procedure, all the objects near the source’s X-ray position on red and blue Palomar Observatory Sky Survey (POSS) plates in the north and SERC J and ESO B plates in the south were examined. Objects which showed unusual colours (eg. a strong blue excess) were examined first at the telescope. If a positive identification was made but there were other unusual objects in the error circle, then these were also examined. If there was no positive identification or if there were no unusual objects within the error circle, then all objects were examined in and around the X-ray error box, until a plausible counterpart of the soft X-ray source was found.

2.2.2 Active galaxies The discovery of so many active galaxies was an unexpected surprise. Twenty-two USS AGN were independently spectroscopically identified during observing runs at the William Herschel and Anglo-Australian Telescopes (see Chapter 5). Five of these ‘identifications’ were confirmed when the Optical Catalogue of QSO’s (Hewitt and Burbidge 1987; hereafter HB) and the Einstein Extended Medium Sensitivity Survey (hereafter EMSS; Gioia et ai. 1990; Stocke et ai. 1991 and Mac- cacaro et ai. 1992) were subsequently searched for AGN whose positions lay within the X-ray error circles of the USS sources. A further 31 USS AGN identifications were obtained from these catalogue searches. E0132—411 had been previously op­ tically identified by Kriss (1982; although it had not been recognised by him as a soft-spectrum object).

Details of the optically identified AGN are presented in Table 2 .1 (the X- ray data are given in Table 4.1). The Table is in three sections; the first section contains the secure USS AGN and the second the non-secure sources. The third section contains sources on the borderline which fail to satisfy the present USS criteria.f These non-USS sources are listed for information only as possible ultra- soft AGN. Measurements of optical and X-ray parameters have been made as for the secure and non-secure sources, but these have not been considered in the analysis and discussion of the results.

f Since Cordova and Kartje (1989), minor revisions have been made to the criteria for inclusion in the USS; see C92 for the details of these changes. 44 Chapter 2

2.2.2.1 Identification confidence

GO c 0o 00D

1I K X

8 © t o

-3 0

-6 0

-6 0 0 0 30 6 0-3 8 (RA) - arcseconds

Figure 2.3 : The positions of the optically identified AGN relative to the position of the X-ray source. The separations, £(RA) and <$(Dec), are in arcseconds. The cross, plotted at the origin, represents the X-ray position. Secure USS sources are indicated by boxes around the asterisks; all other objects are non-secure. The solid circle represents a distance of 30" from the X-ray position and the dashed circle a distance of 60".

I have plotted the relative spatial separations of the optically identified AGN from the USS X-ray source in Figure 2.3. The X-ray position lies at the origin. The solid and dashed lines represent distances from the X-ray to the optical position of 30" and 60" respectively. The diagram shows that all but two of the AGN lie within V and almost 50% within 30". The two that lie outside are E0957-f561; a double quasar which has extended soft X-ray emission and E0114-002; a low redshift

(z=0.04) Seyfert 1 , newly identified in the optical identification programme (see lhe Ultra-bolt burvey 45

0.04

Key

0.03 simulation c USS AGN •B ja • f i | 0.02 .a T3 E 'Z, 0.01

0.00 0 50 100 150 Distance from X-ray position (arcseconds)

Figure 2.4 : The distribution of separations of the USS (X-ray) source from the op­ tically identified AGN (histogram). Also plotted as a isolid curve, is the cumulative function which describes the probability that the nearest field would lie at a given distance from the USS source by chance.

Chapter 7 for details on individual objects). I have tested the association of USS sources with AGN by comparing the dis­ tribution of USS AGN optical-to-X-ray separations with a simulated distribution for any random field star which may lie as close to the X-ray source. I measured the star density in each USS field and performed a Monte Carlo simulation to determine the probability of finding a field star at a given distance from the X-ray source. The cumulative function, summed over all the identified AGN fields, which describes the probability that the nearest field star would lie at a given distance from the USS source by chance is shown by the solid curve in Figure 2 .4 . The his­ togram shows how the separation of the USS source from the AGN is distributed. The USS AGN distribution is significantly more peaked to lower separations than expected if they were just chance associations with objects randomly dis­ tributed on the sky. Formally, there is a probability of <1% that the two distri­ butions are drawn from the same parent population. Thus, even without taking into account the low surface density of AGN at this magnitude compared to stars, 46 Chapter 2

I can demonstrate that the association of USS sources with AGN is more likely than with any other type of star in the field. Thus, statistically, ultra-soft X-ray emission is clearly a property of this AGN sample.

2.2.2.2 Finding Charts In Figure 2.5, the finding charts for all of the USS sources are presented, except those only listed in the EMSS which may be found in Maccacaro et ai. (1992). These finding charts were made using facilities kindly made available to us by the Space Telescope Science Institute in Baltimore. The AGN is identified by the number 1 on the finding chart, except for E0957+561 which is a double (gravi- tationally lensed) quasar; in this case the two components are labelled 1 and 2 . Other labelled objects were examined while searching for the optical counterpart, either spectroscopically in the case of stars, or visually using the WHT TV camera for the two galaxies in the fields of E1425+169 and E1805+700. The numbers on the charts correspond to the ‘Star No.’ in Table 5.1.

Figure 2.5: Opposite and on the following page - finding charts for all of the USS AGN except those only listed in the EMSS. Each chart is 5/ X 5/ and is centred on the USS X-ray position. Charts for 5 non-USS AGN are included (see Table 2.1 for the list of these objects). The Ultra-Soft Survey

E0039 - 019 E0111 -015 . E0114 -002 E0129 - 066

1 1 \ >3 — • • — 11 ’ I - ' 1 1 : T5 •

E0131 - 408 E0132- 411 E0150- 102 E0337 - 267

1 I 2 ! Ql'M 11

E0436 - 433 E0844 + 377 E0845 + 378 E0906 + 111

12

11 1 1 1 1

E0944 + 464 E0952 + 442’ E0957 + 561 E1011 + 496

ij : Tr • ’ *2 n li • f • #

E1028 + 310 E1040+123 E1059 + 730 E1146 + 558

1 I 1 I 1 1 11

E1208 + 322 E1213 + 378 E1215 + 692 E1217 + 695

The Ultra-Soft Survey 49

E1218 + 693 E1227+140 E1228 + 123

1 I . 2 1 . T* 31 t l

E1304 + 342 E1346 + 266 E1352 + 183

'• ,, ‘ » ' 'V ' • i t . 1 1 T« • r .

■ ■ ■

E1401 + 098 E1423 + 201 E1423 + 242

— # 1 I ■M.: 1 l • 12

... • . .

E1425+169 E1511 . E1519 + 279

I 2 1 I • — 1 I it - I 2 .T. , < ' • . - v %

E1640 + 537 E1654 + 352 E1657 + 326 • *

• ' 1.1 . 11 * It

. Ik ♦

E1704 + 608 E1805 + 700 "* E2318 - 423 liir* ’7 ■ >; ttaiKv' 2Y f I Ti '3

CHAPTER 3

Galactic Column Densities

Figure 3.1 : The all-sky N h i map from Stark et a1. 1992. The map is plotted in n h h celestial co-ordinates with 90 North at the top and RA from 0 to 24 left to right. The positions of the secure (squares) and non-secure (crosses) USS AGN are seen to lie in the lowest galactic column regions of the sky.

“I f w e do not succeed, then we run the risk of failure.”

— DAN QUAYLE 52 Chapter 3

3.1 SOFT X-RAY ABSORPTION Before calculating the intrinsic luminosity of any distant object in the Universe, the observed flux must be corrected for the effects of absorption along the line of sight. This is especially true in the soft X-ray region where the absorption cross- section,

AGN. Both components are integrated over 0 .2 keV to 4.5 keV; the soft component is represented by a 10 eV blackbody and the hard component by a 0.7 power-law. Figure 3.2 illustrates the large uncertainty in the derived soft X-ray luminosity which may be incurred due to errors in the N eg. just a 1 0 % error on N h at

~ 2 x 1 0 20 cm - 2 corresponds to a 30% error on the luminosity at 0.2 keV. The hard component luminosity shows little dependence on Nh in comparison.

Soft component

Hard component - >a> »oM 2 42 OO o

0 1 2 3 4 5 6 Nh (units of 10” cm'2)

Figure 3.2 : Models of the soft and hard X-ray luminosity plotted as a function of N h f°r a typical USS AGN. Galactic Column Densities 53

3.2 CALCULATING THE ABSORPTION Figure 3.2 demonstrates that any errors in the amount of absorption will have a large effect on the derived soft component luminosity. It is essential therefore, to calculate the absorption as accurately as possible and to consider carefully the errors involved. Within the bandpass of the IPC, photoelectric absorption by heavy elements dominates, but below the carbon edge at 0.28 keV, helium contributes most to the absorption. For a given hydrogen column density, Nh? and emitted photon energy, E, the flux which reaches the observer is reduced by a factor ea^ Nn, due to the intervening matter along the line of sight, (where a (E ) is the total absorption cross-section). The value for Nh is usually derived from measurements of the HI Galactic 21 cm emission and a is calculated by assuming helium and heavy element abundances relative to hydrogen.

3.2.1 Elemental abundances The elemental abundances assumed in this thesis are taken from Morrison and McCammon (1983) which are based on solar and meteoritic values. Uncertainties on the abundances of elements which are important for absorption in the IPC range are relatively high, eg. ~ 1 0 % for carbon, nitrogen and oxygen; ~30% for neon. Uncertainties on the helium abundance, important below 0.28 keV, are ~l-2%. Also, because they are based on local measurements, the heavy element abundances may only apply in the solar neighbourhood. The absorption cross-section for the IPC range is plotted as a function of energy in Figure 3.3 using the coefficients from Morrison and McCammon (1983). The diagram shows a increasing as the photon energy softens.

3.2.2 Hydrogen column density, N h

The fraction of the emitted radiation which reaches the observer is shown in Fig­ ure 3.4 for values of Nh from lxlO 20 cm - 2 to 6 x l0 20 cm-2, plotted as a function of energy. For an Nh greater than 3 x 1 0 20 cm-2, at energies less than ~ 0.4 keV, more than 50% of photons have been absorbed; at energies less than ~ 0.25 keV, more than 90% have been absorbed. The presence of any ionized hydrogen is neglected when using measurements of the neutral hydrogen emission for N h - However, the errors incurred may be small: while HII may constitute ~27% of the total hydrogen column density in 54 Chapter 3

H+He

Fe-L D Ne •aB CO8 I . CO CO S U /x

>o- 0.160.56 1.08 3.50 Photon Energy (keV)

Figure 3.3 : Net photoelectric absorption cross-section per hydrogen atom as a function of energy using coefficients from Morrison and McCammon (1983). The energy range covers the bandpass of the IPC; boundaries of the Cl, C2 and C3 bands are shown on the x-axis. Absorption edges are labelled.

1.0 0.8 N„=lxl0‘

0.6 0.5 aB 2 0.4 Uh a"S 0.3-a COo •8 § 0.2

0.1 0.16 0.56 1.08 3.50 Photon Energy (keV)

Figure 3.4 : The unabsorbed fraction of the emitted radiation plotted as a function of energy (using coefficients from Morrison and McCammon 1983) for values of N /f from 1X1020 cm-2 to 6X1020 cm-2 . The energy range covers the bandpass of the IPC and the ranges of the Cl, C2 and C3 bands are shown on the x-axis. Galactic Golumn JJensities 55 the Galactic disk (Liszt 1983), atomic hydrogen has been shown to dominate at high Galactic latitudes (Blitz, Magnani and Mundy 1984), where most of the USS AGN lie. Therefore, I have assumed throughout that the contribution to N h from ionized and molecular hydrogen is negligible, ie. N h =N h i *

3.2.2.1 Measurements of Nh i The quantity of the intervening hydrogen is usually derived from measurements of the Galactic 21 cm emission. The all-sky survey of the 21 cm flux by Stark et al. (1992) is generally used to determine the hydrogen column density for extragalactic objects. This survey is largely uncontaminated by stray radiation and therefore provides a good estimate of the HI emission over a 2 ° xZ° region (Lockman, Jahoda and McCammon 1986).

3.2.3 Stark et al. N h i distribution of the USS AGN The distribution of the USS AGN Stark Nh is plotted in Figure 3.5, which shows that most USS AGN lie in regions where the average Nh is between 1 and 3 x l0 20 cm-2. The positions of the USS AGN are also shown on an all-sky map of the Stark 21 cm emission in Figure 3.1.

15

£ 10 O < o ol-l x> Ba 55 5

0 0 2 4 6 8 10 Stark NH (units of 10 20 cm'2)

Figure 3.5 : Distribution of Stark et al. column density for the USS AGN (dotted line). The solid line isolates the secure sources. 56 Chapter 3

These results suggest that the USS AGN are found in the lowest column regions of the sky observed by Einstein. Only 0.9% of the sky at 8 > —40° has

an Nh i < 1 x 10 20 cm - 2 and < 0.1% has an Nh i < 7 x 10 19 cm - 2 (based on an analysis of the 22,000 HI spectra in the Stark survey by Lockman, Jahoda and

McCammon 1986). Einstein observed only 2 % of the whole sky, and there is only one pointing in the 8 lowest column regions identified by Lockman, Jahoda and McCammon (1986; this was sequence no. 7764 which lies within the region centred on 1 0 h 45m 00s, +57° 2 0 ' 0 0 ", and there is no identified USS AGN in this IPC image). The Lockman et al. regions were mapped at a resolution of 2 1 ' with the 140 ft NRAO telescope at Green Bank. If AGN with USS-type soft X-ray excesses are found to dominate in low column regions, then it would imply that these components are common in all AGN, but the soft component is absorbed if the object lies behind an N h > 3 — 4 x 102° cm "2.

3.3 HIGH RESOLUTION N HI MEASUREMENTS

While the Stark et al. data is the best available all-sky map of the Nh/, the angular resolution is poor, typically 2° x 3°, and the measurements are only an average over this region. It has been estimated that 90% confidence errors in N h i introduced by small-scale structure in the HI distribution, may be ± 1 x 1 0 2° cm - 2

(Elvis et al. 1986). An error of this size for an Nh of 2 x l 0 20 cm-2, results in a factor of ~9 error in the flux at 0 .2 keV. Therefore, I have tried to obtain more accurate values of N h i by using data with higher spatial resolution.

3.4 IRAS INFRARED CIRRUS MAPS

An indirect way of obtaining finer maps of the galactic neutral hydrogen is from the distribution of the infrared cirrus. Correlations between the Galactic HI emission (based on the Stark et al. data) and infrared interstellar dust emission at 60 pm and 100 pm (using data taken by the Infrared Astronomy Satellite, IRAS) have been reported by Boulanger and Perault (1988) and Rowan-Robinson et al. (1991).

Figure 3.6 Opposite : Examples of the 100 p m IR A S cirrus maps for the USS AGN. The maps cover an area of 130/ X130/ and have a resolution of l l. The USS AGN position is marked by .a circle. Galactic Column Densities 57

RA (Minutes) RA (Minutes)

Galactic Column Densities 59

Boulanger and Perault (1988) found a sky-average linear relation between the

1 0 0 /zm intensity, 1(100), and the HI column density N///> such that:

1(100) = 0.85 x 1 0 ~ 2 0 N Hi where 1(100) is in MJy sr - 1 and N h i is in cm-2. This correlation was confirmed by Rowan-Robinson et al. (1991), who present maps of the IRAS 100 /zm and Stark 21 cm HI emission covering the whole sky. Boulanger and Perault found that on scales less than a few hundred parsecs, the IRAS 100 /zm and Stark et al. HI distributions are well correlated. There are large scale variations in the ratio (on scales of tens of degrees) which may reflect a dependence of the IR emission per H atom on the interstellar radiation field intensity (Boulanger and Perault).

3.4.1 IRAS maps of the USS AGN fields

I have obtained IRAS maps of the 1 0 0 /zm emission for all of the USS AGN (including non-USS objects). These maps, which have a resolution of 1', were made by Mark Jones at Queen Mary and Westfield College. A selection of these is presented in Figure 3.6. The maps reveal that in some cases the USS AGN is seen through ‘holes’ in relatively dense cirrus, eg. E0114-016, E1254+221 and E1614+308. E1644-029 lies sandwiched between two regions of strong 100 /zm emission. The corresponding IRAS N h is 8.99 which throws doubt upon the AGN identification (it is drawn from the EMSS). While it may be possible that the AGN is seen through a hole which is on a smaller scale than the resolution of the IRAS maps ( 1 '), since this is a non-secure object, it is also possible that the 0.25 keV excess is not real. Some objects, eg. E1028+310, are seen in flat, featureless regions of the cirrus and others are observed close to strong 1 0 0 /zm sources, eg. E0111-015 and E1255+017. These

1 0 0 /zm sources may, in some cases, be related to the AGN itself. Maps for all of these objects are shown in Figure 3.6.

Mark Jones has converted IRAS 1 0 0 /zm fluxes to N h i by measuring the background cirrus and scaling this to the Stark et al. survey. At each position, a mean ratio, R, of 1 0 0 /zm to 2 1 cm emission over a 6 ° x 6 ° region was calculated

(assuming Stark values for N h i )- The 1 0 0 /zm background was taken to be the mean flux over an annulus with inner radius 2' and outer radius 5', to remove any contribution from the AGN itself. In cases where there was significant structure in this region, the mean flux within a V circle was used (see Table 3.1). Where there 60 Chapter 3

is significant source emission, the cirrus has been estimated from the surrounding

background by interpolation. The ratio R was then used to convert the 1 0 0 pm background flux to the Galactic N//j. Values of Nh interpolated from the Stark

et al. survey and derived from the IRAS 1 0 0 pm cirrus maps are listed in Table 3.1. Assuming therefore, that the Stark et al. data represents the best available

measurement of the H I emission over a 2° x 2 ° region around the source, and that the small-scale structure in the infrared cirrus traces the neutral hydrogen

distribution, then the background 1 0 0 pm flux may be used to calculate N hi for each USS AGN position.

6 12

5 8 c_>E Z IRAS b O 4 < 4 o • 3 3 O o 0 S o X) z 2 E 8 co o o £ Stark 1 4

0 0 0 1 2 3 4 5 6 2 4 6 8 10 Stark N h (units of 1020 cm'2) N h (units of 10” cm'2)

Figure 3.7 : (a) Plot showing the N n i derived from IRAS lOO^m maps versus the Stark et al. values (diamonds). The solid line indicates IRAS= Stark, (b) A comparison of the IRAS (upper histogram) and Stark et al. (lower histogram) column density distributions for the USS AGN (secure and non-secure), where measurements of both were available. The dashed lines indicate the medians of the distributions (Stark=2.41, IRAS=1.98).

3.4.2 Stark versus IRAS

Values for N h i derived from the IRAS 1 0 0 pm cirrus are compared with Stark et al. values in Figure 3.7. Most IRAS values are lower than Stark; the average difference (Stark minus IRAS) is 0.17xl02° cm - 2 with a 90% scatter of ± 1 x 1 0 20 cm-2. This is in agreement with the results of Elvis et al. (1986) who estimated that 90% errors introduced by small-scale structure were ±1 x 10 20 cm-2 . The distributions of the Stark and IRAS column densities are compared in Figure 3.7b. The median of the IRAS distribution (upper plot) is 1.98; lower than Galactic Column Densities 61 the Stark median of 2.41 (medians are indicated by the dashed line). On average, the IRAS Nh i is 5% lower than the Stark N h i (the average IRAS-to-Stark ratio is 0.95 with a 90% error of ±0.35). These results suggest that the USS AGN are seen through relatively low columns when considering the N h i on Stark et al. scales (2 ° x 2 °).

3.5 JODRELL BANK NHi In a further bid for more accurate measurements of the neutral hydrogen col­ umn density and as a test of the IRAS cirrus method of determining Nh/? I have obtained 21 cm observations of some of the USS AGN with a resolution of 12'. These were taken with the Lovell Telescope at Jodrell Bank by Phil Smith and Graziella Branduardi-Raymont. Reduction of the data, including brightness cali­ brations and the stray radiation correction, was performed using standard software at Jodrell Bank. The background was fitted with a low-order (typically order 2-3) polynomial and then subtracted before calculating the flux in the line (Smith, pri­ vate communication). Finally, the integrated brightness temperature in the line, T b (v)dv was converted to a column density using the equation: o o T B(v)dv /-oo These measurements of N hi are compared with the IRAS cirrus Nhi in Fig­ ure 3.8. The plots show that the Jodrell Bank measurements are, with only one exception, all lower than IRAS. A straight line fitted to the IRAS-to-Jodrell ratio plotted in Figure 3.8b, shows that this ratio is almost constant with the IRAS

Nhi? indicating that there is a systematic error in one set of measurements. At the same time that these USS AGN pointings were made, the Jodrell Bank telescope was also used to map the 2 1 cm emission in a very low column density field used for the ROSAT Deep Survey pointing. A comparison of the IRAS cirrus map of the Deep Survey region (scaled to the Stark survey of N h i) shows a similar discrepancy, ie. a ~ 20% deficiency in the Jodrell Bank fluxes relative to IRAS (Jones, private communication). This region is one of the eight mapped in HI at 2 1 ' resolution by Lockman, Jahoda and McCammon (1986). The IRAS map agrees well with the flux and structure in the Lockman et al. map (Jones, private communication). These results imply a systematic error in the Jodrell Bank measurements. The observations have been corrected for background and stray radiation at 90° 6 'o e b £ 5 c_>

^ 4 1.5 3 z a

0 0.5 0 1 2 3 4 5 6 IRAS Nra (units of 10” cm*2) IRAS Nm (units of 10” cm'2)

Figure 3.8 : (a) Plot showing Jodrell Bank measurements of N f j l against IRAS for the USS AGN (secure and non-secure sources). The solid line represents Jodrell=IRAS; the dashed line represents a linear least-squares fit to the data, (b) The IRAS-to-Jodrell NHI ratio as a function of IRAS Nh/« Again, the solid line represents Jodrell=IRAS; the dashed line represents a linear least-squares fit to the data.

to the pointing direction in the sidelobes of the telescope, however this correction

is crude and may affect the flux determination by ~ 2 0 % for weak 2 1 cm spectra (Pedlar 1991; private communication). Willacy (1991) has reported a deficiency

of 1 2 % in her Jodrell Bank data. A further complication for the USS AGN is that these objects lie mostly in the Galactic poles, 90° away from the plane of the

Galaxy where the 2 1 cm radiation, ie. the contaminant in the sidelobes, is strong. Therefore, the usefulness of these observations is limited. However, the IRAS- to-Jodrell ratio is nearly constant with N h i and the greatest scatter about a straight line fitted to the ratio, lies at the lowest column densities where the errors on N hi are largest (see Figure 3.8). The correlation between the IRAS and Jodrell values (a linear correlation coefficient of 0.94 with an associated probability of more than 99.9%) adds weight to the validity of using 100^z fluxes to determine N hi-

3.5.1 Results from another high resolution 21 cm survey

Elvis, Lockman and Wilkes (1989; hereafter ELW) presented Green Bank N hi measurements of 98 AGN with a resolution of 2 1 '. They quote accuracies for the column densities of ~ 1 x 10 19 cm - 2 or ~5%, whichever is the larger. Three

USS AGN appear in this sample, E1028+310 (ELW N hi = 1-80 x 1 0 2° cm-2 ),

E1704+608 (ELW N h i = 2.26 x 1 0 2° cm-2 ), and E1352+183 (ELW N h i = 1.84 x Galactic Column Densities 63

1 0 20 cm 2). IRAS measurements have been made for E1028+310 and E1352+183 (these maps are shown in Figure 3.6) but there is no 100 fim data at the position of E1704+608. The ELW and IRAS measurements agree well for E1028+310 (map quality B; see Table 3.1) but for E1352+183, the ELW value is significantly larger than IRAS

(1.18xl020 cm - 2 and a map quality A); the value interpolated from the Stark et al. survey is 2.11 xlO 20 cm - 2 (see Table 3.1). The IRAS map for E1352+183 shows that this AGN lies in a structured region which, when the difference in resolution is also taken into account, may account for the discrepancy. This, and the IRAS cirrus fine structure, suggest that there may be significant errors in N hi even on scales of 2 1 ', although a larger sample (ie. more than 2 objects!) is needed to test the significance of possible errors.

3.6 SUMMARY The results presented in this chapter have illustrated that accurate measurements of the absorbing column density are essential if a realistic determination of the intrinsic soft X-ray (~0.2 keV) flux of any object is to be made. I have shown that errors incurred by using low resolution measurements of N h i , eg. the Stark et al. data, may be as high as ± 1 x 1 0 20 cm-2, based on the assumption that the small-scale structure of the lOOjLtm cirrus traces the Nh* Errors of this size imply large uncertainties on the flux at ~ 0 .2 keV. In a bid to obtain more ac­ curate values for the Nh than the Stark survey could provide, I have obtained IRAS 100/zm measurements for most of the USS AGNj, and these will be used in the subsequent reduction of the X-ray data. Where IRAS measurements are not available, the N h i interpolated from the Stark et al. data is used, except for E l704+608 where the value given in Elvis, Lockman and Wilkes

(1989) is used. The Jodrell Bank 2 1 cm measurements are unreliable because of an apparent systematic error, possibly due to the stray radiation correction, and are not considered further. ^unpLci o

Table 3.1 : Galactic column measurements

N h (units of 1 0 2° cm 2) Map Object Stark Jodrell IRAS quality z SECURE USS

E0111-015 3.67 4.51 4.62 AS 0 .1 2 E0114-016 4.30 4.16 B 0.34 E0131-408 2 .1 1 1.84 B 2.36 E0132-411 2.14 1.91 B 0.27 E0136-250 1 .2 0 1.03 1.13 B 0.31 E0310-557 1.80 1.59 A 0.23 E0331-365 1.40 1 .2 2 C 0.31 E0845+378 3.36 2 .2 1 2.98 B 0.31 E0944+464 1.30 0.84 1.52 B 0.35 E0957+561 1.28 0.7 2 CS 1.41 E1028+310 1 .8 6 1 1.16 1.90 B 0.25 E1040+123 2.70 2.60 B 1.03 E1146+558 1.30 0.56 - D 0.44 E1208+322 1.87 1 .1 0 1.40 - B 0.39 E1227+140 2 .1 1 2.91 B* 0 .1 0 E1251-005 1.60 1.39 B 0.43 E1254+221 2.60 1.70 C 0.19 E1255+220 2.60 1.92 B 0.26: E1255+017 1.70 2.32 BS 0.16 E1334+038 2 .0 0 2.03 B 0.14 E1346+266 1.43 0 .8 8 1.18 B 0.92 E1401+098 1.98 1.84 B 0.43 E1423+201 2.54 2 .1 0 2.83 B 0 .2 1 E1425+169 1.75 1.44 1.94 BS 0 .2 2 E1519+279 2.69 2.94 3.98 A 0.23 E1614+308 2.40 2 .0 1 A 0.27 E1704+608 2.701 - D 0.37 E2034-228 3.80 - D 0.26 - E2318-423 1.98 1.90 B 0 .2 1

Map quality: A Good. Error on N/f i IX 1019 cm - 2 . B Fair. Error on Nj/ i 2X 1019 cm-2 . C Poor. Flux unreliable. Error on N// > £ 2X 1019 cm-2 . D No data. S Source contamination in integration region. * Integration over 1* radius circle at source position. Galactic Column Densities

Table 3.1 (cont.) : Galactic column measurements

Nh (units of 10 20 cm 2) Map Object Stark Jodrell IRAS quality z NON-SECURE USS

E0007-357 1 .2 0 1 .1 2 B 0.05 E0039-019 3.91 2.28 2 .6 8 B 0.35 E0114-002 3.53 2.42 2.43 B 0.04 E0129-066 2.50 2.75 3.55 B 0 .2 2 E0141+020 3.00 1.9 2 B 0 .0 2 E0150-102 1.77 1.71 1.94 B 0.36 E0200-089 2 .1 0 1.77 B 0.77 E0337-267 1.16 - D 0 .1 1 E0436-433 2.53 2.85 A 0.07 E0844+377 3.61 2.27 2.94 B 0.45 E0952+442 1.15 0.94 1 .0 0 A 0.47 E1008+348 2.50 1.19 B 0.14 E1011+496 0.83 0.77 0.9 2 BS* 0 .2 0 E1059+730 4.03 3.64 A 0.09 E1215+692 1.59 1.50 D 1 .1 2 E1217+695 1.60 1.27 1.42 B 0.63 E1218+693 1.60 1.05 1.42 B 0 .1 1 E1352+183 2 .II 1 1.60 1.18 A 0.15 E1423+242 2.70 D 0.65 E1511+671 2.73 - D 0.31 E1640+537 2.54 3.22 B 0.14 E l644-029 9.20 8.99 B* 0.26 E1657+326 2.42 2.54 A 0.09 E1805+700 5.68 4.10 B 0.19 NON-USS

E0906+1H 4.37 3.41 A 0.16 E1213+378 1.50 1.42 C 0.82 E1228+123 1.70 2.30 B 0 .1 2 E1304+342 1 .0 2 0.89 B 0.28 E1654+352 2.28 1.62 B 0.80

: Value uncertain. 1 Column density also measured by Elvis, Lockman and Wilkes (1989; see Section 3.1). 2 Estimated value for the background 100/mi flux. Error on Nj/ = ± IX 1020 cm-2 .

CHAPTER 4

Analysis of the X-ray Data

This unique sample comprises 53 AGN with an unusually strong observed flux at ~0.25 keV. In this Chapter, I present the X-ray data upon which the selection is based (see Chapter 2 for details of the selection criteria). The X-ray spectral fitting is described and finally, I present and discuss the separate hard and soft X-ray component luminosities which are calculated from the fitted models. The X-ray ‘spectra’, ie. the counts in the Cl, C2 and C3 bands, are given in Table 4.1 (columns 4, 5 and 6). Also listed are (1) the IPC X-ray position in J2000 co-ordinates, (2) the IPC sequence number, (3) the IPC countrate (in the 0.16-3.50 keV range) and (7) the total exposure time.

“The real question for 1988 is whether we’re going to go forward to tomorrow or past to the - to the back!”

— DAN QUAYLE 68 Chapter 4

Table 4.1 : USS AGN X-ray data

(1) (2) (3) (4) (5) (6) (7) Object X-ray position (J2000) Seq Countrate C l C2 C3 Exp Name RA Dec No rate err cts err cts err cts err time(s)

SECURE E0111-015 01 14 28 -01 16 41 5394 1.8 0.1 15.7 6.9 4.0 4.8 8.1 7.6 13968 E0114-016 01 17 08 -01 24 37 5394 0.8 0.1 30.6 11.9 8.6 8.6 22.6 11.5 13968 E0131-408 01 33 50-40 38 13 2578 0.4 0.1 80.0 19.6 16.9 14.4 92.0 18.9 28587 E0132-411 01 34 58 -40 56 04 2578 1.0 0.1 136.5 19.3 34.3 15.0 3.5 17.9 28587 E0136-250 01 38 40 -2 4 50 30 3996 3.0 0.5 32.4 9.0 10.8 4.5 15.6 6.7 3663 E0310-557 03 11 47-55 32 09 6465 5.9 0.9 21.6 6.3 7.2 4.2 19.0 5.6 1397 E0331-365 03 33 13 -36 20 04 3058 4.4 1.0 13.5 5.2 4.3 2.7 1.7 2.7 772 E0845+378 08 48 20 37 40 45 1840 1.9 0.4 23.5 6.0 6.8 3.7 5.8 5.4 2018 E0944+464 09 47 19 46 15 06 5985 1.2 0.4 21.9 6.7 2.7 5.5 -2.1 6.6 3511 E0957+561 10 01 20 55 53 55 423 2.9 0.8 14.3 4.1 2.8 3.2 -2.4 3.3 846 E1028+310 10 31 37 30 47 18 4256 0.7 0.2 17.7 8.7 -5.7 7.0 8.7 8.2 6595 E1040+123 10 42 45 12 03 37 497 2.6 0.3 52.3 9.8 14.3 7.7 58.1 10.0 4310 E1146+558 11 48 50 55 36 26 6101 0.9 0.2 35.6 9.3 -12.4 8.0 7.0 8.3 5997 E1208+322 12 10 39 31 57 19 3966 4.9 0.6 30.8 6.4 10.8 5.2 16.7 6.2 1944 E1227+140 12 29 32 13 46 42 2124 2.3 0.4 36.5 9.8 12.5 8.5 38.4 11.9 4388 E1251-005 12 53 35 -00 47 41 7039 1.0 0.2 36.2 10.6 2.3 8.5 5.6 7.9 12093 E1254+221 12 56 58 21 53 36 2136 1.3 0.2 23.9 8.2 8.4 4.9 14.2 7.1 4867 E1255+220 12 57 49 21 44 41 2136 0.9 0.2 15.2 7.3 1.5 5.9 1.3 8.1 4867 E1255+017 12 58 20 01 31 41 9156 1.4 0.3 43.5 9.2 5.7 6.9 22.2 8.3 4790 E1334+038 13 37 09 03 36 24 5547 1.2 0.1 37.9 10.5 10.4 8.2 47.0 11.1 12628 E1346+266 13 48 34 26 22 08 293 1.5 0.3 58.1 11.3 16.6 7.6 -16.1 9.8 6498 E1401+098 14 04 13 09 37 27 6684 6.7 1.1 14.5 4.5 3.2 2.5 19.2 4.7 1529 E1423+201 14 26 12 19 55 28 4396 3.3 0.7 25.0 6.0 1.0 2.9 -0.9 2.9 1763 E1425+169 14 27 45 16 41 26 6077 1.2 0.2 94.9 17.5 -3.5 11.6 43.2 14.9 14136 E1519+279 15 21 31 27 44 07 1796 1.9 0.3 35.0 9.9 10.9 8.7 -3.2 11.9 7048 E1614+308 16 16 52 30 45 26 3548 0.9 0.1 32.8 11.9 8.4 8.7 40.4 11.6 16484 E1704+608 17 04 38 60 44 53 4208 2.8 0.4 28.3 6.8 8.2 5.3 29.8 7.8 3229 E2034-228 20 37 30 -22 42 46 8390 2.9 0.6 18.9 5.6 3.6 3.7 15.7 6.3 1717 E2318-423 23 20 59 -4 2 03 36 6385 1.7 0.2 81.6 12.4 23.4 9.3 33.6 11.1 9768

(1) Einstein IPC X-ray position. (2) IPC sequence number. (3) IPC countrate (0.16-3.50 keV) in units of 0.01 counts per second. Errors lcr. (4) Total counts in the 0.16-0.56 keV band. Errors lcr. (5) Total counts in the 0.56-1.08 keV band. Errors lcr. (6) Total counts in the 1.08-3.50 keV band. Errors lcr. (7) Exposure time in seconds. Analysis ol the A-ray Data

Table 4.1 (cont.) : USS AGN X-ray data

Object X-ray position (1950) Seq Countrate C l C2 C3 Exp Name RA Dec No rate err cts err cts err cts err time(s)

NON-SECURE E0007-357 00 10 19 -35 26 38 4518 2.1 0.4 19.5 8.0 6.2 5.0 10.4 6.7 3050 E0039-019 00 42 20 -01 42 09 5393 1.6 0.2 19.6 6.7 3.0 3.6 -9.3 4.0 9482 i

E0114-002 01 17 00 -00 00 42 6083 1.0 0.2 9.7 5.4 o ►—» 2.7 8.2 4.9 12116 E0129-066 01 32 15-06 24 53 5768 1.7 0.5 3.0 2.4 -0.9 1.8 -0.4 2.5 3556 E0141+020 01 43 57 02 20 48 3464 2.0 0.4 19.7 5.9 6.0 5.6 11.5 6.4 2570 E0150-102 01 53 23 -10 00 28 5179 2.6 0.4 0.2 2.7 -2.3 0.7 3.6 2.3 4820 E0200-089 02 03 24-08 43 57 5163 1.3 0.2 24.3 9.9 8.5 5.7 7.9 8.7 5135 E0337-267 03 39 13 -26 36 15 2096 1.8 0.2 23.7 6.3 5.1 3.9 -5.6 4.7 8482 E0436-433 04 37 57 -43 15 34 4011 1.1 0.2 6.0 4.2 -0.4 2.6 0.0 3.4 7387 E0844+377 08 47 15 37 32 14 1840 2.5 0.6 2.9 2.9 -0.2 1.2 2.3 2.1 2018 E0952+442 09 55 30 43 57 50 6853 3.2 0.6 13.8 9.1 2.1 6.6 16.4 5.6 3161 E1008+348 10 11 49 34 37 52 2702 7.1 0.9 20.3 4.9 3.7 2.9 10.3 3.8 1530 E1011+496 10 15 01 49 26 40 4414 16.0 1.2 20.2 5.3 5.6 2.7 -0.3 1.2 3970 E1059+730 11 02 30 72 46 38 1948 2.5 0.5 2.1 2.1 0.4 1.0 2.0 2.3 3107 E1215+692 12 18 15 68 59 53 5803 0.8 0.2 3.4 2.7 0.6 1.4 -0.8 0.3 13485 E1217+695 12 19 23 69 14 12 5803 0.9 0.1 5.9 6.2 -0.3 4.0 -5.5 4.1 13485 i

E1218+693 12 20 40 69 05 30 5803 3.2 0.3 3.6 5.6 o 4.2 12.9 5.4 13485 E1352+183 13 54 36 18 05 23 851 15.3 1.4 43.0 6.8 10.0 3.6 6.0 3.2 1324 E1423+242 14 25 53 24 04 22 3898 4.7 1.0 3.3 2.3 0.6 1.0 1.3 1.4 1123 E1511+671 15 12 27 66 56 44 2727 1.2 0.3 6.8 4.7 1.9 2.2 0.0 2.6 5547 E1640+537 16 42 01 53 40 01 8351 2.2 0.3 9.3 5.0 -5.7 2.7 -0.5 4.2 8386 E1644-029 16 46 48 -03 03 51 2494 1.2 0.3 13.7 10.6 2.1 9.0 33.1 10.6 8711 E1657+326 16 58 59 32 37 17 4954 1.2 0.2 9.3 9.0 -1.1 6.9 18.2 9.4 5490 E1805+700 18 05 21 70 06 23 5689 0.6 0.1 1.7 4.7 -0.7 3.5 -9.5 3.5 19709

NON-USS E0906+111 09 09 02 10 59 28 4959 7.7 0.8 6.5 3.4 2.2 2.5 3.1 3344 to i i i

E1213+378 12 15 30 37 32 18 5803 0.6 0.2 5.9 -0.3

4.1 X-RAY SPECTRAL MODELS As with all proportional counter detectors and particularly given the relatively poor spectral resolution of the IPC, the relationship between detected counts and incident flux is dependent on the assumed spectral model. To derive representative source fluxes, we first need to quantify the spectral shape of these sources.

4.1.1 Two-component modelling By definition, the USS sources are bright in the 0.16-0.56 keV band. However, this does not preclude the presence of a hard component; indeed examination of the raw count distribution in Table 4.1 reveals that many (about a third) of the sources have significant counts in C3 while counts in C2 are low. Therefore the spectra are parameterized in terms of two components; the ‘soft’ and ‘hard’ components. There are only 3 spectral points available for each AGN, so the ability to constrain the spectrum of any one object is limited.f Instead, I consider the USS AGN as a class and determine a mean soft component index for the sample as a whole. The hard X-ray component was represented by a power-law, while both power-law and blackbody models were tried for the soft component. For all fits, the absorbing column was fixed to the IRAS cirrus measurement of N h i where available, otherwise N h i was interpolated from the Stark et al. (1992) data, except for E l704+608 where the value given in Elvis, Lockman and Wilkes (1989) was used. All column densities and a full discussion of their derivation are given in Chapter 3.

4.1.2 Model parameter ranges. As the first stage in finding the best-fit ‘mean’ two-component model for the data, I have searched for the appropriate parameter ranges of the soft and hard X-ray components, ie. the power-law index, a, for the hard component and either power- law index or effective blackbody temperature, kTeff, for the soft component. The definitions for a and kTeff are given by the following equations, where F„ is the flux per unit frequency interval and v is the frequency. For the power-law model: F„ oc v ~ a

t Because of the low statistical quality of most of the USS data and the poor resolution of the IPC detector, little would have been gained by using the full number of detector Pulse Height Analyser channels. Analysis ot the A-ray Data 71 and the blackbody model:

27xhvz 1 F„ oc . hv (ekTeff — 1 ) ie. the Planck function, where h is the Planck constant, c is the velocity of light, k is the Boltzmann constant and Teff is the blackbody temperature. The term ‘effective temperature’, used throughout this thesis, refers to the quantity kTeff and is expressed in units of energy.

4.1.2.1 x 2 grids

These parameter ranges were found by searching over a ‘grid’ of indices for the soft and hard components to find the preferred values for the two-component model. At each grid point, the corresponding two-component model was fitted to the USS spectra and the reduced x 2 f°r the fit was calculated. Both hard and soft indices were fixed, leaving just the normalizations free. Only the low-redshift (z<0.5) secure sources were used for this purpose (see Section 4.3). These x 2 grids are presented in Figure 4.1, for two-power-law models and soft blackbody plus hard power-law models.

4.1.2.2 Grid results

The grids show that the data prefer the steepest possible (opposing) slopes for both components. This is required for objects with significant C3 counts, to fit the resulting deep trough in C2. The soft component is best fit with either a very steep soft power-law index or a very low blackbody temperature. Blackbody models for the soft component are preferred over power-law models (at a given hard component power-law index), which is in agreement with results found by Urry et al. (1989). In particular, it proved difficult to reproduce the C2 trough using a power-law description. 72 Chapter 4

1

3.00 — 3

*oS — 2.00 • GM 5 £

O 1.20 — a* 7 g G 9 - v fO oO 11 « CO

13

v 151__ 1.00 0.90 0.80 0.700.60 0.50 0.40 0.30 Hard component power-law index

24 v 22

20 S3 TJ 0 18 seg 3 4—> 16 s 8 14 o1

CO 12

10 ' v v / / 8 1 lA ______A______L_Z ____I______I___Z __I______I_____Z 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20

Figure 4.1 : Reduced X? grids for the two-component model fits to the secure USS AGN. The upper plot shows X 2 f°r two-power-law models and the lower plot assuming a blackbody model for the soft X-ray component (where the blackbody temperature was fixed in frame of the observer) and a power-law model for the hard X-ray component. Analysis 0 1 the A-ray Data 05

4.2 FINDING THE BEST-FIT MODEL 4.2.1 Fitting the hard X-ray component... As the data quality of this sample is poor, the hard component index was fixed at an appropriate value. It has long been established that hard X-ray components of

AGN in the 2 - 2 0 keV range may be fit by a power-law with the ‘canonical’ energy index, a, of 0.7 (see Section 1 .2 .2 .1 ). For all subsequent fits to the USS AGN spectra therefore, the hard X-ray power-law index was fixed at this value.

4.2.2 ...and the soft X-ray component Having fixed the hard component, the soft component index (a or kTeff) was allowed to vary until a best fit was found. This left, as free parameters, the mean spectral index of the soft component together with the normalization of the two components for each source. Thus the number of degrees of freedom was n- 1 , where n is the number of AGN fitted.

4.2.2.1 Three soft component models Two different versions of the blackbody model are possible: it may be assumed that the blackbody temperature is the same in the observer frame or the rest- frame of each AGN. Thus, the soft components were fitted with three different models: 1. a blackbody with a mean observer frame temperature, 2 . a blackbody with a mean rest-frame temperature and 3. a power-law with a mean index.

4.2.3 A comparison of the models The rest-frame version of the blackbody model converged to a kTeff of 9 eV; the observer-frame model had not converged down to 8 eV. The soft power-law model had not converged at the steepest slope tried, which was 16. For all three types of model, the quality of the fit is not very sensitive to the index (this can be seen o 6 te s ,ye r for the power-law and ^ -frame blackbody models in Figure 4.1). Thus, for the purposes of calculating separate soft and hard X-ray luminosities and comparing the three models, the IPC data were re-fit, having fixed the soft component index. The values adopted were kTeff=7.85 eV for the observer-frame blackbody, k T efF=10 eV for the rest-frame blackbody and o = 1 0 for the power-law. The observer frame temperature of 7.85 eV was chosen because it corresponds to a rest-frame temperature of 1 0 eV at the mean redshift of the low redshift secure AGN (for objects with a z<0.5, z=0.27). The hard component was represented by an a=0.7 power-law in each case. 74 Chapter 4

Fitting only the low redshift (z<0.5) secure sources, gave x2s of 1.0 for an observer frame kTeff of 7.85 eV and for a rest-frame kTeff of 1 0 eV. A higher \ 2 ? 1.2, was obtained when fitting only the low redshift (z<0.5) secure sources with a = 1 0 . A more detailed comparison of these models is shown in Figure 4.2.

4.2.3.1 Observer frame versus rest-frame blackbody.

SOFT COMPONENT LUMINOSITIES Figure 4.2a shows the ratio of observer-frame to rest-frame soft component lu­ minosities plotted as a function of redshift. This ratio is less than 1 below a redshift of ~0.25 and increases with redshift. This is due to the flattening of the observer-frame slope relative to the rest-frame model (when both are considered in the rest-frame). This ‘hardening’ means that the number of photons with higher energies increases relative to the rest-frame model (with a corresponding decrease in the number of soft photons). Therefore the luminosity integrated over 0 .2 to 4.5 keV increases. The blackbody temperatures (and therefore their slopes) are equal for both models at z=0.27; here the ratios are close to 1 with little scatter. In Figure 4.2c, this ratio is plotted as a function of the column density. There is no correlation between the ratio and Nh- The apparent ‘ceiling’ in the data at a ratio of 1 is due to the concentration of sources at z~0.25 (see Figure 8.1).

HARD COMPONENT LUMINOSITIES The ratios of the corresponding hard component luminosities, shown in Figure 4.2e, show very little scatter in comparison to the soft.

4.2.3.2 Blackbody versus power-law models.

SOFT COMPONENT LUMINOSITIES A much larger degree of scatter is seen when comparing soft component lumi­ nosities of a blackbody model with a power-law. The ratio of (observer-frame) blackbody to power-law soft X-ray luminosities increases with redshift (see Fig­ ure 4.2b). This is due to the blackbody slope flattening relative to the power-law in the rest-frame as the redshift increases. No similar correlation is seen for the rest-frame model: in this case the slope of the blackbody is constant with redshift. Figure 4.2d illustrates that the blackbody/power-law ratio increases with the absorbing column density for the observer frame blackbody model.f Figure 4.3

f The data point at very high ratio, seen in Figure 4.2b is not shown because it has a high column density (N h j = 9-2) and is out of the range of this plot. Analysis of the X-ray Data

10.0 100.00 Ii-H I 10.00 (b) £5o 3 * * * * * * * * „ ** * * * * * * * 00

1

10.0 100.00

10.00 & o £3o JCO % 1.00 £4) r\*£ * oom 00m C-’ C-’ M ^ 0.10 1 o £3o u J3 0.01 3 4 0 1 2 3 4 5 Nh (units of 10” cm'2) Nh (units of 10” cm'2)

y w urfvniMf>aifMfr' ^ mow/ uam / m/

0.2 0.3 0.2 0.3 Redshift Redshift

Figure 4.2 : A comparison of the soft (Lsoft) and hard (Lhard) component X-ray lumi­ nosities integrated over 0.2 to 4.5 eV for three different models fitted to the USS AGN spectra; observer frame: soft component represented by a 7.85 eV blackbody fixed in the observer’s frame (kT=7.85eV), rest frame: soft component represented by a 10 eV blackbody fixed in the AGN rest-frame (kT=10eV), and power-law: soft component rep­ resented by a power-law with energy index, (*=10 ((*=10). The hard component was represented by an Ot=0.7 power-law in each case, (a) Lsoft ratio for the observer frame and rest-frame models plotted as a function of redshift. (b) Lsoft ratio for the observer frame and power-law models plotted as a function of redshift. (c) Lsoft ratio for the observer frame and rest-frame models plotted as a function of column density, N//. (d) Lsoft ratio for the observer frame and power-law models plotted as a function of N h - (e) Lhard ratio corresponding to the observer frame and rest-frame models plotted as a function of redshift. (f) Lhard ratio corresponding to the observer frame and power-law models plotted as a function of redshift. demonstrates that this is also true for the rest-frame blackbody model but with less scatter. Also plotted in Figure 4.3 as a solid line, is the simulated soft component blackbody/power-law ratio as a function of column density for a ‘typical’ USS AGN (I have used the data for E0132-411), which shows close agreement with the data.

10.0

o •? XX £ 1.0 X X o1—1

coO

0 1 2 3 4 5 NH (units of lO^cm'2)

Figure 4.3 : The ratio of rest-frame-blackbody to power-law soft com­ ponent luminosity plotted as a function of N h - Also plotted as a solid line is the simulated ratio for a typical USS AGN.

HARD COMPONENT LUMINOSITIES Again, the scatter in the ratios for the hard component is small compared to the soft component ratios, and shows no dependence on redshift (see Figure 4.2f) or column density.

4.3 X-RAY LUMINOSITIES

The separate hard and soft X-ray luminosities are listed in Table 4.2. These were calculated assuming a Friedmann universe with a Hubble constant, Ho=50 km s - 1

Mpc - 1 and deceleration parameter, qo= 0 . Calculating the soft component luminosity from the IPC data alone has proved to be a problem. While these blackbody and power-law models are adequate de­ scriptions of the soft component within the IPC range, they are wholly inappro­ priate when extrapolated to energies lower than ~0.16 keV; higher quality X-ray spectra plus additional UV-to-EUV data would be required to construct a more Analysis ol the X-ray Data 77 realistic model. Therefore the soft component models can only be used to describe the spectrum reliably within the IPC range. Due to the effects of redshift, this sets a lower limit to the energy range over which the luminosity may be calculated; at the same time it is necessary to extend this range to the lowest possible energy in order to encompass a significant fraction of the soft component. Therefore, broadband soft X-ray luminosities are integrated over 0 .2 to 4.5 keV which allows realistic values for the soft X-ray component for all AGN with a redshift of 0.5 or less (25 out of 29 secure sources and 20/24 non-secure sources). Hard component luminosities are also integrated over this range.

Monochromatic luminosities are calculated at 0 .2 keV for the soft component, and 2.0 keV for the hard. Upper limits to the hard component normalization were derived individually by selecting the highest value which did not violate the count error margins in any band.

Soft X-ray component luminosities at 0 .2 keV for all three models are given in Table 4.2. Hard and soft component luminosities integrated over 0 .2 to 4.5 keV and the hard component luminosity at 2 .0 keV are given for the 7.85 eV blackbody observer-frame model only. The corresponding hard component luminosities for the blackbody rest-frame and power-law models are not listed because they differed from the observer-frame model by only 2 % on average.

4.3.1 Soft-to-hard X-ray ratio - defining a sx The soft-to-hard X-ray luminosity ratio is described by the quantity a 8X<> which I have defined as follows:

. Lo^keV •>og—~ — ft*! D2keV where L0 .2 keV is the monochromatic luminosity at 0.2 keV (soft plus hard com­ ponent luminosities) and Ii2keV is the monochromatic luminosity at 2 keV. The values of a sx listed in Table 4.2, correspond to the observer-frame model. 78 Chapter 4

Table 4.2 : USS AGN X-ray Luminosities

T ® Object Lq.2keV Lq 2keV 0.2keV hgoft L2 keV hfiard z N h (1) (2) (3) (4) (5) (6) (7) (8) (9)

E0111-015 29.0 29.0 28.5 44.5 25.3 43.4 3.7 0.12 4.6 E0114-016 30.9 31.0 29.8 46.4 26.1 44.2 4.8 0.38 4.2 E0131-408 27.8 45.9 2.36 1.8 E0132-411 28.8 28;8 28.6 44.3 <25.3 <43.4 > 3.5 0.27 1.9 E0136-250 28.5 28.4 28.5 43.9 26.4 44.5 2.1 0.31 1,1 E0310-557 28.3 28.4 28.4 43.8 26.5 44.6 1.9 0.23 1.6 E0331-365 29.0 28.9 29.0 44.4 <26.3 < 44.4 > 2 .7 0.31 1.2 E0845+378 30.1 30.2 29.5 45.6 26.0 44.1 4.1 0.31 3.0 E0944+464 29.3 29.3 29.1 44.8 <29.3 <44.2 > 0.0 0.35 1.5 E0957+561 < 27.9 <45.9 1.41 0.7 E1028+310 28.4 28.4 28.3 43.9 *25.3 *43.4 3.2 0.25 1.9 E1040+123 27.7 45.8 1.03 2.6 E1146+558 29.6 29.5 29.3 45.1 < 26.2 <44.3 >3.4 0.44 1.3 E1208+322 29.6 29.6 29.4 45.1 26.8 44.9 2.8 0.39 1.4 E1227+140 27.5 27.6 27.5 42.9 25.4 43.5 2.1 0.10 2.9 E1251-005 29.5 29.5 29.2 45.0 <25.8 <43.9 > 3 .7 0.43 1.4 E1254+221 27.5 27.6 27.6 43.0 25.6 43.7 1.9 0.19 1.7 E1255+220 28.7 28.7 28.5 44.2 <25.1 <43.2 >3.6 0.26 1.9 E1255+017 27.9 28.0 27.9 43.4 25.4 43.5 2.5 0.16 2.3 E1334+038 26.8 26.9 27.0 42.3 25.3 43.4 1.5 0.13 2.0 E1346+266 <26.9 <45.0 0.92 1.2 E1401+098 30.2 30.3 29.7 45.7 27.2 45.2 3.1 0.44 1.8 E1423+201 29.6 29.6 29.2 45.1 <25.6 <43.8 >3.9 0.21 2.8 E1425+169 28.3 28.3 28.2 43.8 *25.5 *43.6 2.8 0.22 1.9 E1519+279 30.2 30.2 29.5 45.7 <25.1 < 43.2 >5.1 0.23 4.0 E1614+308 28.2 28.2 28.0 43.7 25.9 44.0 2.3 0.27 2.0 E1704+608 29.9 29.9 29.3 45.4 26.7 44.7 3.2 0.37 2.3 E2034-228 30.2 30.2 29.5 45.7 26.2 44.3 4.0 0.26 3.8 E2318-423 28.2 28.2 28.2 43.7 25.7 43.8 2.5 0.21 1.9

(1) Logio of the soft X-ray component luminosity at 0.2 keV for a blackbody with an observer- frame Teff=7.85 eV, in erg s-1 . (2) Logio of the soft X-ray component luminosity at 0.2 keV for a blackbody with a rest-frame Teff=10 eV, in erg s-1 . (3) Logio of the soft X-ray component luminosity at 0.2 keV for a soft power-law with index = 10, in erg s —1. (4) Logio of the soft X-ray component luminosity integrated over 0.2 to 4.5 keV for a blackbody with an observer-frame Teff=7.85 eV, in erg s—1. Analysis of the X-ray Data 79

Table 4.2 (cont.) : USS AGN X-ray Luminosities

T 3 Object Lo.2keV Lq.2keV l 0.2 keV hgoft hlkeV Lhard &8X z N h (1) (2) (3) (4) (5) (6) (7) (8) (9)

E0007-357 25.8 26.2 26.5 41.3 25.1 43.2 0.7 0.09 1.1 E0039-019 30.4 30.5 29.8 45.9 <26.1 <44.2 >4.3 0.35 2.7 E0114-002 25.6 26.0 26.0 41.1 *24.0 *42.1 1.7 0.04 2.4 E0129-066 29.9 29.9 29.3 45.4 <26.3 <44.4 >3.6 0.22 3.6 E0141+020 24.5 25.0 25.2 40.0 23.6 41.7 0.9 0.02 1.9 E0200-089 <26.9 <45.0 0.77 1.8 E0337-267 26.6 26.8 27.1 42.0 <25.1 <43.2 >1.5 0.11 1.2 E0436-433 27.1 27.3 27.3 42.6 < 24.9 < 43.0 > 2 .2 0.07 2.9 E0844+377 31.1 31.3 30.1 46.6 26.6 44.7 4.6 0.45 2.9 E0952+442 28.9 28.9 28.6 44.4 27.0 45.1 1.9 0.47 1.0 E1008+348 27.2 27.4 27.7 42.7 26.0 44.1 1.2 0.14 1.2 E1011+496 28.3 28.4 28.7 43.7 <25.2 <43.3 >3.1 0.20 0.9 E1059+730 27.9 28.0 27.7 43.4 <25.2 < 43.3 > 2 .7 0.09 3.6 E1215+692 <26.7 <44.8 1.12 1.6 E1217+695 <27.0 <45.1 0.63 1.4 E1352+183 27.9 28.1 28.3 43.4 26.1 44.2 1.8 0.15 1.2 E1423+242 <27.2 <45.3 0.65 2.7 E1511+671 29.9 29.9 29.4 45.4 <25.5 <43.6 >4.5 0.31 2.7 E1640+537 28.8 28.8 28.4 44.2 <25.7 <43.8 >3.1 0.14 3.2 E1644-029 31.7 31.6 29.8 47.2 26.1 44.2 5.6 0.25 9.0 E1657+326 26.4 26.5 26.4 41.9 25.1 43.2 1.3 0.09 2.5 E1805+700 29.4 29.4 28.7 44.9 <26.1 <44.2 >3.3 0.19 4.1

E0906+111 29.8 29.8 29.5 45.3 <26.4 <44.5 >3.4 0.16 3.4 E1213+378 <27.1 <45.2 0.82 1.4 E1227+140 27.8 27.9 27.8 43.3 <25.3 <43.4 >2.5 0.10 2.9 E1228+123 28.1 28.2 28.2 43.6 <26.2 <44.3 >1.9 0.12 2.3 E1304+342 28.1 28.1 28.3 43.6 <25.9 <44.0 >2.2 0.28 0.9 E1654+352 <26.6 <44.7 0.80 1.6

(5) Logio of the hard X-ray component luminosity at 2.0 keV corresponding to the Teff=7.85 eV observer-frame model, in erg s—1. (6) Logio of the hard X-ray component luminosity integrated over 0.2 to 4.5 keV corresponding to the Teff=7.85 eV observer-frame model, in erg s-1 . (7) Ratio of 0.2 keV to 2.0 keV luminosities (see Section 4.3.1) corresponding to the 7.85 eV observer frame model. (8) Redshift. (9) Absorbing column in units of 1020 cm-2 (sec Chapter 3).

* X-ray spectrum poorly fit by the model. 80 Chapter 4

4.4 SUMMARY By comparing the X-ray luminosities calculated under three different assumptions, I have demonstrated that errors on the calculated soft component luminosities can be significant (typically factors of a few). These dominate over errors incurred in the measurement and calculation of the absorption (~30% at 2xl0 20 cm-2 ; see

Chapter 3 and Figure 3.2) and the errors in the raw data ( 1

Optical Data

5.1 SPECTROSCOPY The spectra of the USS AGN were taken during three observing runs; two on the 4.2m William Herschel Telescope (WHT) at the Observatorio del Roque de los Muchachos, La Palma and one with the 3.9m Anglo-Australian Telescope (AAT). The WHT spectra were taken using the Faint Object Spectrograph (FOS; Allington-Smith et ai. 1989) mounted at the cassegrain focus of the telescope, in 1988 February and June. The FOS is a fixed-format high throughput spectrograph, which is equipped with a dye coated GEC CCD. It covers the spectral range from

3600 A to 5000 A in second order and from 5000 A to 1 0 0 0 0 A in first order, with dispersions of 5 and 9 A pixel-1, respectively. The A AT spectra were taken in 1989 January using the RGO Spectrograph and Image Photon Counting System (IPCS) in the blue and the Faint Object Red Spectrograph (FORS) in the red. The IPCS was operated in the range 3500 A to 5600 A with data collected in 2048 channels at a dispersion of 1.5 A pixel-1 . The FORS spectra covered the range 5400 A to 1 0 0 0 0 A in 584 channels with a dispersion of 1 0 A pixel-1 . A narrow spectrograph slit (between 1.0 and 1.5 arcseconds) was used for the WHT and AAT spectra except for the observation of E0132^111 which was made with a wide slit (4.5 arcseconds) for photometric accuracy. Generally, the seeing was in the range 1 to 2 arcseconds. All spectra were taken with the spectrograph slit at the parallactic angle. Details of the spectroscopic observations made at the WHT and the AAT are listed in Table 5.1. 82 Chapter 5

Table 5.1 : Observation Summary

Source Star Classification Telescope Date of Redshift No. Spectrum SECURE USS E0132-411 1 AGN AAT 07 JAN 89 0.27 E0944+464 1 AGN WHT 10 FEB 88 0.35 E0957+561 1 double WHT 16 FEB 88 1.41 2 quasar 99 r> E1028+310 1 AGN WHT 11 JUN 88 0.25 E1146+558 1 AGN WHT 08 JUN 88 0.44 E1227+140* 1 AGN WHT 21 FEB 88 0.10 E1346+266 1 AGN WHT 12 FEB 88 0.92 E1423+201 1 AGN WHT 22 FEB 88 0.21 2 star 99 99 E1425+169 1 AGN WHT 11 FEB 88 0.22 2 spiral galaxy no spectrum NON SECURE USS E0039-019 1 AGN AAT 07 JAN 89 0.35 E0114-002 1 AGN AAT 07 JAN 89 0.05 2 late-type star 3 star E0129-066 1 AGN AAT 07 JAN 89 0.22 E0337-267 1 AGN WHT 11 FEB 88 0.11 2 late-type star 99 99 3 Ha + G-band star 99 99 E0436-433 1 AGN AAT 07 JAN 89 0.07 2 star n n E0844+377 1 AGN WHT 17 FEB 88 0.46 E1215+692 1 AGN WHT 21 FEB 88 1.12 E1217+695 1 AGN WHT 22 FEB 88 0.64 2 star 99 Y1 E1218+693 1 AGN WHT 14 FEB 88 0.11 2 star 99 Y> 3 M-star 99 Y i E1511+671 1 AGNWHT 11 JUN 88 0.31 2 K-star 99 3 SAO star no spectrum E1640+537 1 AGNWHT 07 FEB 88 0.14 2 K-star E1657+326 1 AGNWHT 17 JUN 88 0.09 r> 18 JUN 88 E1805+700 1 AGN WHT 19 JUN 88 0.19 2 galaxy no spectrum 3 star r> 19 JUN 88 NON-USS E0906+111 1 AGN WHT 07 FEB 88 0.16 E1213+378 1 AGN WHT 21 FEB 88 0.82 E1228+123 1 AGN WHT 21 FEB 88 0.12 E1304+342 1 AGNWHT 12 FEB 88 0.28 E1654+352 1 AGN WHT 08 JUN 88 0.80

* There are two USS X-ray spectra for this object (see Table 4.1). Optical Lfata

5.1.1 Data reduction The spectra were extracted and sky-subtracted using Mukai’s (1990) implemen­ tation of Horne’s (1986) optimal extraction algorithm. Wavelength calibrations were derived from Cu-Ar and Hg arc spectra. Secondary calibrations for the WHT spectra were obtained from night sky lines. Observations of photometric standards were made during each night using the same instrumental setup as for the pro­ gramme stars and these were used for flux calibration. To remove atmospheric

absorption features in the red, the spectrum of an F 8 star, taken from the SAO Catalogue and adjacent in time and sky position to each target, was measured.

5.1.2 Spectral analysis

5.1.2.1 Line identification and redshift The spectra taken at the WHT and the AAT are shown in Figure 5.1. These are plotted in the rest-frame of the AGN but no correction has been made for the effect of light lost through the spectrograph slit. The expected positions of some emission lines and Fell blends commonly seen in the optical spectra of AGN are indicated. One or both of the familiar H/?/O[III]AA4959,5007 and Ha/[SII]AA6717,6731 groups of lines are clearly identifiable in most of the spectra. The positions of [OIIIJA4959 and [OIIIJA5007, when observed, were used to determine the redshift of the AGN. For the high redshift AGN, the observed positions of the emission lines were compared with the rest positions of lines which often appear in AGN spectra, such as MgIIA2798, [CIII]A1908 and [CIV]A1549, until a match was found. A gaussian profile was fitted to each line to determine its centroid and the AGN’s redshift was calculated by averaging the of individual lines.

5.1.2.2 Measurement of line parameters I have measured the flux, equivalent width and FWHM of each emission line and the results are presented in Table 5.2 together with upper limits on the more common lines that are not seen. A gaussian profile was fitted to each line with the local continuum represented by a second order polynomial. The line flux, equivalent width and FWHM were usually calculated directly from the data after the fitted continuum had been subtracted. For the weaker lines which are only a few pixels wide, the FWHM was taken to be that of the gaussian fit (these are marked ‘g’ in the table). In cases where lines were blended, a gaussian profile 84 Chapter 5 whose FWHM was fixed to that measured for unblended lines of similar origin was fitted to each line. The flux and equivalent width were then calculated from this gaussian fit (in these cases, the flux, FWHM and equivalent width of the line are marked with an ‘f’ in Table 5.2). The FWHM listed in Table 5.2 have not been deconvolved from the instrumental profile (see Section 5.1.2.5 for details of the instrumental line widths).

5.1.2.3 Blended Fell emission. Fell emission is a well-known feature of AGN spectra. Osterbrock (1977) showed that it was present in 90% of Seyfert 1 galaxies. From the analysis of laboratory spectra, 675 energy levels between 900 A and 50,000 A are presently known (Jo­ hansson, 1986) and the entire Fell spectrum is complex. Lines within multiplets are blended with each other and multiplets also overlap and are blended. This may produce a ‘false’ continuum in a spectrum so that the actual underlying continuum is disguised. A model is required to measure the Fell emission reliably and to esti­ mate the strength in the Fell spectrum. For the purposes of this thesis however, I follow previous practice (eg. Stephens 1989) and measure the flux in two relatively well-defined optical Fell blends between 4500-4680A and ~5100-5500A. For each region, the underlying continuum was represented by a first order polynomial and the flux was summed from the data after subtracting this contin­ uum. For E0132-411 and E1227+140, the 4500-4680 A blend appears to be very strong and is blended with a feature that lies beneath the H/9/[OIII]AA4959,5007 lines which may also be due to Fell emission. This was represented by a gaussian profile and subtracted from the data before the flux and equivalent width of the remaining iron features were measured. Fell blend information is presented in Table 5.3. E0Q39 —019 2 = 0.35 E 0 1 1 4 —002 2 = 0.04 _ E0129 — 066 2 = 0 .2 2 Data Optical iue .: pia pcr fteUS AGN. USS the of spectra Optical 5.1: Figure O 2 > o • » o *» o C O 2 O0 o0 o> ° ? * < < * A * A > 0 • c it fi E C < w It ft.0 ■ M b e E • • • c • • • c ft. it o E • •

=t i I I I 2 9 O0 o> « * o 9 £ £ * *# < A % * < < c u • ft.0 E • c c > b i. E • 0 > 0 c 0 e w c it ■ t. o • o It m

85 • X Chapter EO057+5O1 • z»1.41 . E1028+310 • z-0.28 . . El 148+888 • saQ.44 Optical Data Optical Figure 5.1: 5.1: Figure continued. 1 t • e • • oo «• «• ne «• at © © © o i i Ig i i i i u u 1 1 t I L I J H 1 u n x ) i l l 3 * ^ ------C A U 1 ------MO H r •*© H 87 Chapter E1304+342 zao.20 , El 340-1-200 • z~0.02 El 423-*-201 • z b Q .2 1 Optical Data Optical I I did mu I I « A Figure 5.1: 5.1: Figure jljijm&n inu continued. * o A i i i i ,.uij tin mu i i I

89 2

totlv l M*j •ui al y lu ta iu u • jt * |M » ilu q •tuoJtaluv » »|» * « ji a u * o 90 Chapter 5 Table 5.2 : USS AGN Line Param eters. u Upticad H g 00 D • * • D D D O • C P • O j a CO O H 2 • CM P • • O f H ID CD ID • • 0 0 ^ 2 CO • * © • © cod®® N CD . P o c • W« H • S H l 2 D l O ^ O.S T 5 X J O CD f-"T CO © © — r CH H1 CO rH I—( © V ~ t H , V V os N H V V « o e cd M M M H M H D M H H CO rH rH y g O CM ® O I/O ID rH © © 2 ° O ; _ O CM © ^ CM CO ® ^ ^ CO rH H ^ CM ® H CM CM g © ^ ' *"< ^ © ■<*« 2 ° CM £ ° ° d lO tjH 00 ^ N <0 © 5 0 O O 2 CO © CO © Tjc © Hjc O CM ^ N 2 ^ 3 0 £ o c CD O CO VVH VVr * V° « © « VV rH V **-° ^ HM V V O g £ £ S | = 2 d d ® ® Dot” N®«5 d d ^ ^ P d ® ® rH* « ® ® d P cd ^ 2 ^ S ^ d H N d g 5 « « d d ® d g ® N d OCD^S ” t ® d g d o D ® l d d ® ® d ® W d 2 <=> d d B O © H d d © 2 d d © ® © d d 2 © d d H © > d ® g r g r^ g > t ® 0 0 d .>* © A n a © £ - r rH a—I c ^ m © aM -M* CM © rH 1 m ® a— O 1 * © ^ n a d O M r » a-t ® O “ ^ - K O l ® * N " a d N - ® k ^ O d c N d O ^ J P ® _ rH ® © i m £ i © o - N W V “ ® 0 CO VJ. 1 03 J H | rH I I I I _ ^ V o I ^ I I r I I I I - 1J H _ 3 . CD 0 g © —1e . g J V 1 - S CT

* F- * ^ n CN O H W r OO W W «NW^W W5N ” ^ w t: o r o wt: W5N OO^ ” «NW^W W W Wr H CN n ^ O F-* — t- D H M O H O D M o M O © bO © CM o b O O C O H r bo CM CD O LO rH CO © CM rH © ; 5 0 0 CD t«- ►—I H rH rH r~r ^ ffi w 3 M , . . - H © rH N- © . . . . , , © © CM © „ ---- t >0 «b. rH ID CO O CO CO ■'f O 0 5 CO ID O C O b - r f i O © © © bO N - - N bO © © © O i f r - b O C O ID CO 5 0 O ■'f CO CO O CO ID rH «b. O O O O oC b h oo o - o o o h bO CO bo bO O CO 0 0 3 0 ^ N CO N 3 0 CO - f w ! n

© © © © © © © © o O r w w w w w w w o I I I + I © I - N I CM I 03 I rjc I © j T ^ H CM rH © 5 t yj- ata H rH CN oo H H H © f H © rH rH rH o H M © o CM rH © O OO © O OO © rH 03 O © © © cd © - N © © H H iyj © © rH © © OO rH © OO cm © © O rH H H H H H H rH rH rH rH rH rH rH v ) X C H r V V V V V V V V ^ © ^5 VV ° i ii i i ii i i ii i i i ii i i i iii H 2 H2

CM d OO d © © ID © rH VV 5 H V 2 O OO oo

CM © OO co CO © rH O © © © CM r> rH © rH rH co N- © ©

^^ ii N N o U-t O © © © O © CM © - N © © rH rH CM © © © rH OO CO h h

d d d® ^ ®^ ® ® ^ ^ ® cd® d o t d d d d M« CM CD © N; CM cd CM h CM © CD d © rH 1 cd rH d rH \/ V V \/ V V V V V V V V V CM —1 OO H H ^ rH rH cm 1 o « t

d d cd 03 © CM © CM Is* © rH © © HrH rH d rH }< V CO CO -^ ...... bON- ^ CO ^ CO ^ H CO 3 0 bO co O b bO O CO N OO N ^ ^

© o H CO rH rH CO H - N h ° rH ^ ^ rH N- 0 0 © © © rH © © CM © © rH rH rH rH ID rH CM CO ^ ^ ^ ‘H CM S H CS rH CS - ® ! O H O® f- ^ O H I- CD g ® d d r-J d 2 ® g O *u CM CO j

© © o o © V co*

• M® O C ^© CD ® ^ • O © CM©®® • O

CM O O b O 00 0 O bO iO © O b O CO 0 0 O O H H N O O O N rH00 CM CM 00OID O00 rjc O O OO ^ © O j H Nh Nh H j O •'tf H © kI k © - N rH © 0 0 2 .CD

^ ® ^ ® ^ d M O CM ©^ ®^S5 5 S ^ ® ~ I ' O ^ o O H H M< rH ^ rH 3 0 ~ O 5 U O • - • CO O • I O • Is- • O • ^ O CM CM rH rH - rH 1-1 t *« it? CO OO H 'X

V 1 3 2 5 s i J o> 1""* V oo> cs io J o H CD O rH ^ O D l M C n CO © © d CM rH CM t ID CM O . H • H rH • • rH © . • © 2 " © © co H© g © ^ r © ^ . . ^ © rH ^ ^ © g © rH 00 CON « M , CM J« ^ © © o f o o O O © cm O CM O CM 0 rH © 03 d CM 5X1 N o o N WO ° ^ 510(0 510(0 ^ ° © © / \ i 0 0 M C / \ © o ? « r? H ° V 2 t ° ° co © © t f CM d © © ® CM rH CO

rH rH O © O rt< rH [s. © O rH © rH rH 03 H ®

cm oo h CM ° 1 d H d O rH rH * ° ° V v v CO oocm 0 0 1 1 d ® 2 cd

© ^ © I - D C . ^ ^ 22 ^ d © © rH © CM CM

OO tjh CM © d 01 ° d CM OO 1 o o © ^ o o ^ ^ ^ 0 CO rH rH CM 0 rH

N- 03 CM O 3 0 3 0 O © \/ V V d d ° 2 o 2 CM 0 0 © O b I'll £ n n . 0 0 ^ ^

N NH N- © © rH rH HrH rH 00 CD h 1 a J n i V —a 3 0 *£< © rH © rH © © © 0 rH t>» © OO co ™ o o

^ ^ N ^ CM © N CM © CO CM © Tt< rH £ O © h h

° 03 rH d d CO d ^ 00 00 © © 0 0 ^ N- ^ © O b . .

© © cm © © h rH rH rH rH rH *«

0.79 — 0.46 — 0.38 0.34 0.31 0.30 0.29 0.26 0.25 0.24 0.19 — 0.25

2170 — 1110 — 1280 1420 1840g 1250g 1320g 2300 870 840 2390 — 2120 92 Chapter 5

c o t—i o 00 O o CM CD o CO HH * * O 00 CD 05 CD to CM CD . lO Tj5 O CD C-’ o n O d rH o O © - d d rH CM 05 o CM rH I—I CD O t " CO CM CO?s CO CD '---- ■ O rH d

co e O N (O O 00N- 05o CO 05o 05 N 0 0 60 rj< o n n COO t- CD t- to o ^ if) O °LO t•j u h oo to rH CM H O M rH 05 rH o 00 CD o rH (—1 CM ® 3

CM TjJ to o co 05 co rH CO «o o— d 0 M O -cf Cm W N O O r-H CM 05 O co to *** tS

r- t—t CD 00 o 05 CMo Cdoo CO o rH to o 60 00 o o H • o co to rH N- • tO CO to co 05 • rH rH oo d 00 o 00 to o o CD o oo d co o co CD d rH CO

0 5 t i to 0 5 oo o <0 0)00 o> co . co o CO Tjt o 60 © O O to K co o t—I M tO rH O ■ tO CO tO OO HPJ'I® ,H • cm ^ ©£ 00rH t o o c o rH co d rH O h S ° »

rH 0 ° cd ; © o CO rH » 00 to « S o c d . o O • O Cd to • • o 0 0 go o oo O 05 d d ® © . O > 0 5 : co O CM . CO o O O o V V ° rH

6 0 CO tO 60 05 CO T*t **2 0 0 CO ^ 6 0 CO T$< *JH 6 0 0 ^ O 00 • • W rH • tO O o CO W to CO o . r-i 0 CO CM 05 O N .05 ^ c o g S CO O 2 O CO rH d o to O H . H ® CM CM O CM . CM O CM O CM v v ° ^

05 N- CO CO 0 5 CO O oo © oo o to 00 T*< HH CO 60 CO rj< t o *JH co co co to to CM tO O rH tO ® ^ :• © -!cd g S O 05 — to g o d o w • ° o • CO © v • « V O CM ° CO V o

O tO CM CO CO O to *** © ^h CM ^ 0 5 O O O i o tj« to CO o J r l 0 5 ^H tO co • CO CO S w Cd ^ d g S 0 5 d to o « 2 0 0 O CM - I d g S d . o v • ° CO o w • ^ CO V ° v <=> CM \ / o ^

0 5 CO CO O o rH tO 05 Cm to 0 0 o o CMoo O o 0 5 CD © rH N- © oo E rH CO CM co o CD Tt< to CO CO CD to rH CM c o n H tO«.° O rH d o CO co o 05 to O O O d d to d s § d rH d o V oo v V rH . o d d o « £ V V

CO -v^ (0 t > o u « s v HH « V CO CM OO o oo to © © to © Cm © 05 CM © rH rH Cm © to oo © © CD rH CM © © rH © rH N- © CM CM a to CM rH © CM iO CM rH c t to to w. 0 5 o ■<* © © CM © © © © © £ 0 © © © tO © r-4 O v 05 © d rH u V rH © © o tS V V © P h CO CO CO o oo 00 o V CO° ^i> o J H N CM CM rj< Tjt 4) ° rH . oo O rH a CO £ * CM hJ oo a CO © o © © o CO © © © © CM © CM c o OO CM o CO CD © 00 CO © CM CD _* © 0 0 CM CO o S o rH CD CD CM CM O rH rH rH © o rH M5 o CO to CM 0 0 © CO < rH o © © cd CD CO o tO P d d d

tO cm a (0 CM o u CM »—* 0 0 oo CD CD ID 05 CM tO CO to CO CO tO + + + + + + + U CD oo CD CO tO 4) O to CM rH 05 05 05 o rH CM 3 3* O O o rH i—1 rH £ o W W W (3 (3 (3 (3 E1217+695 1.9 0.0 5.78 — <0.13 — <0.21 <0.32 <1.09 <1.27 <0.06 6.43g 0.16 1.31 <1.89 JLJ&ta. Upticad 9.9 — <0.5 — <0.8 <1.3 <5.8 <5.7 <0.3 40.6 l.Of 8.4 <15.9 0.58 — 0.28 — 0.26 0.25 0.22 0.22 0.22 0.16 0.16 0.16 0.12 2550g — 1700f — 1640f 4800f 4990f 4880f 1460f 5290g 1570f 1650 4800f w d co in co 0 0 rH in O d d CM cm © rH oo m c CO H rH rH rH CM CM rH CM H CO o> CO CM r~- CM in cm V + V V V V V V V V V I 1 1 i • • o> o> o CO CO CO o> co o rH 00 >03 o> co 00 N d rH d co V V V V i I 1 1 • d co CM d co CM CO CO CO d d d CM CM CO o CO d CM cm co CO CO CM d d CM CO d rH o CMo o> 1 1 1 | co CM C C <4H o> rH CM C CM C O oo o O O t- CO O 00 © O CM oo O 03 CO o O rH 0 0 rH O O CM rH rH rH 00 rH rH CM O rH CO CO rH rH CO rH rH m m m m 60 1 1 1 1 w oo d © rH CM d d in od oo CO CO d CM CM oo 03 CM CM o o CM cm 00 CM i CO rH rH rH in o o OC CO CO CO o o ooCO ^ © V V + V V V V V V V — t 60 Tt< 0 0 6 60 © 0 6 I 1 1 1 I i

'- l 00 CM rH 00 03 05 CO rH d > t d CO in rH CO rH rH o rH rH CM CO V I 1 1 1 rl O rl o d 00 CO co l> d - r d CM o CO d 03 CO 03 o d d CO oo d d o rH d CO 1 1 1 I C C C C 03 OO C CO O © CM o CO O . s J CO eo o t- o 00 eo © o CO CO 03 © © CO CO O 00 o o co t- rH O O rH x* co rH m m m m 60 60 m m 60 I 1 1 w 00 © CM lD rH CMd rH <7> 00 in 3in 03 CM co x* x* CO in CO 00 rH in CD CO rH CM 00 CM co in H CO ” N ” CO H CM rH o 5 oo ^ V V V + V V V V V v v O eo v v 600 1 1 1 1 • CM d d in oCOcq CMO co H rH rH lO CM CO 00 CM CM rH 03 © 03 o V V V V 1 1 1 1 o o o oo d O o M CM CMo rH o d oo in d rH d X* co d d in CM CO 03 i 1 1 I C C CO O m c C C oo O 03 rH rH o O O 03 O O CM m 03 00 co 03 03 O CM oo O rH rH o rH rH rH rH co rH o rH rH x* rH rH rH m m m m 60 060 60 1 1 1 1 d d 03 d 00 w CM CM d d rH CM 03 CO rH CO CO oo d rH CO o o CM T* in d 1—I CO Tt«CM rH inin oo o rH + V H ° V V V 60 1 1 I 1 1

CM OCO CO 03 d © CMco oo 00 HjC CO 03 rH © CM in CM rH CM < CO x* in V V n 1 1 I 1 r1 CM o o CO rH d co o CM o o co d co 03 d x* 03 o 00 CO o d d o rH in CM i 1 1 1 I ( Q O CM O 03 o CO O CM CM O O O co f x CM rH oo o CM CM Tt< CO O rH rH 03 rH m o in m m 1 1 1 © d w co © d rH 00 oo CO CM d o d d o rH rH 03 CO CO CM CO o 00 rH CO O C CO CO • CO rH - 2 10 2 r-5 O O N ^ V + V toco 1 1 1 © rH d rH CO oo qCMcq in rH rH 00 co 03 rH V V 1 Q - b o CM o O CM CO d 00 d o o C" 00 o d oo o d co o o 03 1 1 I SCM m c C 03 C O CM O o rH c-~ 0 0 CM 03 03 O O rH 00 O rH CO rH o rH rH © rH iH m m 1 03 rH (3 CM CMin rH cq 00 CO xtf rH CM o rH CO rH m CO ncq in rH 00 CM CO © x* o 00 CM S rH O O H 0 0 oo rH CM + V V V V Y y ° 600 60 1 1 l 1 1 • rH cq qCMcq rH CM d CM rH rH CO in rH Me in rH in in > l O x* rH V V 1 1 1 1 03 o cq rH 03 00 CM CO qo cq rH rH rH o CM rH rH rH 03 rH rH CD rH CM 03 © rH rH 1 i 1 1 «M < o t- m c O 03 co o o 00 Mcq CM 03 O CM O rH 03 CO o co O CM 00 03 CO 03 t'- rH O rH co rH co o in o t" O rH n to rH rH rH V m 60 1 1 1 1

d d d co in W CO o d d CM d 00 rH cq m CD 03 03 o o CO m d rH d h qin cq rH rH rH rH CM m rH CO o> CM CO d CM o + V V V 60 00CM6000 60 *! 1 1 1 1 00 d 3 0 o rH CD CO in d rH CMo in CD CM 00 CM o CS CM rH m d rH rH X* CO rH V V V 1 1 1 I o o d 00 o CO 00 CM CM o 00 o o 03 o d d o o d o oo O OO o oo d d o 0 0 d CM o o rH rH rH CO d o CO 1 1 1 1

^ 0 0 o 0 5 OO lO O b»co i-Hco O « o ^J J O© (O 05 ® ” s to >—■ o ^ o © LO © o» 0 0 o rH xj* CM xt* o x* 50 tot - <—> CO • © rH CO

CO o O CM 0 0 o CM O 05 CM tO O 50 ^ to bO J N H -# 50 xt* CO ^ 05 to O <£> ffl xt* • b- 50 © oo lO ” o A N y—Si *■ rH 50 xt* eo xt* rH N O xt* to ~ xt* • CM to

to o o / CO to 05 o toeo O

(ft H o CM o o oo b- rH Cm xt* o xt* Cm 0 0 o 05 Cm CM o o IO O OO O CM to O c d id 0 o id lO O rH to CM lO oo W r*H CO cd o 50 rH cd CM VV O VV xt* V to o lO o CM V o rH

b* to CM CM o CM CM io «XH io b^ xt* o to eo xt* O CM CM rH O rH b- o to rH 0 5 I > to O b- CM o 05 CM rH id xt* CM b~" o to xt* t*H CO rH o co oo o to CM cd rH rH cd rH rH rH rH 50 o rH O o o

i > CO CM 6 0 xt* xt* o t+H 05 iO xt* *M 50 to Xt* o OO CO CM o O o CM tn rH CM O oo CM rH b- xj* CM t* OCM y-H xt* CM fH xt* CM to eo 05 cd o LO O rH O V rH O rH 00 VV cd rH o rH o o V o •*» H-3 4) T 3

rH «X o o i o eo o tO rH 50 O rH CM xt* xt* CO xt* © H-j ^ to M CM id oo io CM• j: xj* H . CO • b- to «=! r-H 00 ^ i ►— \ 5 ' rH CM a! 5 o 00 o

6 0 60 05 O N O XX 60 6 0 xt* c m CM 50 CM Cm xt* CM oo Cm crteo *—* ' © * * *-H O O O O g t o ^ * CM to b - lO O eo CM t> to CM 05 05 CM * rH cd cd CO 05 o IO O Xjfl v V • n rH . CO VV oo cd V LO v v O rH O o rH cd V

Cm cf -2 60 60 05 *£* CM rH 60 60 CO Cm to CO CM 60 6 0 b- 60 O <15 3 *- o O • O O rH 50 00 CM g LO b- IO rH lO O x ^ « W xt* © 00 CO CM id CO xt* 05 o xt* *OT -*3 xt* xjj o 0 5 V CM eo 00 'I OO xt* CO m h j CM rH V V o CO o CM cd d rH c d V ’§ g 45 O d CM *o 50 CM xt* Cm to CO 00 c m b- rH b- Cm «* © 0 0 o O O o cm e o CO O oo t> b- »^4 q w CO CM cm' » 05 rH id O lO 0 rH ■*» • IO CM "O o td t> VV 50 cd V oo Pm v V d CO O CM cd rH 45 V V 45 CO ' o i-H rH O 00 rH O Cm 00 00 00 Cm 05 05 o c m J 3 > O O * • CO rH CO O io CO t- a w rH CO d ^ d » xt* cd CM CM 05 CM 0 o TS o cd t> VV xt* cd V to a o V V d CO cd CM V cd rH ^ «> g 3 * a 3 00 *t* xt* <£* xt* CO CM Cm oo xt* o c m CO o eo Cm g s d • • CO ® CM O CO O to O «H £ s o n i m ^ s xt* b - o Px (0 ^ 3 p « s o V 4) -M V 05 CM tO OO rH 50 to 'X 50 CO t o o lO CO oo • • eo ° rH * rH <7? rH CM CO b- g CM O O . rH b-’ xt* CM i d S (0 • ° CM CO CM t - o V V ° 5rt O W • CM CO o v fn CO v O rH cd cd d (0 V O 45 -m A h o . s 8 ^ V eo > g a xt* 45 o CO • H CO to s 60 w *H 45 CO d £ 00 t-H rH xb 05 o 50 05 xt* o -----x _f> £ 5-H rH xt* 05 to i o 00 U5 • < - g P

Table 5.3: USS AGN Optical Fell Blend Parameters

Object name FeII&/ue FeIIred Ratio H(3 ratio 4500-4680A 5100-5500A FeII&/ue/FeIIred FeIIfc/ue/H/? FeIIre<*/H/?

E0039-019 *32 30 1.1 1.4 1.3 2 25 31 3 1.3 0.95 E0114-002 <28 < 4 4 — < 1 .2 < 1 .9 < 1 9 < 2 8 1.5 1.6 E0129-066 38 65 0.59 1.8 3.1 46 86 0.86 0.75 E0132-411 23 14 1.7 1.2 0.74 79 52 0.29 0.26 E0337-267 6.6 < 4 .7 — 1.1 < 0.79 29 < 2 0 0.23 0.23 E0436-433 < 3 1 < 2 7 — < 11 < 9 .8 < 2 3 < 1 8 1.3 1.5 E0844+377 6.9 11 0.61 0.20 0.33 27 55 0.26 0.20 E0906+111 < 11 < 1 9 — < 0.83 < 0.76 < 3 3 < 6 2 0.32 0.30 E0944+464 2.8 3.6 0.76 1.3 1.8 24 38 0.12 0.095 E1028+310 <2.0 < 1 .9 — < 0.90 < 0.85 < 3 9 < 41 0.052 0.045 E1146+558 <0.49 < 1.1 — < 1 .2 < 2.6 < 9 .0 < 2 2 0.054 0.049 E1213+378 < 0.82 — < 0.32 < 1 7 0.050 E1215+692 <0.72 — < 6 2 0.011 96 Chapter 5

Table 5.3 (cont.) : USS AGN Optical Fell Blend Parameters.

Object name FeII&/ue FeIIred Ratio Hj3 ratio 4500-4680A 5100-5500A FeIIfc/ue/FeIIred FeIIf,/ue/H/? FeIIred/H/?

E1217+695 10 20 0.51 1.6 3.1 59 160 0.17 0.13 E1218+693 < 7 .3 < 7 .6 < 2 .9 < 3 .0 < 2 5 < 2 5 0.29 0.30 E1227+140 18 < 1 3 > 1.5 2.4 < 1 .7 50 < 3 6 0.37 0.36 E1228+123 <12 < 1 6 < 1 .3 < 1 .7 < 31 < 4 2 0.38 0.38 E1304+342 9.6 12 0.84 0.71 0.84 31 50 0.32 0.23 E1346+266 1.9 0.98 30 0.063 E1423+201 40 41 0.97 0.75 0.77 30 42 1.3 0.97 E1425+169 2.4 2.2 1.1 0.52 0.47 31 31 0.078 0.069 E1511+671 15 11 1.3 0.89 0.68 62 56 0.24 0.20 E1654+352 4.5 0.73 36 0.12 E1657+326 < 5 .9 < 8 .5 < 2 .4 < 3 .5 < 1 8 < 2 5 0.33 0.34 E1805+700 < 2.8 < 3 .2 < 1.2 < 1 .4 <20 < 2 5 0.14 0.13

1 (Flux in units of 10“ 15 ergs cm-2 s—1) 2 (Equivalent width in A) 3 (Continuum flux in units of 10-15 ergs cm-2 s-1 A-1, may include Fell emission, Balmer continuum, etc.) Opticeil Data, 97

Table 5.4 : Spectral classifications of the USS AGN.

Object Name Redshift Log Lv FWHM H (3 Stephens Present km s-1 Classification Classification

Secure USS EO111-015 0.12 39.31s 1130s NL Sey 1 EO132-411 0.27 40.46 1930 NL Sey 1 E0845+378 0.31 40.18s 3450s Sey 1 E0944+464 0.35 40.23 1320 NL Sey 1 E0957+561 1.41 41.75 3250* Double quasar E1028+310 0.25 39.92 1780 NL Sey 1 El 146-1-558 0.44 40.15 5180 Sey 1 E1227+140 0.10 39.82 3290 Sey 1 Sey 1 E1346+266 0.92 40.64 1840 NL QSO E1401+098 0.43 40.76s 8270s quasar E1423+201 0.21 40.78 2750 QSO E1425+169 0.22 39.76 2520 Sey 1 E1519+279 0.23 39.99s 2210s Sey 1

Non-secure USS E0039-019 0.35 41.21 1520 NL QSO E0114-002 0.04 39.69 2980 Sey 1 E0129-066 0.22 40.51 1310 NL Sey 1 E0150-102 0.36 40.34s 3400s quasar E0337-267 0.11 39.71 1340 NL Sey 1 E0436-433 0.07 40.14 2180 Emission line galaxy E0844+377 0.45 40.74 2220 Sey 1 QSO E1008+348 0.14 39.69s 3310s Sey 1 E lO ll-f 496 0.20 BL Lacw E1059+730 0.09 39.91s 6630s Sey 1 E1215+692 1.12 40.13 5830* Sey 1 E1217+695 0.63 40.81 5260 Sey 1 E1218+693 0.11 39.84 1410 HII region galaxy E1352+183 0.15 41.01s 4540s quasar E1511+671 0.31 40.47 1650 NL Sey 1 E1640+537 0.14 970 NL Sey 1 E1657+326 0.09 39.73 570 Sey 1.9 E1805+700 0.19 39.89 1700 NL Sey 1

Non-USS E0906+111 0.16 40.17 3780 Sey 1 E1213+378 0.82 40.44 3310 Sey 1 E1228+123 0.12 40.01 1680 Sey 1 NL Sey 1 E1304+342 0.28 40.42 2060 Sey 1 Sey 1 E1654+352 0.80 40.84 3620 QSO

Notes to Table 5.4 : * FWHM of MGIIA2798 ( E/3 out of range). s from Stephens (1989). W Classification from Wisniewski, Sitko and Sitko (1986). 98 Chapter 5

5.1.2.4 Continuum parameters. A power-law model was fitted to each spectrum having first removed the emission features. No attempt was made to subtract other contributions to the optical con­ tinuum such as contaminating starlight, Balmer continuum emission and residual blended Fell emission. This optical power-law index, a opt, is an indication of the overall observed shape of the spectrum and may not accurately reflect an intrinsic underlying power-law in the optical region (but see Section 6 .2 .2 ). A power-law was not an appropriate model for some low redshift objects where the contribution from the underlying galaxy is strong towards the red; E0436-433 is an extreme example of this and was not measured. Optical power-law indices are listed in Table 5.2.

5.1.2.5 AGN spectral classification The AGN have been classified using the criteria in Stephens (1989). According to this scheme, a ‘quasar’ appears stellar, it has very broad permitted lines (with FWHM typically several thousands of km s-1) and an optical continuum luminos­ ity in the V band, log Lv > 40.60. A ‘QSO’ is a radio-quiet quasar. A ‘’ has a stellar or semi-stellar nucleus and log Lv < 40.60. There are two main types of Seyfert galaxies; 1 and 2 . ‘Seyfert 1 ’ galaxies have broad permitted lines with FWHM up to 1 0 4 km s - 1 and narrow forbidden lines. In ‘Seyfert 2 ’ galaxies, the permitted and forbidden lines have similar widths of about 500 km s - 1 and a ratio [OIIIJA5007/H/? >3 (Shuder and Osterbrock, 1981). A narrow-fine (NL)

Seyfert, quasar or QSO is defined as one which has an HFWHM < 2 0 0 0 km s- 1 (Goodrich 1989). For the purpose of this spectral classification only (in order to be consistent with Stephens), I define the V band optical continuum luminosity L„, as the luminosity at 5500 A (assuming Ho =75 km s - 1 Mpc - 1 and qo— 1 ). The luminosity was derived from the flux of the power-law continuum model (fitted to the rest- frame spectrum) at 5500 A, except in the case of E0436-433 where the continuum flux was measured directly from the spectrum. The flux was corrected for fight lost through the slit (see Section 5.1 for details) before the luminosity was calculated. The Hf3 FWHM has been deconvolved from the instrumental profile by assuming that the observed width is the quadrature sum of the intrinsic B./3 and instrumental FWHM, and that these fines have a gaussian profile. Instrumental FWHM were measured from the calibration arc spectra, at 14.5 A for the FOS and 16.0 A for Optical Data 99 the FORS. There are two objects, E0114-002 and E0436-434, whose H/? lines were measured with the IPCS; it was not necessary to deconvolve these lines due to the higher resolution of this detector (see Section 5.1). The optical continuum luminosity at 5500 A, deconvolved H/? FWHM and the classification for each spectrum are listed in Table 5.4. Also listed are details from Stephens (1989) for those USS AGN which she observed. (The observed FWHM for the USS AGN are listed in Table 5.2.)

5.2 CCD IMAGING

Direct CCD images of the sources listed in Table 5.5 were obtained at the Isaac Newton and Jacobus Kapteyn Telescopes of the Observatorio del Roque de los Muchachos in March and May 1989 respectively. The filters used and exposure times are given in Table 5.5. Observations of flux standards and sky flatfields in red and blue filters were taken at the beginning and end of each night. All of the images were corrected for bias and flat-field variations using image processing software developed at the Milliard Space Science Laboratory.

5.2.1 Magnitudes

Each source was fitted with a gaussian profile superposed on a first order polyno­ mial to represent the sky background. Source counts were summed after subtract­ ing the background, and count rates were converted to magnitudes by comparing them with the count rates measured for the flux standards. The AGN magni­ tudes have been corrected for differences in the airmass between the object and the standard. Integrated magnitudes in B and R filters, which include flux from any underlying galaxy, are presented in Table 5.5. Errors on the magnitudes are ~ 10%.

5.2.2 Extended sources

A measurement of the spatial extent of the AGN in these CCD frames is also given in Table 5.5 as an indicator of the presence of an underlying galaxy. This is the g of the gaussian fitted to the AGN (see Section 5.2.1) in arcseconds,

Other details, eg. morphology of the underlying galaxy, and some of the images themselves may be found in C JL cy7.

Table 5.5 : CCD Observations of the USS AGN. Object Name Date Telescope Filter Exposure Mag* a gal Time (s) n E1028+310 10 May 89 JKT R 1800 18.6 < 0 .4 El 146+558 09 May 89 JKT R 1800 18.7 0.8 E1213+378 29 Mar 89 INT R 300 18.3 < 0 .2 99 99 INTB 300 19.0 <0.3 E1215+692 29 Mar 89 INT R 300 19.1 < 0 .3 99 99 INTB 300 20.1 < 0 .4 E1217+695 29 Mar 89 INT R 150 17.1 < 0 .2 99 99 INTB 150 17.5 < 0 .2 E1218+693 29 Mar 89 INT R 500 16.9 1.2 99 99 INT B 500 18.4 1.2 E1228+123 13 May 89 JKT R 1800 16.8 1.0 E1304+342 29 Mar 89 INT R 300 17.8 < 0 .3 99 99 INT B 300 18.3 < 0 .2 E1511+671 09 May 89 JKT R 300 17.9 0.3 99 99 JKT B 400 18.4 0.3 E1640+537 29 Mar 89 INT R 300 16.8 0.6 99 99 INT B 100 18.8 0.9 E1654+352 09 May 89 JKT R 1000 17.5 < 0 .2 99 99 JKT B 1000 17.9 < 0 .2 E1657+326 29 Mar 89 INT R 300 17.1 1.2 99 99 INTB 300 18.5 1.4 E1805+700 29 Mar 89 INT R 300 17.7 0.8 99 99 INT B 300 19.2 0.6

* Integrated magnitude of the AGN (includes flux from the underlying galaxy). ** Gaussian sigma of the AGN in arcseconds <75a/, deconvolved from the seeing profile. Upper limits are given where the AGN profile is point-like.

5.3 OPTICAL LUMINOSITIES. In computing the optical luminosities of the USS AGN I have assumed values of 50 km s-1 Mpc-1 for the Hubble constant, Ho, and 0 for the deceleration parameter, qo. Optical luminosities between 3000 A and 6000 A, Lopt, and at 2500 A, L2500A? in the rest-frame of the AGN are calculated and are listed in Table 5.6. The broad band optical continuum luminosity, Lopti is defined as the lumi­ nosity between 3000 A and 6000 A in the AGN rest-frame. The optical continuum Uptical Data 1U1 flux was measured from the power-law fit to the spectrum where available (for E0436-433, the flux was measured directly from the spectrum). To determine the fraction of light lost through the slit, ‘magnitudes’ determined from the spectra were compared with the accurate CCD magnitudes (see Section 5.2) when both were available (13 objects) and the spectral flux was scaled up accordingly. These scaling factors are listed in Table 5.2. On average, the flux measured from the spectrum was 1.8 ± 0 .2 times lower than the corresponding CCD flux. Where CCD magnitudes were not available, the average scaling factor was used. In cases where I do not have spectra, optical luminosities have been deter­ mined by converting the V magnitude listed in Table 2 .1 to the flux at 5500 A using the equations in Allen (1973), then integrating over the appropriate range assuming a power-law index of 1 .0 (the average optical index measured in my sam­ ple of spectra), having made corrections for the effects of redshift. I have compared luminosities derived using both of these methods for the sources for which I have spectra, and find an average difference of 1 0 %.

The rest-frame flux at 2500 A, (f2500A)> was measured directly from the spec­ trum, or, when this region was not observed, by extrapolating the optical power- law fit to the continuum and scaling using the factors in Table 5.2. In the case of objects for which spectra were not available, f 25ooA was calculated from colour magnitudes using the equations given in Schmidt (1968; these extrapolate the flux to 2500 A by assuming an optical power-law index of 0.7 over the range of 1500 A to 3700 A in the rest-frame of the AGN). If more than one observed colour fell within this range, I have taken the weighted mean value of f 2500A • 102 Chapter 5

Table 5.6 : USS AGN Optical Luminosities

Object m y hopt L2500A &OS &OX z

(1) (2) (3) (4) (5) (6) Secure USS

E0111-015 17.70 43.83 28.22 -0.14 > 1.12 0.12 E0114-016 19.29 44.13 29.27 -1.07 1.21 0.38 E0131-408 19.8 45.25 30.19 0.89 2.36 E0132-411 17.47 44.50 29.59 0.36 1.65 0.27 E0136-250 17.8 44.54 29.69 0.71 1.27 0.31 E0310-557 17.0: 44.62 29.78 0.75 1.26 0.23 E0331-365 18.0 44.46 29.62 0.26 > 1.28 0.31 E0845+376 18.03 44.45 29.88 -0.33 1.47 0.31 E0944+464 18.6 44.32 29.40 0.21 > 1.25 0.35 E0957+561 16.7 46.23 31.40 > 1.35 1.41 E1028+310 19.50 43.92 28.57 0.10 *1.26 0.25 E1040+123 17.29 45.65 31.43 1.43 1.03 E1146+558 19.61 44.17 29.24 -0.19 > 1.17 0.44 E1208+322 16.71 45.45 30.28 0.10 1.31 0.39 E1227+140 17.3 43.70 28.65 1.16 1.26 0.10 E1251-005 18.45 44.53 29.67 -1.99 > 1.48 0.43 E1254+221 16.98 44.48 29.64 0.80 1.55 0.19 E1255+220 18.50 44.12 29.29 1.93 > 1.62 0.26 E1255+017 18.43 43.77 28.94 0.98 1.36 0.16 E1334+038 17.72 43.92 29.09 1.44 1.44 0.13 E1346+266 18.5 44.94 30.57 > 1.38 0.92 E1401+098 16.6 45.46 30.74 0.22 1.37 0.44 E1423+201 16.0 44.86 30.39 0.57 >1.76 0.21 E1425+169 19.1 44.28 28.26 0.42 *1.27 0.22 E1519+279 18.2 44.15 29.81 0.27 > 1.75 0.23 E1614+308 18.01 44.36 29.51 0.56 1.38 0.27 E1704+608 15.28 45.69 30.89 0.63 1.62 0.37 E2034-220 17.8 44.40 29.56 -0.41 1.27 0.26 E2318-423 18.2 44.09 29.71 0.88 1.52 0.21

* Hard X-ray spectrum poorly fit by the model. : Value uncertain. Optical Data 103

Table 5.6 (cont.) : USS AGN Optical Luminosities

Object my h opt L2500 A &OS Otox z Non-secure USS E0007-357 16.0 44.29 29.47 2.21 1.66 0.09 E0039-019 16.1 45.34 30.46 -0.50 > 1.67 0.35 E0114-002 15.9 43.38 28.34 1.19 *1.66 0.04 E0129-066 17.1 44.58 29.81 0.41 > 1.34 0.22 E0141+020 14.16 43.58 28.77 2.08 1.96 0.02 E0150-102 18.1 44.54 29.69 0.36 E0200-089 16.52 45.75 31.21 > 1.65 0.77 E0337-267 15.5 43.53 28.39 0.91 > 1.27 0.11 E0436-433 16.0 43.91 28.64 1.66 > 1.43 0.07 E0S44+377 17.3 44.96 30.39 0.32 1.46 0.45 E0952+442 17.28 45.08 30.31 0.39 1.26 0.47 E1008+348 17.62 44.01 29.18 0.99 1.22 0.14 E1011+496 16.15 44.86 29.92 0.60 > 1.82 0.20 E1059+730 16.32 44.13 28.92 1.72 > 1.42 0.08 E1215+692 19.78 44.53 30.01 > 1.25 1.12 E1217+695 17.5 45.10 31.04 > 1.56 0.63 E1218+693 17.69 43.76 28.93 0.11 E1352+183 16.41 44.52 29.61 0.41 1.21 0.15 E1423+242 17.2 45.35 30.69 > 1.33 0.65 E1511+671 17.7 44.59 29.83 -0.05 > 1.68 0.31 E1640+537 18.1 43.95 28.90 0.38 > 1.22 0.14 El644-029 17.91 44.28 29.69 0.26 E1657+326 17.50 43.47 28.55 1.42 1.33 0.09 E1805+700 18.49 43.84 28.74 -0.94 > 1.01 0.19 Non-USS E0906+111 16.93 44.10 29.17 -0.76 > 1.05 0.16 E1213+378 18.62 44.79 30.07 > 1.12 0.82 E122S+123 17.10 43.93 28.90 0.82 > 1.02 0.12 E1304+342 17.97 44.62 29.74 0.88 > 1.46 0.28 E1654+352 17.58 45.18 31.25 > 1.80 0.80

(1) V band magnitude (see Table 2.1). (2) Logio of the optical continuum luminosity integrated over 3000 to 6000 A in erg s—1 (see Section 5.3). (3) Logio of the optical continuum luminosity at 2500 A in erg s~ 1 Hz-1 (see Section 5.3). (4) Ratio of optical to soft X-ray component luminosities (see Section 5.3.1) corresponding to the 7.85 eV observer frame model. (5) Ratio of optical to hard X-ray component luminosities (see Section 5.3.1) corresponding to the 7.85 eV observer frame model. (6) Redshift. 104 Chapter 5

5.3.1 Optical to X-ray luminosity ratio The optical to X-ray ratio luminosity is parameterized by a ox which, following Tananbaum et a1. (1979), is defined by:

log—^———— = 2.605 x a ox L2keV

^ 2keV is the monochromatic luminosity of the hard component at 2 keV and L 2500A is the optical monochromatic luminosity at 2500 A in the AGN rest-frame. I have defined a corresponding index for the soft component luminosity, a oa which is given by:

log—7 -^2500A = .1.605 x a oa Lo.2keV where L0 .2 keV is the monochromatic luminosity of the soft component at 0.2 keV. Values of a ox and a os are also listed in Table 5.6.

“We’ll let the sunshine in and shine on us, because today we’re happy and tomorrow we’ll be even happier.”

— DAN QUAYLE CHAPTER 6

Infra-red Images of USS AGN

Figure 6.1 : The UKIRT K image of E0129-066.

“Hawaii has always been a very pivotal role in the Pacific. It is IN the Pacific. It is a part of the United States that is an island that is right here.”

DAN QUAYLE 106 Chapter 6

Infra-red images of USS AGN were obtained in 1991, February 6 and 7 using the United Kingdom Infra-Red Telescope (UKIRT) on Mauna Kea, Hawaii. The data were taken with IRC AM 1, a 62x58 array of InSb photodiodes on a pixel scale of 0.62" pixel-1. A summary of the observations including the filters used and exposure times are given in Table 6.1.

E1423+201

Figure 6.2 : The mosaic UKIRT J image of E1423+201. A very faint source lies just to the left of the AGN.

Due to the high background flux at infra-red wavelengths and significant vari­ ations in the background on timescales of ~1 hour, it is essential to obtain accurate Hatfields for each object (Casali, Aspin and McLean, in “IRCAM User’s Guide”). This was achieved by taking a ‘mosaic’ of images (see Figure 6.2 and Figure 6.3). Each object was observed five times placing the object at different positions on the array each time. The flatfield was obtained by median filtering the five frames. Dark frames and observations of flux calibration standards were taken throughout both nights. All images were dark-subtracted and flatfielded using STARLINK software at the Joint Astronomy Centre, Hilo. Infra-red Images of USS AGN 107

Table 6.1 : Infra-red Observations of the USS AGN.

Object Name Date Filter Exposure Mag* Nucleus z Time (s) % E0129-066 7 Feb 1991 J 1000 15.79 34 0.22 6 Feb 1991 H 900 15.23 48 K 500 14.41 57 E0337-267 7 Feb 1991 J 800 15.11 45 0.11 H 900 14.59 47 K 500 13.97 44 E0844+377 6 Feb 1991 J 1500 16.19 70 0.45 H 900 15.59 79 K 500 14.89 83 E0845+378 7 Feb 1991 J 1500 15.84 42 0.31 H 450 15.34 44 K 500 14.45 48 E0944+464 6 Feb 1991 J 1500 16.75 54 0.35 H 900 16.16 68 K 500 15.61 87 E1028+310 6 Feb 1991 J 2000 16.80 74 0.25 H 1200 16.05 75 K 500 15.22 67 E l 146+558 7 Feb 1991 J 2000 16.78 61 0.44 H 1200 16.08 61 K 500 15.35 70 E1227+140 6 Feb 1991 J 1000 15.22 52 0.10 H 600 14.62 53 IC 500 14.04 60 E1346+266 7 Feb 1991 J 2250 17.48 > 7 8 0.92 H 1200 17.15 > 7 1 K 500 16.41 > 8 6 E1352+183 7 Feb 1991 J 750 14.81 61 0.15 H 1000 14.13 67 K 500 13.08 74 E1423+201 6 Feb 1991 J 1000 14.64 57 0.21 H 600 13.70 65 IC 500 12.70 71 E1425+169 7 Feb 1991 J 1500 16.88 65 0.22 H 600 16.43 6 Feb 1991 IC 500 15.60 45

* Integrated magnitude of the AGN (includes flux from the underlying galaxy). 108 Chapter 6

2.° E0337—267

X j

Figure 6.3 : The mosaic J image of E0337-267 which is the central ob­ ject in the field. Infra-red emission from the underlying galaxy is strong relative to the nucleus and is clearly seen on this image.

6.1 IR DATA REDUCTION AND ANALYSIS

The counts were summed in two orthogonal directions across each AGN, along the major and minor axes of the host galaxy where these could be determined. Two gaussians were fitted to each resultant profile; a narrow gaussian representing the point-like nuclear component and a low, broad gaussian for the underlying galaxy. A first-order polynomial was fitted to the sky background. Total (point plus extended) source counts were summed after subtracting this background; the contribution to the total counts from the extended source was calculated after subtracting the fitted narrow gaussian component. The percentage of the integrated flux attributed to the nuclear component in separate J, H and K filters is given in Table 6.1 (this is the average of the percentages calculated along the major and minor axes). No underlying extended component could be detected for E1346-f266, which is a high redshift (z=0.92) AGN; in this case, lower limits to the nuclear percentage are given. Count rates were converted to magnitudes by comparing them with the count rates measured for the flux standards. The AGN magnitudes have been corrected Infra-red Images of USS AGN 109 for differences in the airmass between the object and the standard. Integrated magnitudes in J, H and K filters, which include flux from any underlying galaxy, are presented in Table 6.1. Errors on the integrated fluxes are estimated to be

~ 1 0 %, and errors on the nuclear fraction calculated from the IR images are ~10- 20%.

6.2 MULTIWAVELENGTH SPECTRA The IR colours have been combined with optical spectra (where available, see Chapter 5) and the X-ray colours, Cl, C2 and C3 (see Chapter 4), to produce spectra covering 4 decades in wavelength. These are plotted in Figure 6.4 and have been redshifted into the rest-frame of the AGN (although no redshift correction has been made to the flux). X-ray counts were converted to fluxes using factors derived from the two- component observer-frame model (with a blackbody kTeff=7.85 eV and power-law

<*= 0 .7 ; see Chapter 4 for more details). Upper limits to the counts in C2 and C3 are indicated by an arrow and vertical bars on the flux correspond to the lcr errors in Table 4.1. The horizontal bars indicate the extent of the IPC Cl, C2 and C3 energy ranges. Optical spectra have been corrected for light loss due to the narrow slit (for scaling factors see Table 5.2) but the underlying galaxy has not been subtracted. IR magnitudes were converted to fluxes using calibrations given by Beckwith et a 1. (1976). The IR spectra also include flux from the underlying galaxy. For E0845+378 and E1352+183 where optical spectra were not available, B and V colours taken from the literature are shown. Also for E1352+183, a UV spectrum taken using the short wavelength (SWP) camera on IUE is included.

6.2.1 Multiwavelength fits The composite spectra have been fitted with a three-component IR-to-X-ray con­ tinuum model, comprising a UV/soft X-ray blackbody with a kTeff of 50 eV, a galaxy component in the IR/optical, and an underlying IR-to-X-ray power-law. The composite model is plotted as a thin solid line in Figure 6.4 and the separate components as dashed lines. Since the data only reach into the near-IR, no IR dust component (which may be significant at A > l^m ) was included in the fitting. Three major problems beset these multiwavelength fits; (a) the non-simultaneity of the data (there are years between each dataset), (b) the non-photometric optical spectra (all of the spectra were taken with a narrow slit) and (c) the poor quality 110 Chapter 6

Table 6.2 : Results from multiwavelength fits

Object Power-law lfim nuclearfraction Redshift Notes index multi-A IR

E0129-066 0.64 0.53 0.41 0.22 IR low E0337-267 0.72 0.39 0.45 0.11 IR excess E0844+377 0.70 > 0.89 0.79 0.45 opt/U V + C3 e E0845+378 0.78 > 0.81 0.44 0.31 E0944+464 0.71 0.84 0.61 0.35 E1028+310 0.65 0.63 0.75 0.25 IR + C3 excess E1146+558 0.65 0.38 0.61 0.44 IR + excess E1227+140 0.77 0.53 0.52 0.10 C3 excess E1346+266 0.59 > 0.90 > 0.86 0.92 opt/UV excess E1352+183 0.62 0.52 0.61 0.15 opt low E1423+201 1.12 > 0.76 0.57 0.21 opt/UV excess E1425+169 0.70 0.72 0.65 0.22 IR + C3 excess

of the X-ray data. The effects are obvious in some cases, eg. for E0129-066 and E0337-267, the IR data do not join smoothly with the optical continuum. Other effects may not be as clear; for the X-ray data there is no way of assessing any possible offset with respect to the optical and/or IR. With these problems in mind, the following assumptions were made when fitting the spectra.

6.2.1.1 The soft X-ray excess The (observer-frame) blackbody is only intended to be a representation of the soft X-ray excess so that the IR-to-X-ray power-law may be fitted. A blackbody with the ‘best-fit’ temperature of 7.85 eV (see Section 4.2.3) is unrealistic in the optical and UV, therefore the effective temperature was fixed at 50 eV merely to represent the soft component and not to fit it (see Chapter 4 for details of fits to the soft X-ray component). The multiwavelength plots show that a blackbody with this temperature has little effect in the optical/UV continuum which is dominated by the power-law component (except in cases such as E0129-066 and E0845+378 where the soft X-ray excess is particularly strong). It is not a good fit to the optical/UV big blue bump and no attempt has been made to fit this feature.

6.2.1.2 Underlying power-law

The index of the underlying IR-to-X-ray power-law was allowed to be free in the fitting procedure, and was constrained by the minimum flux in the IR/optical the fits shown in Figure 6.4 are listed in Table Table in listed are 6.4 Figure in shown fits the spectrum and the the and spectrum nr-e mgsofUSA N AG USS f o Images Infra-red

For two objects, E1227+140 and E1425+169, there is a significant excess in excess significant a is there E1425+169, and E1227+140 objects, two For L°§ v Fv Log vFv Log vF, otnu mdl cmrsn a ot -a bakoy ih Tf=0 V galaxy eV, power-law. kTeff=50 IR-to-X-ray with underlying blackbody and 1988) X-ray Kriss soft from a (taken comprising spectrum model, continuum infra-red J, H and K integrated fluxes with optical spectra (where available) and the and available) (where spectra optical with fluxes integrated K and H J, infra-red three X-ray colours; C l, l, C colours; X-ray three Figure 4 1 - 10 -1 14 -1 -10 4 -1 -12 -10 14 6.4 6.4 E0844+377 E0337-267 E0129-066 Mliaeegh pcr o h US G cmiig h integrated the combining AGN USS the of spectra Multiwavelength : C2 and and 516 15 C3 C2 dt. oe-a nie (, ie yF = = F„ by given (a, indices Power-law data. and and C3. lte a a hn ie s h three-component the is line thin a as Plotted Log v Log .2 6 - and Table 6.3. Table and 8191718 v a ) for for ) a

111 112 Chapter 6

E0845+378

-1 4

E0944+464

_ _ _ \

-1 4

E1028+310

- 1 4 L 14 1516 17 18 19 Log v

Figure 6.4 : (continued.) the C3 flux over the underlying power-law; there is also marginal evidence for a C3 excess in the spectra of E0844+377 and E1028+310. When considering the implications of these fits to the X-ray data, it should be borne in mind that the IPC data in general are poor for the USS AGN (see Chapter 4); the errors on the X-ray fluxes, plotted as vertical bars, correspond to \o errors on the IPC counts. imra-rea images 0 1 a u i\ 110

El 146+558

-10

-12 .+ + +

-1 4

E1227+140

-1 4

E1346+266

- 1 4 14 1615 17 18 19 Log v

Figure 6.4 : (continued.)

6.2.1.3 The host galaxy contribution

The host galaxy was modelled using the spectrum in Kriss (1988) as a template. In general when there was an offset between the IR and optical data, the shape of the optical continuum was used to assess the galactic flux. However, in the case of E1352+183, an optical spectrum was not available thus the galaxy component was fitted to the IR colours. The additional UV data were used to further constrain 114 Chapter 6

E1352+183

Pu> - 1 0 - > bJO

-12 K

-1 4

E1423+201

-1 4

E1425+169

-10

-12 + +

- 1 4 15 16 17 1814 19 Log v

Figure 6.4 : (continued.)

the fit to the nuclear continuum. The two optical points for E1352+183 are low relative to the IR and UV - 1 have assumed that this is due to variability and have not included them in the fitting procedure.

The nuclear fraction at 1 /rni, calculated from these fits after subtraction of the host galaxy contamination, is compared with that derived from the two-gaussian fits to the IR profiles (see Section 6 .1 ) in Table 6 .2 . Lower limits are given in cases mira-rea images ui u j j /iw y 110 where no galaxy contribution was required. In general, there is good agreement between these two methods for separating the nuclear flux, including cases where the IR and optical data do not join smoothly, eg. E1352-f-183 and E0337-267. The notable exceptions are E0845+378 (no optical spectrum is available for this object so the galactic contribution is poorly constrained in the multiwavelength fits) and El 146+558 (the IR image is very faint and the underlying galaxy could not be well-determined).

THE IR EXCESSES Apparent IR excesses above the multiwavelength fits are seen in several objects

(see Table 6 .2 ). For E0337-267, E1028+310 and E1425+169, the nuclear fractions at 1 /im calculated independently from the IR and optical data agree, therefore the excess may be due to inaccurate scaling of the optical spectrum. There is less agreement for E1146-f558; in this case the galactic fraction derived from the UKIRT data (61%) is much higher than that derived from the multiwavelength fits (38%), and may account for the apparent excess.

6.2.2 Results from the multi wavelength fits 6.2.2.1 Continuum power-law slopes In order to measure the change in slope of the underlying continuum from the optical to the IR, I have fitted power-laws in the optical and IR for both the nuclear and total spectra. The power-laws were fitted to the optical and IR separately; for the nuclear IR spectra I subtracted the galactic contribution calculated from the UKIRT data, and for the nuclear optical spectra I used the galactic fraction from the multiwavelength fits. The power-law indices are given in Table 6.3. Also listed is the index for the underlying IR-to-X-ray power-law.

6.2.2.2 The big blue bump

Only in three cases out of the 1 2 shown (E0844+377, E1346+266 and E1423+201) is there firm evidence for the big blue bump in the optical data, which would be seen as an excess above the underlying power-law. E0844+377 and E1423+201 also have very strong soft X-ray excesses above the power-law, suggesting a link between the big blue bump and the 0.25 keV-excess. A comparison of the underlying power-law index with the optical nuclear index provides further evidence for the presence of the big blue bump. For E0129-066, E0844+377 and E1423+201, the slope of the nuclear continuum in the optical is 116 Chapter 6

Table 6.3 : Optical and IR power-law indices

Object IR-to-X-ray Optical IR total nuclear total nuclear

E0129-066 0.6 0.8 0.2 0.8 1.5 E0337-267 0.7 2.1 1.5 0.3 0.2 E0844+377 0.7 0.0 0.0 0.6 0.8 E0845+378 0.8 0.5 0.9 E0944+464 0.7 1.0 0.8 0.7 1.1 E1028+310 0.7 1.7 0.7 1.1 0.8 E1146+558 0.7 1.3 0.5 0.7 0.9 E1227+140 0.8 1.6 0.8 0.5 0.6 E1346+266 0.6 0.4 0.4 0.3 0.3 E1352+183 0.6 1.3 1.5 E1423+201 1.1 -0.1 -0.1 1.5 1.8 E1425+169 0.7 1.4 0.8 0.7 -0.2 significantly steeper than the slope of the underlying power-law. For E l346+266 the optical slope is not significantly steeper and the apparent optical/UV excess may be due to inaccurate scaling of the optical spectrum.

6.2.2.3 Evidence for dust emission IR-to-optical spectra of the galactic nuclei (ie. having subtracted the galaxy com­ ponent), combining optical spectra derived from the multi wavelength fits with the IR fluxes calculated from the UKIRT images, are shown in Figure 6.5. This dia­ gram, and an examination of Table 6.3 show that in some cases, the nuclear IR slopes flatten or rise to the red relative to the optical, indicating the presence of a separate component due to dust emission. The IR spectrum of E1346+266 in particular shows quite a dramatic ‘dip’ at lp m . A dip is also clearly seen for two of the big blue bump candidates E0129-066 and E1423+201 (which rise longwards of 1/zm), but not as well for E0844+377 (although the spectrum does flatten in the IR). Note that the hard X-ray component is not significant in E0129-066 and E1423+201, but strong in E0844+377 (in C3); a relatively dominant underlying power-law in E0844+377 may be masking a separate IR component. For E0337- 267 and E1028+310, the spectra fall either side of lp m . This may be because the galactic contribution has not been completely subtracted (but note that the nuclear fractions at lp m calculated independently from the IR and optical data agree within the bounds of error). mira-rea images ui sikjuv

I r

Jv^iEO129-066 .

E1227+140 + + + ^ ^ l l E0337' 267

Hh > to I—Ho wAvjS^Qg44+377

I E0944+464 11423+201

E1028+310 E1425+169

+ - + -H- .+ \JU» J______I 14.0 14.5 15.0 14.0 14.5 15.0 15.5 Log v Log v

Figure 6.5 : Nuclear IR-to-optical spectra of the USS AGN for which IR colours and optical spectra are available. The contribution from the galaocy was calculated from the multiwavelength fits for the optical spectra, and from the UKIRT IR images for the IR fluxes. 118 Chapter 6

6.3 LUMINOSITIES Nuclear IR fluxes at lp m and 1.65 pm in the rest-frame of the AGN have been calculated by linear interpolation from the J, H and K magnitudes and these were converted to luminosities assuming Ho=50 km s- 1 Mpc - 1 and qo=0. These luminosities, together with luminosities at 2500 A (see Chapter 5), 0.2 keV and

2 keV (see Chapter 4) are listed in Table 6.4. Also listed is the index of a power-law fitted to the three nuclear IR colours, a m (where F„ oc v~ aiR).

Table 6.4 : Infra-red Luminosities.

Object Name z L(l/im) L(1.65//m) L2500A ho.2 keV L2 keV OtlR

(1) (2) (3) (4) (5) (6) (7) E0129-066 0.22 29.55 30.02 29.81 29.88 <26.29 1.5 E0337-267 0.11 29.38 29.49 28.39 26.55 <25.06 0.2 E0844+377 0.45 30.26 30.39 30.39 31.11 26.55 0.8 E0845+378 0.31 29.88 30.07 29.88 30.14 26.04 0.9 E0944+464 0.35 29.75 29.95 29.40 29.28 <26.13 1.1 E1028+310 0.25 29.59 29.77 28.57 28.43 *25.28 0.8 E1146+558 0.44 29.94 30.15 29.24 29.55 <26.19 0.9 E1227+140 0.10 29.32 29.47 28.65 27.47 25.37 0.6 E1346+266 0.92 30.13 30.56 30.57 <26.94 0.3 E1352+183 0.15 29.88 30.16 29.61 27.88 26.08 1.5 E1423+201 0.21 30.24 30.64 30.39 29.59 <25.69 1.8 E1425+169 0.22 29.39 29.34 28.26 28.30 *25.49 -0.2

(1) AGN redshift. (2) Luminosity of the nuclear component at 1pm in erg s_1. (3) Luminosity of the nuclear component at 1.65/zm in erg s—1. (4) Luminosity at 2500 A in erg s-1 (see Table 5.6). (5) Luminosity of the soft X-ray component at 0.2 keV corresponding to the observer frame kTe//=7.85eV model in erg s—1 (see Table 4.2). (6) Luminosity of the hard X-ray component at 2 keV corresponding to the observer frame kTe//=7.85eV model in erg s—1 (see Table 4.2). (7) Slope of a power-law fitted to the infra-red data.

* Hard X-ray spectrum poorly fit by the model. CHAPTER 7

Notes on Individual Objects

7.1 A BRIEF INTRODUCTION This section begins with a more detailed look at the secure USS AGN, E0132-411. Following in R.A. order, are notes on the remaining objects. AGN marked with an asterisk (*) are secure identifications.

7.2 E0132—411 This object (z=0.267) was independently identified by Kriss (1982) and was also discussed by Cordova et al. (1989). It was contained within a deep Einstein IPC field (effective exposure time 28.5 ksec). Consequently, though it is not an in­ trinsically bright source, (IPC count rate 0 .0 1 c s-1) the X-ray spectral data on this object have good statistical quality. It exhibits a soft overall X-ray spectral distribution with little evidence of a hard spectral component. X-ray spectral fits using all of the Einstein IPC PHA channels may be found in C92. E0132-411 is also relatively bright optically (V=17.4), therefore an accurately flux calibrated optical spectrum of this AGN was obtained in order to investigate its multiwave­ length spectral distribution. It has also been observed with IUE as described in Section 7.2.2.

We’re going to have the best-educated American people in the world.

— DAN QUAYLE 120 Chapter 7

7.2.1 Optical spectrum The optical spectrum of E0132-411 is shown in Figure 7.1. The continuum rises

towards the blue and is fitted with a power-law index = 0 .8 (in F^). The Balmer lines are strong and relatively narrow (FWHM H/?=2080 km s-1) and are visible down to He. Forbidden lines of [Oil], [OIII], [Neill] and [NeV] are also seen but [01] and [SII] are weak. Perhaps the most striking feature is the presence of strong red and blue optical Fell blends. Fell emission between 4500 and 4680 A is

particularly strong; it extends blueward beyond H 7 and blends with another Fell feature which lies beneath H/? and [OIII]AA4959,5007. At 1.7, E0132-411 has the highest ratio of blue to red Fell flux of the sources fisted in Table 5.3.

2.5

2.0 [OH] Ha

H-edge

1.0 |H5 [Offl] He I [OIII] Hel [SII] 0.5

Fell Fell Fell 0.0 L_ 2000 4000 6000 70003000 80005000 Wavelength Angstroms

Figure 7.1 : The optical spectrum of E0132-411 taken at the Anglo-Australian Tele­ scope (AAT), plotted in the rest-frame of the AGN. No correction for redshift has been made to the flux. The expected positions of optical emission lines typically seen in AGN are indicated.

7.2.2 UV spectrum Two observations of E0132-411 were made in the ultraviolet using the short wave­ length camera (SWP) on the International Ultraviolet Explorer satellite ( IUE). The first spectrum was taken on 1990 January 3 at the Villafranca Tracking Sta­ tion (VILSPA) and the second on the night of 1990 June 4 at the NASA Goddard Space Flight Center. Both were taken in low resolution mode and covered the Notes on Individual Ubjects 121 range of 1150 A to 1980 A with a resolution of 5 to 8 A pixel *. The two weighted spectra have been added together and the result is shown in Figure 7.2.

15 Sill Silll

OVI 10 NV

OI |+SiIV 5

1100 1200 1300 15001400 Wavelength Angstroms

Figure 7.2 : The UV spectrum of E0132-411. This spectrum is derived from the weighted sum of two exposures taken with IUE and is plotted in the rest-frame of the AGN. No correction for redshift has been made to the flux. The expected positions of UV emission lines typically seen in AGN are indicated.

The UV spectrum shows strong Lya emission at 1216 A (F ~ 2 x 10-13 ergs cm - 2 s-1) and possibly Ly (3 at 1026 A (F ~ 2 x 10-14 ergs cm - 2 s-1 ). The continuum flux at 1200 A (in the rest-frame) is ~ 2 x 10-15 ergs cm - 2 s- 1 A " 1 and the continuum between 1000A and 1500A was best fit using a power-law with an index of 0.7 (c.f. 0.8 in the optical).

7.2.3 Multi wave length spectrum The optical and ultraviolet data for E0132-411 are combined with the X-ray data in Figure 7.3 to produce a multiwavelength spectrum. The X-ray count rates have been converted from counts per second to flux using the two component spectral model that has a blackbody temperature of 10 eV (in the rest-frame of the AGN) and a hard power-law index of 0.7. This model is plotted as a thick solid line in Figure 7.3. Fitted models with blackbody temperatures of 15, 25, 50 and 100 eV are also shown for comparison, plotted as thin solid lines. Note that as the blackbody temperature increases, the normalization of the hard X-ray power-law decreases, 122 Chapter 7

thus the 10 eV model fit provides a reasonable upper limit to the flux in the hard X-ray spectrum.

-10

-11

> _12 PL, > Ot>0 J -13

-14

-15 14 15 16 17 18 19 Log V

Figure 7.3 : The multiwavelength spectrum of E0132-411. Plotted as a solid line are the optical spectrum taken at the A AT (also shown in Figure 7.1) and the UV spectrum taken with IUE (also shown in Figure 7.2). The X-ray spectrum is represented by a two component model, a soft blackbody with kTefF=10eV in the rest-frame of the AGN and a hard power-law with an index of 0.7, and is plotted as a thick solid line. Also plotted, as thin solid lines, are fitted models with a rest-frame kTeff of 15, 25, 50 and lOOeV (see Section 7.2.3). The three X-ray data points represent the counts detected in the C l, C2 and C3 bands, and were calculated using conversion factors from counts to flux derived from the lOeV two-component fit. The multiwavelength spectrum of the Seyfert galaxy Mkn 841 (from Arnaud et al. 1985) is plotted as a dashed line for comparison.

The spectral distribution of E0132-411 is compared with that of the Seyfert galaxy, Mkn 841 (z=0.037) using data taken from Arnaud et al. (1985). The spectrum of Mkn 841 has been redshifted to the same distance and normalized to the optical spectrum of E0132-411. The relative distribution of flux between the optical, ultraviolet and soft X-ray bands is very similar in the two objects. The major difference between them is that the hard X-ray component in E0132^411 is depressed by more than an order of magnitude relative to Mkn 841. lyotes on individual uojecis 1 £6

7.3 THE REMAINING USS AGN

* = secure USS source

E0039—019 z=0.35 : This is a new USS identification which I have classified as a narrow-fine (NL) QSO. No hard X-ray component is detected (ie. there are no significant counts in the C3 band). The optical continuum in Fa rises steeply to the blue (optical power-law index in Fj,, a opt = 0.3). The optical spectrum shows narrow Balmer fines and MgIIA2798, and strong red and blue optical Fell blends. It appears point-like on the quick V finding chart.

E0111—015 * z = 0 . 1 2 : This X-ray selected object, classified by Stephens (1989) as a NL Seyfert 1 , has a strong hard X-ray component. The V magnitude is given in the EMSS as 17.70 (from CCD photometry) and by Stocke et al. (1983) as 19.2. The Balmer fines are strong down to He and the [OIII] fines are also strong. The optical continuum is flat. Blue and red Fell blends are present but the red is weak. It appears slightly extended on the finding chart.

E0114—002 z=0.04 : This is a new USS identification, which I have classified as a Seyfert 1 , and has a strong hard X-ray component. The optical continuum is flat (a opt = 1.9) and shows strong features from the underlying galaxy. Ha is relatively narrow compared to H /3 which may be contaminated by underlying Fell and features from the host galaxy; [OIII] and SIIAA6717,6731 are also seen. It appears extended on the finding chart.

E0129—066 z= 0 . 2 2 : This is another new USS identification which is only weakly detected in X-rays with no hard component. The optical continuum rises slowly to the blue (power-law index = 0.8 in F„), the Balmer fines are strong and narrow and the optical Fell blend emission is also very strong. I classify this object as a NL Seyfert 1 and it is point-like on the finding chart.

E0131—408 * z=2.36 : An optically selected secure source which has by far the highest redshift of all the USS AGN. The corresponding range of the soft X- ray excess in the rest-frame of the AGN (ie. the Cl range), is ~0.5-1.9 keV. The counts in the 1 .9-3.6 keV(rest-frame for C 2 ) range are relatively low but strong again between 3.6 and 1 1 .8 keV(C3). It is point-like on the finding chart which shows no other objects closer to the X-ray position. Lya and CIVA1549 have been detected (HB). 124 Chapter 7

E0132—411 * z=0.27 : See Section 7.2.

E0141-f-020 z= 0 . 0 2 : This object is better known as Mkn 573, a Seyfert 2 galaxy. Strong ultra-soft X-ray emission is unusual in a Seyfert 2 nuclei; these are believed to have a thick, obscuring torus which lies edge-on to the line of sight (see Section 1.3.6), therefore we would expect all soft X-rays to be absorbed before reaching the observer. It is a very low redshift USS AGN and the hard X-ray component is strong. The permitted lines are very narrow; FWHM H/? =

300 km s- 1 and FWHM HeIA5876 = 460 km s - 1 (De Robertis and Osterbrock,

1986). Radio maps of Mkn 573 at 6 cm (Ulvestad and Wilson 1984) reveal a triple source whose outer radio components are separated by 3". The [OIII] A5007 line emission is aligned with the radio 6 cm emission, but [OIIIJA5007 emission at the position of the outer radio lobes is relatively weak (Haniff, Wilson and Ward 1988).

E0150—102 z=0.36 : Also known as PHL 1 2 2 0 , this is an X-ray selected source. X-ray counts are very weak in all three bauds. From the spectrum in Stephens (1989), it appears to be similar to E0845-}-378, ie. it has broad and low Balmer lines, the continuum rises steeply to the blue and the [OIII] lines are very weak. Classified by Stephens as a Quasar/QSO.

E0337—267 z= 0 . 1 1 : The signature of the underlying galaxy is strong in the optical spectrum of this low redshift USS AGN. The optical continuum is flat

(&opt = 2 .1 ) and the spectrum shows narrow Balmer lines and the presence of red and blue Fell blends. E0337-267 is a new USS identification and no hard X-ray component is detected. It appears point-like on the finding chart.

E0436—433 z=0.07 : The optical spectrum of E0436-433, shown in Figure 7.4, is dominated by light from the underlying galaxy which resembles those of A2029 and MKW3s, both elliptical galaxies which lie at the centre of clusters with cooling flows (Johnstone et al. 1987). Balmer line emission is weak and blended with the galaxy features; the presence of Fell emission is difficult to determine due to features from the host galaxy. It is a new USS identification and no hard X-ray component is detected.

E0844+377 z=0.45 : The number of X-ray counts detected for this object in all three bands is low, but the hard X-ray component is relatively strong. The optical continuum rises steeply to the blue ( a opt = 0 .0 ), and shows very strong and narrow emission lines. Blue and red Fell blends are weak and the red blend ivotes on maiviauai UDjecis

2.0 [Neill] Ha

[OIII] Hel

& 1.0

g E 0.5

o.o L_ 3000 40005000 6000 7000 8000 9000 Wavelength Angstroms

Figure 7.4 : The optical spectrum of E0436-433 taken at the Anglo-Australian Tele­ scope (AAT), plotted in the rest-frame of the AGN. No correction for redshift has been made to the flux. The expected positions of optical emission lines typically seen in AGN are indicated. is weaker than the blue. Mgll has a strong narrow component and an underlying broad component blended with Fell. Classified in this paper as a quasar/QSO and in Stephens as a Seyfert 1 . This source was X-ray selected. Hutchings et ai. (1982, hereafter HCCGM) report that of all the objects in their sample where they could detect ‘fuzz’, this has the highest redshift. They also report that it has an extended arc of nebulosity to the NE and appears to be in a cluster. Hutchings, Crampton and Campbell (1984, hereafter HCC) note that E0844+377 is irregular and may be interacting. They derive a corrected ratio of luminosity in the nucleus to luminosity in the surrounding “fuzz”, Ln:Lf, of 3.0.

E0845+378 * z=0.307 : This object has a weak hard X-ray component. The optical spectrum in Stephens shows strong Fell emission and the presence of [OIIIJAA4959 and 5007. However, Margon, Downes and Chanan (1985: hereafter MDC) report that [OIII] is absent but confirm that Fell is strong. Also from the spectrum in Stephens, the optical continuum rises to the blue and the Balmer lines are broad and low. HCC note that E0845+378 has a halo or broad tail and a corrected Ln:Lf of 0.9. It is an X-ray selected object.

E0906-+-111 z=0.16 : This object also appears in the EMSS and no longer 126 Chapter 7 meets the new USS criteria (see Section 2 .2 .2 ). No hard X-ray component has been detected with the IPC (ie. counts in C3 are not significant). The optical spectrum shows a flat continuum ( a opt = 1.9), relatively broad Balmer lines compared to the USS AGN (FWHM E/3 = 3860 km s " 1) and strong [OIII]AA4959,5007. Optical

Fell emission is very weak. This AGN is classified as a Seyfert 1 .

E0944+464 * z=0.35 : This secure source is a new USS identification and is classified as a NL Seyfert 1 . It has no significant hard X-ray component. The Balmer lines are very narrow (deconvolved FWHM H/? = 1320 km s-1) and the [OIII] AA4959,5007 lines are relatively strong. Red and blue Fell blends are seen in the spectrum; the blue blend in particular is strong. The optical continuum rises to the blue ( a opt = 1 .0 ).

E0952+442 z=0.47 : This AGN has a very strong hard X-ray component and was selected on the basis of its UV excess (HB). It also appears in the EMSS where the presence of Mgll is reported.

E0957-f-561 * z=1.41 : This radio selected object is a gravitationally lensed double quasar, the first lensed quasar to be discovered (Walsh, Carswell and Wey- mann 1979), and is variable (see HB). A detailed discussion of its soft X-ray emission may be found in C92. It has a high redshift and no hard X-ray com­ ponent is detected in the C3 band. The optical power-law index is steep ( a opt = 0.3) and MgIIA2798 is weak and broad with absorption features either side. The CIIIJA1908 line is broad and stronger than Mgll, and may be blended with under­ lying Fell emission. There is evidence of a strong and narrow CIVA1548 line, but it lies in the noise at the blue end of the spectrum. The two spectra were resolved by Wills and Wills (1980); both spectra show MgIIA2798 as strong as CIIIJA1908 and the lines have similar profiles, but the continua are flatter than seen in the USS spectrum (shown in Figure 7.5) and rise to the red. In a recent spectroscopic survey of the objects in the field of E0957+461, Gar­ rett, Walsh and Carswell (1992) have found that there are probably two clusters of galaxies in the foreground. The lens itself, a cD galaxy, is a member of one of these clusters (at z=0.36); both clusters may have a significant role in the lensing of the distant quasar (Garrett et aI. 1992).

E1008-+-348 z=0.14 : This X-ray selected object has a strong hard X-ray component in addition to the soft. Optical spectra are published in Stephens lvoies on muiviuuai kjujccis

3.0 j[SII]

2.5 [HeU |[°m] p i n

to | |[NeIII]

I He H-edge | 0.5

0.0 L_ 1500 2000 2500 3000 4000 45003500 Wavelength Angstroms

Figure 7.5 : The optical spectrum of the double quasar, E0957+561, plotted in the rest-frame of the AGN. No correction for redshift has been made to the flux. The expected positions of optical emission lines typically seen in AGN are indicated.

(1989) and Kriss and Canizares (1982) and there are no significant differences between the two. The optical continuum is flat (in F a). The Hj3 FWHM is relatively broad for a USS AGN (3310 km s-1) and optical Fell emission is not detected (Stephens 1989). Classified by Stephens as a Seyfert 1.

E1011-J-496 z= 0 . 2 0 : This radio-selected source has been identified by Wis­ niewski, Sitko and Sitko (1986) as a BL Lac object. The redshift of 0 .2 0 is uncertain and is based on possible membership of the cluster A950. Wisniewski et ai. give an olqx ~0.9, (cf. the value in this thesis of >1.82 which is calculated after the subtraction of the soft component) and an optical spectral index = 1.16. We note that, according to the scheme in the EMSS, this object would be classified as a galaxy, not as a BL Lac, by virtue of its unusually high a ox (ie. >1.82). It is seen in radio wavelengths as a compact core with a faint extended halo (Machalski and Condon, 1983). A slight extension may be seen on the finding chart. No hard X-ray component is detected.

E1028-f 310 * z=0.25 : The optical spectrum shows very strong and narrow emission lines; blended Fell emission is weak. This source has a flat optical con­ tinuum slope ( a opt = 1.7) and lines in the Balmer series are strong down to He. 128 Chapter 7

It appears point-like on the CCD R image. A new identification and classified in this thesis as a NL Seyfert I.

E1040+123 * z=1.03 : Other names for this AGN include 3CR 245 and 4C 12.37. This is a superluminal radio source (HB) which has a strong hard X-ray component. Radio maps at 6 cm and 20 cm by Saikia et al. (1990) reveal a triple source with a flat-spectrum nucleus and a jet towards the western component which is beamed and curved (Foley and Barthel 1990). The optical spectrum shows a relatively narrow MgIIA2798 line (FWHM = 3750 km s_1; Foley and Barthel 1990). Narrow permitted line widths and core-dominated radio sources are both associated with a disk-shaped BLR seen face-on (see Section 7.4. 1 .1 ). This source is variable in the optical and high frequency radio ranges (HB).

E1059-F730 z=0.089 : This is a weak X-ray detection but the hard component is relatively strong. It also appears in the HGLS and in the EMSS. From the spectrum in Stephens, the red Fell blend appears stronger than the blue. Ho; is strong but [Nil] is not seen whereas MDC note that the [Nil] emission is strong. The Balmer lines are broad and the H/3/fOIII] lines are badly blended possibly due to underlying Fell. The slope of the optical continuum appears flat. This

AGN has been classified by Stephens as a Seyfert 1 . It appears slightly extended on the quick V finding chart. The V magnitude is listed in the EMSS as 16.32 (from CCD photometry), and as 14.7±0.5 in Chanan, Margon and Downes (1981). HCC give a ratio of major to minor axis, b/a of 0.42 and a ratio, Ln:Lf of 0.57. They describe it as a spiral with an edge-on appearance (possibly due to local obscuration) that may be interacting, and a group/cluster member (the closest object, 15" away, is a peculiar C-shaped galaxy [HCC]). HCCGM add that it is one of the two most flattened systems in their sample and has dimensions typical of a large spiral galaxy. This is an X-ray selected source.

E1146-f-558 * z=0.44 : This secure USS AGN is a new identification and is classified as a Seyfert 1 . The hard X-ray component is weak and the slope of the optical continuum is flat ( a opt = 1.3). The Balmer lines are broad; only Ha is strong and H/9 is very weak and broad. Blended Fell emission is weak and may underly the H(3 and [OIII]AA4959,5007 lines. The [OIII] lines are also very weak. MgIIA2798 is very broad and stronger than Ha. The object is extended on the CCD R image. Notes on Individual Objects 129

E1213+378 z=0.82 : This relatively high redshift object is a new identification,

but no longer meets the new USS criteria (see Section 2 .2 .2 ). The slope of the optical continuum is steep (

E1215+692 z= 1 . 1 2 : This is another new USS identification with a high

redshift of 1 .1 2 and no significant detection of a hard X-ray component. The optical continuum is quite steep ( a opt = 0.5) and MgIIA2798 is broad and may be blended with underlying Fell emission. There is also evidence for CIII]A1908 emission but the spectrum is very noisy in this region. It appears point-like on the CCD R image.

E1217+695 z=0.63 : This is a new USS identification. No hard X-ray com­ ponent is detected. The optical continuum which rises steeply to the blue ( a opt

= 0 ), H/? is very weak and broad (it may be broadened by underlying Fell), [OIII]AA4959,5007 are very weak and MgIIA2798 is weak but narrow. Optical Fell blends are strong compared to H/?. It appears point-like on the CCD R image and many other galaxies are seen close by.

E1218+693 z=0.11 : This low redshift USS AGN is particularly interesting; it is classified in this paper as an HII region galaxy, yet it has a very strong hard X-ray component. These objects are extremely rare: NGC 838 (Mkn 1022) is the only other HII region galaxy among the Einstein serendipitous sources (Ward 1988). The source of X-rays from these objects is not well understood, but may be supernova remnants and massive pop I binary systems (Fabbiano 1986; Ward 1988). The optical spectrum of E1218+693, shown in Figure 7.6, shows a flat con­ tinuum (a opt = 2.0), strong and narrow Ha but H/3 and [OIII]AA4959,5007 are very weak. Optical Fell emission is not seen. It appears to be a face-on spiral on the CCD R image (shown in Figure 7.7); the arms are also faintly visible on the B image. Many other galaxies can be seen in the field.

E1227+140 * z= 0 . 1 0 : Two observations of this X-ray selected object appear in the USS AGN list, one is secure and the other no longer meets the USS criteria 130 Chapter 7

0.6

Hot

H8

Hel g 0.2 E

■el I 0.0 L 4000 5000 6000 7000 8000 Wavelength Angstroms

Figure 7.6 : The optical spectrum of E1218+693 taken at the WHT, plotted in the AGN rest-frame. This has been classified as an HII region galaxy, yet a strong hard X-ray flux has been detected. No correction for redshift has been made to the flux and the expected positions of optical emission lines typically seen in AGN are indicated.

Figure 7.7 : The CCD R image of E1218-f 693 taken at the INT. The scale is 90;/X90,r and the position of the USS AGN is indicated. iY U fc C O U ll J.11LLI V1U.UCU. KSUJCLLO l O l

(see Section 2.2.2). The two X-ray spectra are quite different; both have similar total count rates, but the spectrum of the secure detection shows a strong hard X-ray component whereas the non-USS detection has a weak hard component (see C92 for full details of the X-ray spectral variability). E1227+140 has a companion galaxy 8 .6 " away which is elongated and irregular (HCC). The optical spectrum is plotted in Figure 7.8. The continuum rises slowly to the blue (a opt = 1 .6 ), Ha and H/? are broad (FWHM H (3 is 3400 km s-1), and the [OIII]AA4959,5007 lines are strong. The red Fell blend is very weak whereas the blue blend is strong and blends with H (3. In Stephens, the optical continuum is flat, there is no red Fell and the blue blend is weak. The Balmer lines, particularly H/? whose FWHM was measured by Stephens as 4300 km s-1, are broad. Grindlay et aJ. (1980) found variability in their two spectra of this object which were taken only three days apart. The continuum of the second is more depressed to the blue of H/? than the first and they say that emission features in the H/3 profile and at wavelengths of Fell are marginally significant in the first spectrum but absent from the second. HCC note that this object may be irregular and it may have an associated jet or filament.

0.6

Ha 5 0.4 •f oE XA oe? ~o Hel

g 0.2 E

Fell

0.0 3000 4000 5000 6000 7000 8000 Wavelength Angstroms

Figure 7.8 : The optical spectrum of E1227+140 taken at the WHT, plotted in the rest-frame of the AGN. No correction for redshift has been made to the flux.

E1228+123 z= 0 . 1 2 : This X-ray selected object also appears in the HB cata- 132 Chapter 7 logue and the HGLS but no longer meets the USS criteria (see Section 2.2.2). A hard X-ray component is not detected. It has a flat optical continuum ( a opt = 1.6) and has been classified as a NL Seyfert 1. The spectrum shows narrow Ha and H/?; Fell emission is very weak. The CCD R image shows a blight nucleus with either a faint ring structure or extension and another extended object about 25 arcseconds away. There are no obvious differences between this optical spectrum and that published by Stephens (1989).

f ? u v<- >0 '

Figure 7.9 : The CCD R image of E1228+123 taken at the JKT. The scale is 60" X60" and the position of the AGN is indicated. Note the faint ring structure around this object.

E1255-f-220 * z=0.160 : The redshift of this AGN is listed in the EMSS as uncertain. It is a secure source with no significant hard X-ray component.

E1304-b342 z=0.28 : There is no significant hard X-ray component in this AGN, which was selected on the basis of its UV excess. Previously a non secure USS source, it no longer meets the new USS criteria. It appears in the EMSS, HB and Stephens samples. The optical spectrum shows a continuum which rises steeply to the blue (optical power-law index = 0.1 in F^), strong narrow Balmer lines, weak [OIII] and strong Fell (blue blend is stronger than the red). The optical Notes on individual Ubjects 166

spectrum appears similar in Stephens (1989). It is classified as a Seyfert 1 and is point-like on the CCD R image.

E1334+038 * z=0.136 : In the EMSS, the identification for this object is listed as ‘tentative’ (ie. it may be a galaxy, LINER etc.) and it is noted that H ft may have a broad base and that Mgll is marginally detected. It has a very strong hard X-ray component and is X-ray selected.

E1346-1-266 * z=0.92 : No hard X-ray component is detected in this source which also appears in the EMSS survey as well as the EXOSAT High Galactic Latitude Survey (HGLS). There is another HGLS source very close by; a z=0.60 AGN with a V magnitude of 18.1 which lies approximately 3' to the east. It has a lower EXOSAT count rate (0.0022 c s-1) than the z=0.92 AGN (0.0068 c s-1, Mittaz 1991) and is further away from the USS position. The slope of the optical continuum of E1346+266 is steep, ( a opt = 0.4), and the emission features are weak. It has been classified as a NL QSO/quasar on the basis of the low FWHM of H/9 (1903 km s-1) but this line is weak and at the red limit of the spectrum where the signal is relatively poor. The FWHM of MgIIA2798 is 2600 km s " 1 and may be broadened by underlying Fell emission.

E1352-+183 z=0.15 : This X-ray selected source has one of highest X-ray count-rates but a weak hard X-ray component relative to the soft. HCC report the presence of a halo or broad tail and a corrected Ln:Lf of 7.5. The spectrum in Stephens (1989) shows broad Balmer lines, weak [OIII] and blue Fell stronger than red (both are clearly seen). The optical continuum rises to the blue: it is also a PG quasar and the power-law index was measured by Neugebauer et ai. (1987) as 1.45 at wavelengths between 10.1 and 1.0 pm and 0.43 between 1.0 and 0.3 pm.

(see also the notes for E1704+608). Classified by Stephens as a Seyfert 1 .

E1401-+098 * z=0.44 : MDC report that this source is located near to the Zwicky cluster ZC 1400.4+0940 and 7’ from NGC 5438. The spectrum in Stephens shows an unusual H/? profile; it appears to be square-shaped and is blended with a feature that underlies the [OIII] lines and may be due to Fell emission. H 7 is also very broad but again this may be partly due to blending with [OIII]. The optical continuum rises to the blue, possibly due to strong Balmer continuum. Classified as a quasar or QSO. This is an X-ray selected source with a very strong hard X-ray component which shows evidence of X-ray spectral variability (see C92). 134 Chapter 7

E1423-f-201 * z=0.21 : This is a new identification of a secure USS source which I have classified as a QSO. No hard X-ray component is detected. The optical continuum rises steeply to the blue (aopt = -0.1), the Balmer lines are strong and [OIII] lines are very weak. Both optical Fell blends are very strong.

E1423-f-242 z=0.65 : Also known as 4C 24.31 and PKS1423T242, the redshift of this radio-selected AGN is uncertain. Although the Einstein count rate is rela­ tively high, the number of counts detected in all three bands is small and it has a weak hard component.

:v.v •: • / • # '

Figure 7.10 : The CCD R image of E1640+537 taken at the INT, showing the presence of many other faint galaxies lying close by. The scale is 150” X 150” and the position of the USS AGN is indicated.

E 1 4 2 5 + 169 * z = 0.22 : I have classified this AGN as a Seyfert 1 galaxy. It is a new secure USS identification which has a strong hard X-ray component. The optical continuum is mostly flat but rises to the blue at the blue end, possibly due to Balmer continuum emission (aopt = 1.4). Both red and blue Fell blends are seen but the red blend is weak.

E1511-f-671 z=0.31 : E1511+671, a new USS identification which is classified as a NL Seyfert 1, has no significant hard X-ray component. The optical continuum rises to the blue (optical power-law index = 0.0 in F„), the Balmer lines are strong ivoces on maiviauai UDjeczs

E1640+537.

* b a n / j e t ?— .* *o

0 10 20 30 40 50 arcseconds

Figure 7.11 : A contour plot representing the R CCD image of E1640+537. This elliptical galaxy also has an extended halo plus a faint bar or jet lying along the major axis (indicated on the diagram). The full CCD image shows many other galaxies in the field; at least three other faint galaxies can be seen in the cen­ tral arcminute shown in the diagram. Contours lie at [5220,5320,5500,5700,6200,7300,10000,20000] counts.

and narrow (FWHM H/? = 1800 km s_1), the [OIII] lines are weak. Both Fell blends are strong and some individual lines are resolved. It is slightly extended on the CCD R and B images.

E 1519+ 279 * z= 0 .2 3 : This AGN is X-ray selected yet there is no significant detection in the C3 band. It appears in the MDC and the Stephens samples. From the spectrum in Stephens, the optical continuum rises to the blue (this may be due to a strong Balmer continuum), the Balmer lines are narrow (FWHM H (3 = 2210 km s-1 ), [OIII]A5007 is weak and red and blue Fell blends can be seen. MDC note that Ho is strong and [OIIIJA5007 is very weak. It is classified by Stephens

as a Seyfert 1 .

E1614+308 z=0.27 : This is drawn from the EMSS catalogue and may be a BL Lac object. Lines of MgIIA2798 and H/? are reported in the EMSS.

E1640-J-537 z=0.14 : This is a new USS identification and has no hard X-ray component. The optical spectrum has poor quality; the continuum is flat ( a opt = 136 Chapter 7

Figure 7.12 : The CCD R image of E1657+326 taken at the INT revealing an extended object with many other galaxies close by. The scale is 150;/ X 150/; and the position of the USS AGN is indicated.

1.6) and there are very weak and possibly broad Balmer lines. The [OIII] lines are also weak and Fell is not seen. On CCD R and B images, the underlying galaxy appears to be elliptical with a faint bar or jet lying along the major axis (see Figure 7.10 and Figure 7.11). The galaxy is also surrounded by a halo which suggests the presence of a cooling flow.

E1654+352 z=0.80 : This is a new identification which no longer meets the USS criteria. No hard X-ray component was detected in this relatively high redshift object. The optical continuum rises very steeply to the blue (aopt = 0.0), H/l and H7 are weak and broad and the [OIII] lines are very weak. The blue optical Fell blend and the blends around the narrow MgIIA2798 line are strong. It appears point-like on the CCD R and B images and is classified in this paper as a QSO.

E1657+326 z—0.09 : A very strong hard X-ray component is detected in this low redshift AGN which is a new USS identification and classified as a NL Seyfert 1. The slope of the optical continuum is flat ( a opt = 2.0), Ha is strong and broad but H (3 is very weak and may be narrow. The [OIII]AA4959,5007/H(d and [SII]AA6717,6731/H/3 ratios are high, suggesting the presence of a cooling flow. Features from the underlying galaxy can be seen and are difficult to separate from Notes on Individual Objects 137

Figure 7.13 : The CCD R image of E1805+700 taken at the INT. There are at least two other galaxies within 40" of the object and many others are seen in the full CCD image (covering 41 X61). The scale is 3,X3/ and the position of the USS AGN is indicated.

both red and blue Fell features. CCD R and B images reveal that this Seyfert nucleus apparently lies within an elliptical galaxy which has an extensive surrounding halo (see Figure 7.12): further evidence for a cooling flow. There is another point-like blue object 16" to the east, ie. away from the X-ray position. Many other galaxies can be seen in the field.

E1704-(-608 * z=0.37 : This is another PG quasar (see Green, Schmidt and Liebert, 1986) and shows variability in its X-ray spectrum over a two-year period - see C92 for a full description. A low resolution spectrum from 0.3 pm to 10.1 pm is published in Neugebauer et al. (1987) and is fit with a broken power-law of a = 1.67 in the low frequency range (< lpm ) and 0.43 in the high frequency range (the same as for El352+183). It is a radio selected source and other names include 3CR 351 and 4C 60.24. HCC report that this object has a halo or broad tail (also like E1352+182) and an Ln:Lf of 4.0 in the R band and 14 in the B band. The QSO is more elongated in the blue than the red (HCC).

E1805+-700 z=0.19 : This is a very weak X-ray source with no detectable hard El 805+700

0 10 20 30 40 arcseconds

Figure 7.14 : Contour plot representing the R CCD image of E1805+700 (z=0.19). Note the elliptical shape of the galaxy surrounded by an extensive and almost circular halo. This group of objects, of which three are clearly extended, lies in a field crowded with other galax­ ies. Contours lie at [6370,6450,6600,7000,8000,10000] counts.

component. It is a new USS identification classified as a NL Seyfert 1 , although the optical spectrum also resembles that of a giant HII region. The optical continuum is flat (a opt = 1.8 in F^), Ha is quite strong and very narrow and H/? is much weaker. The [OIII] lines are also weak and both red and blue Fell are present (blue is stronger than red). CCD images show a field crowded with other galaxies (see Figure 7.13). The source itself has an elliptical core with an extensive, almost circular halo. There are five other objects within 40" of the source and at least two of these are galaxies (see Figure 7.14) but the identified AGN lies closest to the X-ray position.

E2318—423 * z = 0 . 2 1 : This is a secure USS AGN, which has a strong hard X-ray component. It is X-ray selected and is taken from the EMSS. CHAPTER 8

The Multiwavelength Analysis

This Chapter examines the X-ray, optical and IR properties of the USS AGN in detail. To search for relationships between the various line and continuum parame­ ters, I have computed linear correlation coefficients for each parameter pair in turn and these are summarised in Table 8.1 (X-ray and optical continuum correlations for the secure sources only), Table 8 .2 (IR continuum parameters of secure and non-secure AGN) and Table 8.3 (optical spectroscopic parameters of secure and non-secure objects). Only measured values were used in the correlations, although upper limits may be included in associated plots. No correlation coefficient is given in the Tables if the number of available data pairs (given in parentheses) was less than 5.

“Mars is essentially in the same orbit... somewhat the same dis­ tance from the Sun, which is very important. We have seen pictures where there are canals, we believe, and water. If there is water, that means there is oxygen. If oxygen, that means we can breathe. ”

— DAN QUAYLE 140 Chapter 8

8.1 COMPARISON SAMPLES

The redshift distribution and optical and X-ray continuum properties of the USS AGN are compared with two other large, X-ray selected surveys, the Einstein Extended Medium Sensitivity Survey (hereafter EMSS: Gioia et a1. 1990; Stocke et al. 1991; Maccacaro et al. 1992) and the EXOSAT High Galactic Latitude Survey (hereafter HGLS: Giommi et al. 1991). The EMSS sources are selected from the

0 .2 to 3.5 keV band data of the Einstein IPC detector while the EXOSAT CMA detector used for the HGLS was sensitive between 0.05 and 2 .0 keV (unlike the Einstein IPC, the EXOSAT CMA detector had no intrinsic spectral resolution).

8.2 REDSHIFT

The redshift distribution of the secure USS AGN is shown in Figure 8 .1 together with the distributions for the EMSS and the HGLS. There are no secure USS sources with a redshift of less than 0.1, four out of 29 sources have z>0.92 and the remaining secure USS AGN are distributed evenly over the 0.1 to 0.5 range. In contrast, the EMSS AGN distribution shows a strong peak between 0.05 and 0.20 while the HGLS peaks at the lowest redshifts (z< 0 .1 ). Applying the Kolmogorov- Smirnov test to the distributions yields a probability of 4% that the EMSS and secure USS AGN are drawn from the same parent population, and a corresponding probability for the HGLS and secure USS of less than 1 %. The apparent cut-off at z=0.5 for the secure USS AGN may indicate that there is an upper limit to the ‘temperature’ of the soft component which is moved out of the IPC bandpass at higher redshifts. This was previously suggested by Wilkes et al. (1989) who also saw this cut-off in their sample. The lack of secure

USS AGN at z< 0 .1 is curious; soft X-ray excesses have been discovered in very low redshift (z< 0 .1 ) AGN of other Einstein samples (eg. Urry et al. 1989 and

Wilkes et al. 1989) and 6 (out of 24) of the non-secure USS AGN lie below this redshift. Confirmation of this restricted range (ie. z= 0.1 to 0.5) for the steepest soft X-ray spectra in a larger sample of higher data quality, would have important consequences for constraining models of the soft X-ray ‘engine’ in AGN. I have assumed that a common mechanism produces the soft X-ray excess in all the low-redshift (z<0.5) AGN (ie. that which also produces the big blue bump). The 4 secure USS AGN with z>0.5 are E0131-^108, a faint QSO (z=2.36); E0957+561, the double quasar (z=1.41); E l040+123, a superluminal radio source The Multiwavelength Analysis 141

Secure USS

0.0 0.2 0.4 0.6 0.8 1.0 oc 4

3 EMSS EMSS ’•3 2 <§ N •*3 ^ 1 £ 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.2 0.4 0.6 0.8 1.0 4

3 HGLS HGLS

2

1

Ot___ ■ I____,__ □____ ,___ 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.2 0.4 0.6 0.8 1.0 Redshift Redshift

Figure 8.1 : The redshift distribution of the secure USS AGN compared with the distributions for the EMSS and the HGLS. These axe plotted over the full redshift range of the samples and over z=0 to 1 to show the differences at the lowest ranges more clearly. at a redshift of 1.03 and E1346+266, a new USS identification (z=0.92) which also appears in the HGLS. The unusual nature of the high redshift sources (see

7 for more information) suggests that different mechanisms may be responsible for their observed soft X-ray excess. 142 Chapter 8

8.3 LUMINOSITIES

8.3.1 Soft X-ray luminosities.

The Ultra-Soft Survey was originally designed to select hot, isolated neutron stars by searching for a steep slope in the lowest energy channels of the Einstein IPC

PHA distribution, as evidence for the high energy tail of a ~ 1 0 eV blackbody. The discovery of so many AGN in the USS sample was an interesting and unexpected surprise, but assembling a sample of AGN on the basis of their observed soft X-ray spectra has introduced a strong redshift selection effect. The AGN soft and hard component luminosities are compared over an energy range which is defined in the rest-frame. The observed portion of the steep soft component is rapidly shifted through this energy range as the redshift of the AGN increases.

This is illustrated in Figure 8 .2 a where a typical observed USS-type spectrum is represented by a soft blackbody component with a Teff of 7.85 eV and a hard power-law component with an index of 0.7 (see Section 4.3). Figure 8.2a illustrates how a ‘typical’ USS AGN spectrum registered in the observer frame appears in the rest frame at different redshifts, with the spectrum shifting to higher energies as the redshift increases. Plotted as a solid line in Fig­ ure 8 .2 b is a simulated soft component luminosity distribution (integrated over the

0 .2 to 4.5 keV range in the AGN rest frame), calculated for this ‘typical’ observed spectrum at different assumed redshifts. The dashed line illustrates the distri­ bution expected for a perfectly flat spectrum so that there is no dependence on redshift in the luminosity other than the cosmological distance dependence. The actual soft luminosities for the secure USS sources are also plotted for compari­ son with the models. This emphasizes that the steepness of the soft component spectrum exacerbates the usual selection in favour of higher luminosity sources at higher redshifts.

It is thus clear from Figure 8 .2 that by selecting a sample of objects at different redshifts on the basis of their observed spectral shape, a strong redshift dependence is introduced into the soft component luminosity because it is measured over a restricted band. I am unable to quantify the luminosity of the soft component as a whole, without data which reach further into the big blue bump from the UV and the soft X-ray regions. The blackbody parameters which I have derived from modelling the X-ray data are applicable only within the 0.16 to 3.5 keV range; because it is so steep, the model soon becomes unreliable if I extrapolate longwards The Multiwavelength Analysis 143

-10

Key -11 IPC range extrapolation

-12

-13

0.1 0.2 -14 z=0

-15 0.15 0.20 0.25 0.30 0.35 0.40 Rest-frame Energy (keV)

48

46

44

42

40 U 0.00 0.10 0.20 0.30 0.40 0.50 Redshift

Figure 8.2 : (a) The spectra of 6 hypothetical USS AGN, which have identical observed spectra but different redshifts (from 0.0 to 0.5), plotted as they would appear in the rest-frame of the AGN. The redshift of each AGN is indicated on the diagram. The spectra shift towards higher energies as the redshift increases. The dotted line shows the lower limit of the range over which I have calculated the X-ray luminosities (i.e. 0.2 keV). The solid lines cover the observed range of the IPC and the dashed lines indicate the extrapolated portion of the fit. (b) The redshift dependence of the soft X-ray component luminosity. The solid line represents the simulated soft component luminosity distribution for a typical USS-type spectrum observed at redshifts from 0 to 0.5. Measured values of Lsoft for the secure sources are plotted as asterisks. The dashed line represents a simulated distribution for a perfectly flat spectrum so that there is no redshift dependence in the luminosity, other than the cosmological distance dependence. into the EUV. So when looking for the possible dependence of other parameters

on the soft luminosity parameters (eg. L 80f t, Lo.2keV and a oa) this very strong redshift dependence must be borne in mind.

8.3.1.1 The frequency of ultra-soft X-ray excesses in AGN

The USS and the EMSS are drawn from the same parent sample (ie. the IPC database), therefore the fraction of EMSS AGN which are USS sources gives an indication of the ubiquity of ultra-soft X-ray excesses in AGN.f There are 132 EMSS AGN with a z<0.5 (the Wilkes et al. and USS redshift distribution suggests that the soft component may be shifted out of the IPC range at larger redshifts;

see Section 8.2) and an N^ < 3 x 1 0 20 cm - 2 (above this value, most of the soft X-rays below 0.56 keV are absorbed; see Figure 3.4). Out of these 132 EMSS AGN,

13 are USS sources, representing 1 0 % of X-ray selected AGN.

Table 8.1 : Secure USS AGN X-ray and optical continuum correlations

Linear correlation coefRcent array

T ^ h s o f t s o f t Lhard L 0 p t ho.2keV L2 keV L2500A z & 0 8

T ^ 1.0(25) 8 0 f t f^hard 0.7(17) 0.7(17)

L 0 p t 0.6(25) 0.6(25) 0.8(18)

Lo .2keV 1.0(25) 1.0(25) 0.7(17) 0.6(25)

l>2keV 0.7(17) 0.7(17) 1.0(18) 0.9(19) 0.7(17)

L 2 5 0 0 A 0.6(25) 0.6(25) 0.8(18) 0.9(29) 0.6(25) 0.8(19)

Z 0.8(25) 0.8(25) 0.7(18) 0.6(29) 0.8(25) 0.7(19) 0.5(29)

& 0 8 -0.6(25) -0.6(25) -0.3(17) -0.1(25) -0.6(25) -0.3(17) -0.1(25) -0.6(25)

&OX 0.0(17) 0.0(17) -0.5(18) 0.1(19) 0.0(17) -0.4(19) 0.2(19) -0.6(19) 0.3(17)

&SX 0.9(17) 0.9(17) 0.3(17) 0.3(17) 0.9(17) 0.3(17) 0.3(17) 0.7(17) -0.9(17)

Linear correlation coefficient (number of data pairs)

f The EMSS sources have a S/N>4 in the 0.3-3.5keV band (Stocke et al. 1991) cf.>3 for the USS, thus there are many sources in the USS which do not appear in the EMSS. I he iviuitiwaveiength Analysis 14 &

8.3.2 Hard X-ray luminosities

The hard component X-ray luminosity for the secure USS AGN from 0.5 to 4.5 keV, L/tarrf? (see Chapter 4) is plotted in Figure 8.3 as a function of redshift, with the X- ray luminosities of the EMSS and HGLS samples included for comparison. These are plotted over the redshift range of 0.0 to 0.5, where most AGN in all 3 surveys lie (see Figure 8.1 and Section 8.2). I have converted HGLS counts to fluxes using the conversion graph in Giommi et al. (1991), assuming a power-law energy index of 1.5 (which was the best fit inferred for the EXOSAT sample; Giommi et al. 1991) and the N h listed in the HGLS. The EMSS fluxes used have been taken from Stocke et al. (1991) and were converted from count rates assuming an energy index of 1.0. The hard component luminosities of the USS AGN are on average lower than the X-ray luminosities of the EMSS AGN. The HGLS X-ray luminosities are gen­ erally consistent with those of the EMSS.

M in >o o 0 <>: o. o

0.0 0.1 0.2 0.3 0.4 0 .5 Redshift Figure 8.3 : The hard component luminosity from 0.5 to 4.5 keV, plotted as a function of redshift, for the low-redshift (z<0.5) secure USS AGN (asterisks within boxes). The ‘total’ X-ray luminosities of the EMSS (dots) and the HGLS (diamonds) over the same range, are plotted for comparison.

Figure 8.3 shows that about one third of the USS AGN have a value for Lhard that is significantly lower than the band defined by the X-ray luminosity 146 Chapter 8 distribution of the EMSS (note that there are 11 secure USS sources where there are only upper limits on the hard component flux; only eight have z<0.5), These are the ‘hard X-ray quiet’ AGN. Other USS AGN have an L hard that is comparable to the EMSS X-ray luminosity distribution (although generally lower than the average for the EMSS), indicating that for these objects, a strong soft component is superposed on a ‘normal’ underlying hard X-ray power-law.

47

46

45 o CO 44 o

43

42 0.0 0.1 0.2 0.30.4 0.5 Redshift Figure 8.4 : Optical luminosities of the secure USS AGN (asterisks within boxes) plotted as a function of redshift for the low-redshift (z<0.5) AGN. Corresponding luminosities Me plotted for the EMSS (dots) and the HGLS (diamonds). All luminosities are calculated over the 3000 A to 6000 A range from V magnitudes.

Note that the ‘hard’ component luminosities of the EMSS and HGLS sources will include any photons from a soft component, if present, since multiple spectral components were not differentiated in those studies. Thus if any of the EMSS or HGLS sources have significant soft component emission, their hard component luminosities will be overestimated in Figure 8.3. As an illustration of this, if I calculate luminosities for the USS sources on the basis of a single hard power-law fit to the Einstein spectra, I obtain values that are consistent with those in the EMSS (as expected since both are derived from the same count-limited sample). However, I presume that any contribution from the soft component is relatively small in the majority of the EMSS sources since only a small fraction of Einstein The Multiwavelength Analysis 147

AGN appear in the USS sample. By inference, the same is true of the HGLS since the relative numbers of EMSS and USS AGN suggest that the incidence of observable strong soft X-ray emitting components among X-ray emitting AGN is relatively low when they are selected without spectral discrimination.

8.3.3 Optical luminosities The optical luminosities have been calculated in the 3000A to 6000A range for the EMSS and HGLS AGN from the V magnitudes, assuming an optical power-law index of 1.0. I have recalculated the USS optical luminosities in the same way in order to make a comparison. The results are shown in Figure 8.4 for z<0.5 and demonstrate that most of the USS optical luminosities are typical of other X-ray selected AGN (the only notable exception being E1704+608 which is known to be variable).

8.3.3.1 Relationship between X-ray and optical luminosity

3

2

1

0 0 0

1

2L_ 0.0 0.1 0.2 0.3 0.4 0.5 Redshift Figure 8.5 : The ratio of broadband optical to hard X-ray luminosity (shown separately in Figures 8a and 8b) for the secure low-redshift (z<0.5) USS AGN (asterisks within boxes), plotted as a function of redshift. The ratio of broad­ band optical to ‘total’ X-ray luminosity is plotted for the EMSS (dots) and HGLS (diamonds) for comparison.

Strong correlations between X-ray and optical luminosity have been reported by previous authors in both X-ray selected (eg. Kriss and Canizares 1982) and 148 Chapter 8 optically selected samples (eg. Zamorani et al. 1981, Kriss and Canizares, 1985). Figure 8.5 shows that this correlation also exists in the EMSS and HGLS AGN (although the latter is not as tight).

2.5

2.0

oo -o ;9o

o o h .o .. rssir

1.0

0.5 0.0 0.1 0.2 0.3 0.4 0.5 Redshift Figure 8.6 : The ratio of monochromatic optical to hard X-ray luminosities, Otoxi plotted as a function of redshift for the low redshift (z<0.5) secure USS AGN (asterisks within boxes: see Section 5.3.2 for the definition of olox).

Also plotted is the distribution of o l o x for the EMSS (dots) and the HGLS (diamonds).

I investigate the relationship between X-ray and optical luminosity directly using their ratio parameterized by a ox (see Section 5.3.1). Using the method de­ tailed in Avni et al. (1980) which takes into account the upper limits on L 2 keV-, I calculate an effective a ox of 1.36 ± 0.05 for the USS secure sources (the aver­ age detected a oz is 1.37). For the EMSS sample, I calculate an average a ox of 1.33 ± 0.01 (this excludes BL Lac objects, ‘normal’ galaxies and AGN with an uncertain redshift) and an average of 1.35 ± 0.05 for the HGLS. Due to the red­ shift effect in the soft component luminosities, values for a os and o l 8 X are strongly dependent on redshift (see Section 7.1.1) and it is not appropriate to calculate the corresponding averages. Values of otox for the secure USS, EMSS and HGLS sources for z<0.5 are plotted against redshift in Figure 8.6. For the EMSS, the a oxs are those listed in Stocke et al. (1991). I have calculated values of a ox for the HGLS AGN, where The Multiwavelength Analysis 149 the flux at 2500 A was derived from the V magnitudes using the equations in Schmidt (1968) and the flux at 2 keV was derived using the same method as for the broadband X-ray flux (see Section 8.3.2). Note that values of a ox for the USS have been calculated after the subtraction of the soft component: EMSS and HGLS o0xS include any soft component flux. About two thirds of the USS sources have an olox which lies within the EMSS range; for these objects, the soft component may be superposed on a ‘normal’ un­ derlying hard X-ray to optical continuum. The remaining a oxs are high, suggesting that the hard component is depressed relative to the optical. The ratios of the broad band luminosities (Lopt/Lhard) for the secure USS, EMSS and HGLS, are plotted as a function of redshift in Figure 8.5, and bear out the results of the a ox distributions.

8.3.4 Infra-red luminosities Rest-frame IR luminosities have been calculated at 1 //m, (Li^m) and 1.65/zm

(Li ,6 5/zm) for the twelve USS AGN for which IR images have been obtained (see Chapter 6). These include eight secure and four non-secure sources. Linear regres­ sion correlations between IR parameters and other properties of the USS AGN, ie. X-ray and optical continuum parameters and optical line continuum parame­ ters, are given in Table 8.2. IR luminosities are compared with those of two other samples, Worrall (1987) and Kriss (1988) in Figure 8.7 and Figure 8.8. The Worrall sample is made up of 69 radio-quiet QSOs drawn from the optically selected sample of Avni and Tananbaum (1986), while the 88 Kriss AGN are X-ray observed, radio-quiet QSOs from Kriss and Canizares (1985) and Tananbaum et al. (1986), with li2keV < 6 x 1024 ergs s 1 Hz 1 and a starlight contribution <50% at lp m . Almost all of the Worrall AGN (62 out of 69) are also in the Kriss sample, but both are shown because Kriss has measured the IR flux at lp m , whereas Worrall made her measurements at 1.65pm. Also, the IR luminosities given by Worrall include any flux from the host galaxy so they are compared with the total USS Li.65/am whereas Kriss has quoted fractional contributions to the total IR flux from the host galaxy; I have subtracted these to obtain nuclear IR luminosities and these are compared with the nuclear IR luminosities of the USS. Only one USS AGN is also found in these samples, E1352+183, which appears in both. The IR flux of this object measured from UKIRT images is approximately 30% lower than that 32 32

31 31 £ 30 b O 00 30 5 29

29 28 0.00 0.10 0.20 0.30 0.40 0.50 0.00 0.10 0.20 0.30 0.40 0.50 Redshift Redshift 34 34 33 33 32 32

b o 31 b o 31 5 3 30 30 29 29 28 29 30 31 32 33 34 24 25 26 27 28 29 Log (L2500 a) Log (L ^ )

Figure 8 .7 : The integrated IR properties of the secure and non-secure USS AGN for which IR images were taken (plotted as asterisks within boxes) compared with the sample of Worrall (1987; triangles), (a) Integrated IR luminosities at 1.65/im, plotted as a function of redshift. (b) Optical luminosities at 2500 A, L2 5 0 0 A 1 plotted as a function of redshift. (c) Integrated plotted against L2500A- (d) Integrated Li.65nm plotted as a function of X-ray luminosity at 2 keV, L2keV- used by Kriss and Worrall, who both used the Neugebauer et al. (1987) IR data taken in 1980. Figure 8.8a shows that the 1/mi luminosities of the USS AGN are relatively low for a given redshift, but Figure 8.8b shows that L2500A is also low, thus the USS AGN are faint overall at a given redshift relative to the Kriss sample (the

L2keV for the USS, which is not shown, is also low). Similar results are seen for

Li.6 5 /im when comparing the USS with the Worrall data (Figure 8.7a and b).

8.3.4.1 The dependence of the IR on the X-ray spectrum

The IR luminosity dependences on L 2keV for the USS AGN are compared with the Kriss and Worrall samples in Figure 8.7d and Figure 8.8d. There are only six detections of the hard X-ray component for the IR USS AGN sample, therefore a discussion of their IR/X-ray relationship is limited! However, the plots show that Ihe Multiwavelength Analysis 161

32

31

* o* po 29 p °

28 0.00 0.10 0.20 0.30 0.40 0.50 0.00 0.10 0.20 0.30 0.40 0.50 Redshift Redshift

oo

33 24 25 26 27 28 29 Log (L-2500A) Lofi (Lacev)

Figure 8.8 : The nuclear IR properties of the secure and non-secure USS AGN for which IR images were taken (plotted as asterisks within boxes) compared with the sample of Kriss (1988; diamonds), (a) Nuclear IR luminosities at 1/xm, Liplotted as a function of redshift. (b) Optical luminosities at 2500A, L2500A> plotted as a function of redshift. (c) Nuclear Li^m plotted against L250(M- (dj Nuclear plotted as a function of X-ray luminosity at 2 keV, \j2keV-

the USS AGN generally lie within the range of the comparison samples. Both Worrall and Kriss found a non-linear dependence of the hard X-ray component luminosity on the IR, similar to the dependence between hard X-ray and optical luminosities, ie. they found that the X-ray-to-IR luminosity ratio de­ creases as the IR luminosity increases. The X-ray-to-IR ratio is plotted against IR luminosity in Figure 8.9a for the nuclear 1/xm luminosity, and Figure 8.9b for the integrated IR luminosities. Low-redshift (z<1.0) objects are shown separately from the high redshift (z>1.0) AGN in these plots (all of the IR USS AGN sample have z<1.0). Again, the USS objects are consistent with both comparison samples.

For L ir = L j, Kriss calculated 7=0.73, having made corrections for host galaxy contamination and taking into account X-ray upper limits. Worrall found

7 <1 for the integrated IR luminosities at the 95% confidence level and pointed 152 Chapter 8

3 3

I i Aa ■4 ■4 LAa * a A A

too -5 ■5

6 6 28 29 30 31 32 33 34 29 30 31 32 33 34 Log (Lllun) Log (L 1j6Sjub)

Figure 8.9 : The X-ray-to-IR ratio of the USS AGN plotted as a function of IR luminosity and compared with the Kriss and Worrall samples. All USS AGN lie below z=1.0 and are shown as asterisks within boxes. Upper limits on l/2keV 3X6 indicated by arrows, (a) Nuclear ratios at Iflm for the USS AGN compared with the Kriss sample; low redshift (z<1.0) Kriss AGN are plotted as diamonds and high-redshift Kriss AGN (z>1.0) are plotted as crosses, (b) Total (nuclear-j-galaxy) IR ratios at 1.65pm for the USS AGN compared with the Worrall sample; low redshift (z<1.0) Worrall AGN are plotted as triangles and high-redshift Worrall AGN (z>1.0) are plotted as ‘X’s.

out that subtraction of the galactic contribution would reduce 7 further (assuming that the host galaxy contamination decreases with increasing redshift). She sug­ gested that the synchrotron self-Compton model of Zdziarski (1986) was the most promising mechanism to explain the X-ray/IR correlation. Worrall and Kriss both found that any redshift dependence is weak or non-existent. Fits to the USS AGN data (ie. neglecting the upper limits) indicate a non­ linear dependence of L^m and Li. 6 5 ^m on L/2keV> with slopes <1, consistent with the results of Worrall and Kriss.

Malkan (1984) found a linear relationship between L ir and L x in his sample of radio-loud and radio-quiet Seyfert Is, although Kriss demonstrates that this may be due to the mixing of radio-loud and radio-quiet objects. A linear relationship was also found by Fabbiano et al. (1986) for their hard X-ray selected Seyfert Is but their sample spans a relatively small range in X-ray luminosity.

8.3A.2 The dependence of the IR on optical luminosity

The tight, nearly linear dependence of LiMm on L 2 5 0 0 A reported by Kriss can be seen in Figure 8 .8 a. While a strong correlation is also seen for the USS AGN (a correlation probability of >99.5%; see Table 8 .2 ), the slope is flatter indicating a non-linear dependence of LiMm on L^sooa* A linear least-squares fit to the Kriss m e iviuiiiwaveiengiu analysis j.0,3 data (in log space) gives a value for 8a of 0.94±0.02 for Li^moc L*Jt, whereas for the USS, 8a = 0.5 ± 0.2. Similar trends are seen at Li.es^m in the Worrall data (Figure 8.7). Assuming Li .65/un « L ^ , gives

1.5 1.5

"I 1.0 1.0

-0.5 -0.5 28 29 30 31 32 33 hO£ 0^25OOa) hO£ 0^25OOa).

Figure 8.10 : The IR-to-optical luminosity ratios plotted against optical luminosity for the USS AGN (asterisks within boxes) and compared with the Kriss and Worrall samples, (a) Nuclear ratio at 1/im for the USS AGN compared with the Kriss sample (diamonds), (b) Integrated ratio at 1.65//m for the USS AGN compared with the Worrall sample (triangles).

In his sample which extends to higher luminosities, Kriss found that the IR- to-UV ratio is higher for objects with LiMm< 1030 ergs s-1 Hz-1 than those with higher IR luminosities, but that the ratio remains roughly constant above this threshold. A plot of the IR-to-optical ratio against optical luminosity, illustrated in Figure 8.10, shows this effect very well. For optical luminosities, L 2500A? be­ low ~ io30-5 ergs s-1 Hz-1, the IR-to-optical ratio decreases steadily as L 2500A increases. As L2500A reaches the IO30 5 ergs s-1 Hz-1 threshold, the IR-to-optical ratio flattens and is indeed, roughly constant. This is also seen in Figure 8.10b for

Li.65 /xm in the Worrall data. The USS AGN IR-to-optical ratios all lie on the decreasing Hail’ before the log L 2500A = 30.5 threshold is reached. Therefore, while the IR-to-optical ratios of the USS AGN are consistent with the comparison samples, they are concentrated at low IR-to-optical ratios. This would explain the relatively low values of 8a and 8b for the USS AGN. 154 Chapter 8

Table 8.2 : IR correlations with optical and X-ray parameters

X-ray and optical continuum parameters

Correlation coefficent Probability (%)

Ll fim 1'1.65/tm OLIR 1*1^1 m ^<1.65 pm <*IR

Ll.65 /itn 1.0(12) 100

o l i r 0.5(12) 0.6(12) 87 96 Lo.2keV 0.8(11) 0.8(11) 0.5(11) 100 99 87 L2keV 0.9( 6) 0.9( 6) 0.5( 6) 99 98 63 ^250OA 0.9(12) 0.9(12) 0.4(12) 100 100 80 OCoa -0.5(11) -0.4(11) -0.2(11) 84 77 39

OL q x 0.7( 6) 0.6( 6) 0.0( 6) 84 75 1 oc8X 0.6( 6) 0.5( 6) 0.2( 6) 79 73 24 z 0.6(12) 0.6(12) -0.2(12) 95 95 48 N H 0.4(12) 0.3(12) 0.3(12) 79 70 67 OLopt -0.9(10) -0.9(10) -0.5(10) 100 100 87

Optical line luminosity

Correlation coefficent Probability (%) Ll/zm 1*1 .65 pm OLIR Ll pm 1*1 .65 pm OLIR

H/? 0.7(10) 0.8(10) 0.6(10) 99 99 91 [OIII] 0.6( 9) 0.6( 9) 0.4( 9) 94 93 73 Hel 0.9( 7) 0.9( 7) 0.5( 7) 99 98 77 0.9( 9) 0.9( 9) 0.6( 9) 100 100 92

Optical line FWHM

Correlation coefficent Probability (%) Ll pm hl.Q5fim OLIR Ll pm hl.Q5fim OLIR

H/J 0.0(10) -0.1(10) 0.0(10) 5 16 7 [OIII] 0.2( 9) 0.4( 9) 0.6( 9) 32 67 88 Hel 0.2( 7) 0.0( 7) -0.4( 7) 37 3 65 Ho -0.2( 9) -0.3( 9) -0.3( 9) 35 54 52

Optical line EW

Correlation coefficent Probability (%) Ll fim I<1.65 fim OLIR 1*1 fim 1^1.65/1771 OLIR

H/? 0.4(10) 0.3(10) 0.0(10) 74 52 1 [OIII] 0.4( 9) 0.2( 9) -0.1( 9) 69 47 15 Hel 0.8( 7) 0.6( 7) 0.1( 7) 96 84 18 Ho 0.7( 9) 0.5( 9) 0.1( 9) 96 86 25

Linear correlation coefficient (number of data pairs) Probability of correlation m e iviuiiiwaveiengm Aneuysis 1 0 0

8.4 OPTICAL LINE AND CONTINUUM PROPERTIES 8.4.1 Line widths Examination of Table 5.4 reveals a high proportion of narrow-line objects among the USS sample. A narrow line Seyfert 1 is defined by Osterbrock and Pogge (1985) as a Seyfert 1 or Seyfert 1.5 galaxy with a broad line FWHM less than 2000 km s-1 (see Section 1.2.1). Osterbrock (1987) reports that approximately 15% of Seyfert 1 and 1.5 galaxies belong to this group. Stephens’ (1989) sample of X-ray selected AGN contains 42 potential Seyfert 1 galaxies of which 10 (or 24%) are narrow line Seyfert Is. In contrast, when we apply the same criteria to the USS sample, 9 out of 17 Seyfert 1 galaxies, or ~50%, have permitted lines with FWHM of less than 2000 km s_1.

8.4.1.1 H/3 FWHM distribution. In Figure 8.11, I have compared the number distribution of the deconvolved B./3 FWHM for the USS sample with those from the samples of Stephens (1989) and Mittaz (1991; labelled ‘HGLS’ on the diagram). Only Seyfert Is and quasars/QSOs axe included from each of the three samples.

The Stephens sample

The Stephens AGN axe X-ray selected from Einstein IPC data without regard to the X-ray spectral slope. The sample includes 10 USS objects (5 secure, 5 non- secure) as well as those with harder X-ray spectra and is restricted to objects with redshifts of 0.56 or less. The data on the 10 USS objects have been included with our USS measurements, and are excluded from the distribution for the Stephens sample. The FWHM given in the Stephens paper have been deconvolved from the instrumental width.

The HGLS sample

The HGLS sample is X-ray selected and is made up of active galaxies identified in the EXOSAT High Galactic Latitude Survey (HGLS; Giommi et al. 1991). The response of the EXOSAT low energy telescope was biased to lower energies (0.05- 2.0 keV) than Einstein but the CMA detector used for the HGLS had no intrinsic spectral resolution. Because only a small fraction of the total number of AGN observed by Einstein satisfy the USS criteria, it is likely that an undifferentiated sample such as the HGLS will contain predominantly ‘hard’ objects (as does the 156 Chapter 8

10

Z 8 USS o (inc. Stephens USS) < 6 u-io 4 6 a Z 2 0 10

Z Stephens o (not inc. Stephens USS] < <4-1o <3 X> S z

(not inc. HGLS USS)

0 2000 4000 6000 8000 10000 FWHM Hp km s'

Figure 8.11 : The distribution of the H/? FWHM for the USS AGN (secure and non-secure sources) compared with the Stephens (1989) sample and AGN from the HGLS. All FWHM have been deconvolved from the instrumental profile. The median of each distribution is indicated by the dashed line.

EMSS), despite the softer energy response of the EXOSAT CM A. Indeed, the sample of USS X-ray spectra for which I have optical spectra (secure and non- secure) was best fit with an ‘average’ single power-law of index 6, much steeper than the index inferred for the HGLS AGN (1.5; Giommi et al. 1991). This The Multiwavelength Analysis 157 implies that the HGLS sample is hard in relation to the USS. f Two EXOSAT HGLS AGN are also USS sources, one secure (E1346+266) and one non-secure (E1059-f730). I have deconvolved the HGLS H/? FWHM from the instrumental width (14.5 A; Mittaz 1991) to compare them with the USS sample. There is no redshift restriction for the HGLS (the redshift distribution is shown in Figure 8.1).

The USS AGN - narrow-hne objects?

Figure 8.11 demonstrates that the TL/3 FWHM distribution for the USS AGN is biased to lower widths than the other X-ray selected samples. The median of each sample is indicated by the dashed line and is lowest for the USS AGN (2180 km s-1; cf. 3190 km s-1 for the HGLS and 3050 km s-1 for Stephens). The comparison between the USS and the Mittaz sample is particularly strik­ ing because many of the optical identifications that went into these surveys were made by the same team during the same observing runs and using the same equip­ ment. Therefore, there should be no systematic optical selection bias between these two samples. Also, both sets of optical spectra were reduced using the same soft­ ware so the method of measuring the FWHM for these samples is consistent. An application of the Kolmogorov-Smirnov (K-S) test indicates that there is an 8% probability that the HGLS and the USS are drawn from the same parent pop­ ulation. There is a corresponding 4% probability for the USS and the Stephens samples, f These results suggest that the presence of narrow permitted lines is a preferred characteristic of AGN which show the USS phenomenon.

Narrow lines and radio properties of AGN

Narrow permitted lines are also a characteristic of core-dominated radio sources; a strong anti-correlation has been reported between the broad component FWHM

t The sample of secure USS AGN was best fit with an average single power- law of index 1.9. The difference between these indices is due to the fact that the secure USS are dominated by objects with significant hard components, whereas the sample of AGN with optical spectra has a greater fraction of objects with no detectable hard X-ray flux. $ The K-S test gives a 94% probability that the HGLS and Stephens samples are drawn from the same parent population, implying that the Einstein IPC and the EXOSAT CMA are sampling the same population of X-ray emitting AGN. of the H/? line and the ratio of radio core flux density to extended radio lobe flux density, R (Wills and Browne 1986). These authors also found that whereas narrow lines are seen in both core- and lobe-dominated quasars, broad lines occur mostly in lobe-dominated quasars.

Narrow lines and warm IRAS Seyfert Is

For a sample of warm, IRAS Seyfert Is with X-ray measurements, Ward (1988) found a high incidence of NL Seyfert Is (3 out of 5 Seyfert Is; there were also two Seyfert 1.9s and one HII region galaxy). ‘Warm’ IRAS objects have relatively flat spectral indices between 25-60 pm (de Grijp et al. 1985). Although the sample is small, these results are supported by evidence that 5 out of 5 NL Seyfert Is studied by Osterbrock and Pogge (1985), have warm IRAS colours.

8.4.1.2 Relationship between Balmer line FWHM and aox

Given the high incidence of narrow line objects in the USS sample, I have looked for a relationship between permitted line width and optical and X-ray continuum properties of this sample. There is evidence for an anti-correlation between the Ha FWHM and a ox ie. sources with a hard X-ray luminosity which is strong relative to the optical luminosity, have broader Ha lines (see Table 8.3). The H j3 data, which include additional data points from Stephens, generally support this correlation and are shown, together with the Ha data, in Figure 8.12. This Figure also shows upper limits for those objects where the hard X-ray emission is not detected. In general these do not contradict the suggested correlation. I found no evidence of a correlation of this kind in the Stephens sample or in the Stephens sub-sample of narrow-line objects. However, it may be that an a ox~ FWHM correlation may only apply in the presence of a strong soft X-ray excess. An investigation into the relationship between the broad line FWHM and a ox for a sample of objects which have a range of soft X-ray properties is needed to clarify this (or more specifically, the proper subtraction of any soft X-ray component is required to allow a comparison with the USS AGN value for a ox). Blumenthal, Keel and Miller (1982) reported evidence of a positive correlation between the half-width at zero intensity (HWZI) of H/3 and a'ox (the ratio of the luminosities at 5000 A and 2 keV), ie. in the opposite sense to the relation seen in the USS, but they point out that their correlation is mostly due to a few extreme points (and they have made no subtraction of any soft X-ray component). The Multiwavelength Analysis 159

8000 8000 Key USS AGN detection [jj] USS AGN lower limit *->■ 6000 Stephens detection [o] 6000 Stephens lower limit a a 4000 4000

2000 2000

1.0 1.2 1.4 2.0 1.0 1.2 1.4 1.6 1.8 2.0 a„ a

Figure 8.12 : The FWHM of Ha and H/? plotted as a function of olox for the USS AGN (secure and non-secure sources). FWHM measured by us are shown as asterisks and those taken from Stephens are plotted as diamonds. A box around an object indicates a measurement of the ot ox; lower limits to OLox axe indicated by the arrows. All FWHM have been deconvolved from the instrumental profile.

8.4.1.3 Line widths and continuum luminosities There are no significant correlations of the broad fine FWHM with hard X-ray, soft X-ray, optical or IR luminosity for the USS AGN. A correlation was reported between the H(3 FWHM and (total; ie. including any soft component photons) X- ray luminosity for the Stephens objects (I calculate an associated linear correlation probability of 99.9% for the Stephens sample). Evidence for a correlation between Balmer line HWZI (which is taken as a representation of the innermost radius of that line-emitting region) and the X- ray luminosities of Seyfert galaxies have been reported (eg. Kriss, Canizares and Ricker 1980) but this was not seen by Blumenthal, Keel and Miller (1982) in their sample of 23 quasars. A correlation has been reported between the H/2 HWZI and bolometric luminosity, L boh of Seyferts, but again no corresponding correlation was found for quasars (Padovani and Rafanelli 1988; Padovani 1989). There is evidence for a correlation of the HWZI of CIVA1549 with L boi for quasars, but there is no corresponding correlation for the FWHM of CIVA1549 (Padovani 1989). 160 Chapter 8

K

o' o' £ 3 o'o'o' rH rH w_ pH rH rH rH o' 2 «T 2 5 "« oT'^'T r - i f f i q E C J5 o o o J• o ) ) I—r» O <0 CO O CO CO rH CO , o O O O 10 co —< 10 _h r-i _h _< rH - h - h Qj t-h i—i t-h i—< ( ( 0 8 n co io re . . 2 j a

o 5 S o S ^ o [OIII] [OIII] 0 0.7(11) 0.7(16) 0 0.9(16) 0 0 0 0

-0.5(16) 1 1 T ) _ 0 0 0 0 0 10

EC rooo0 jc Hios a oq o 05 « n o v s . a a 1.0(15) 0.9(10) 0.9(15) 1 0.1(16) 0.5(10) 0.7(15) 0.8(16) 0.9(16) J o cji o 0 o

) ) ) ) 6 6 6 6 „ oon'oco o ( ( ( 4) ( ( 4) ■JJj , _ ( 4) 'JJj __, 'ho 3 ( ( ( ( 5 ^ o '^ r'' e- 5 co h rr 5; 1-10 0 •-> ES o rH co cm 10 . Mgll Mgll 0.1 0 0.6( 0.6( 7) 0.1 0-4( 5) - 0.5( - 7) 0.4( —’ 0.7( 5) — — —' — — — *J J o d o o o “ “0.1 ( ( “0.1 5) -

OcT t^CD^O^iir 00 CD (O O to „ « o o « 0 0 0 , rH rH rH H r H H rH rH rH rH rH rH §■ cToo'o'o' §• a ' CO ID cT 8- 10' CO CO CO 00 O) DO o 10 Tf N Tf 0 e 0 ? ? Tr ,r ? ? ? ? 6 d ? ? ? ? ( j ^ d d o ^

r H h * S ^ S H N N 10 ’T M1 10 f W f H H *> ( 4) •» n 0 N 00 N S xe OO ^ ^ S S xe ^ t— CD CO o' x O) 00 0.7( 0.7( 7) 0.8( 0.8( 7) ? 0.7( 7) O d d o O O

rH S S ^ N H h h N m 'T in Tf u) v M o H ______( ( 4) O 0 CO CO H CM aEe HrH CH CC S K e q rH^ qO ' ^ q O x 00 COCO 0.2( 0.2( 7) 0000 0.3( 7) ( 0.1 7) oOd o o O o0 o 0 o 0 o 00 I 1 I I 10 t «) 0 W) w ^ m h m h o) o) h a h o) 0 rH r H r H H On 1—1 1—1 1—1 rH H r H rH 8 N N N Cl ^ d o' m cm c m m 00 to m in i-1 m d d cp 0 0 d o o ' d d -0.4(10) -0.7(15) -0.6(14) -0.6(15) 1 1 1 T ?? ? d6

) oo'tTtT'o'co' oo't^'e^'lH'co' co'o'o'coo'coo' T“H T—I 1 1 10 1—1 — rH rH rH rH rH rH rH rH rH rH rH ( N 6 00 CM »q 1-1 CO 'a' CO CO CD K o 'er co co o'p'eT . 0.6(17) 0.7(17) 0.8(16) 0 o d d d d ______1 1 o d d d d ddddddo1 1 ) 3 oo'co't-'e-'o'co' 3 oo'tl'eT'o'co' oo't-'e^'o'co' 3 oo't^'t^'a'co'oo'eC't3'avco' 5 co'o'o'co'o'eo'o'

g H rH rH rH rH10 g0 rHH rH1—1 rH i—l rH rH gO rHrH i-H rH rH rH H rH ^ rH rH rH^rH^ rH rH rH

( 0 O IO ' ” ”0 ' 10 H CO N ”H q wm i n ' n < co s jo q o n v h n cq . CH CH 1.0(16) 1 0.9(17) j ^ddoHri0.8(17) j d o0 d 0 o '0 c 0 ^ )<^5 h,j3 c0 j 0 d 0 o 0 d 0o jdi-Jddodd

^ e - s ^ s ^ r(r H SN N O1 if C- N ^ H' i n 'S N TJI h ^ W W u) if U) 'C «> •«< ' '^h ’'co' o' 0 0 0 0 0 5 ^ 0 0 00 00 00 CH co o ^5CH 0 0 0 0 0 ^ co e- co j c^> 0 0 o0 j d1 o Oo j 0 0 o0 jo o cp ) i n i p w 0 w 10 i n h i n **■ **■ ^‘C^STS'S? ^ ^ “s' o'»o' ^V m if rn'io' *5“ ^-h' O) a h A rH- 05 C

^ rH rH rH10 rH J| rH rH rH^ rH rH^ rH ^ rH rH rH rH ^ t*H^ H ^

( J i CH c s 8 (N H S H IO CO CO CO If CD

. w aaNWioHrt o«

0.8(14) 0.8(15) 0.9(15) O 0 ddddddo0 dddddo0 0 0 d 0 odddddO d d d d o ddddddo *J J 1 ji- 3 i1 1 i T i1 j i 1 j 11 -g

oo'eZ'tT'o'co' 00 N N rH CO CO O O CO O CO o W o, _ i —1 rH r H rH a rH rH rH rH _ _ g o o CO C5 00 C> O) O§• W N S O •C o0 CO CD CO CH C 4 1 i q 0 COCD j 0.9(17) 0.8(17) 0.9(16) 0 J dddddo0.9(10) J oO o O ’d O d O dCji ^ odddd0 0 0 0 0 ^ Jo odddd *g T 1 T 1 ■ 1 1 g 3 h N N r i S N ^ h •tj a"P'c3'a' in v ’f m ic w ^ 0 0 u Ji 00 00CO 00 0 0 00CO -c wX r H e cf w o o q' c 0000 0000 o> p -c co e- co 0 J cji 00d d oO J ooo0 0 0 J o0 o 0 o’ o’ Jo ocj> l

2 ' ^ i ' u T ' ^ ' i o ' o ' w mm if if m m o0 m m w mm ^i f w i n h inm *. h a a h a h a v H ^ rH rH rH rH rH rH1—1 rH r -* rH1“ I rH1—1 s rH^rH rH^ rH rH^ rH rH rH rH rH ^ o 0 0 „ .. . ,CH ...... ^ h ' ^ h ' c o ' o ' m ?' w .. •». r H 0 0.... i n N N c c « o 00'0'cq o' rH CO

J ^>000000.9(14) 0.9(15) 0.9(10) 0.9(15) J ocpodcji o <^> 0 0 cji J 00000o d d d d J 0000000 u 1 i n i f m 0 m kO kO |H 10 v rH O) O) H O) H O) rH rH rH rH rH rH rH rH s-^ rH rH rH s a c 0 HHCDOlf H 0 0 tO N S O ^ rH-—..... n rH *j - .... ri 2 0 oo'o ' cd o ' o cq' 0.9(15) ^ J cj> o o 0.9(14)o 0.9(15)o 00.9(10) j 0O cj> Cj 0 0 0 0 <^> cp J odddd0 0 0 0 0 ]j J 000^00 1 <3 S «w « S n o - s cixct^ 0 < a e a «aJ=r^ « c S «CL«o.ircr -S 06 .s oboi a a k ft »sass . wscaaa | | ^ j; -2 b. c*3 d, 3 j;uu fO V o « E5 •? J 03 .C .C .g ^3 J H j j j k< The Multiwavelength Analysis 161

8.4.2 Other Line Parameters

8.4.2.1 Line Luminosities

Strong correlations have been reported previously between the H/3 luminosity and the (total) X-ray continuum luminosity of AGN, and also between the H/? lu­ minosity and the optical continuum luminosity (eg. Kriss and Canizares 1982; Blumenthal et a!. 1982). Stephens’ (1989) sample shows similar results for the [OIII] A5007 luminosity as well as H./3 although the H/9 luminosity is more tightly correlated than [OIIIJA5007 with both optical and X-ray luminosities.

Broad line luminosities

In the case of the USS AGN, the Balmer line luminosities are strongly correlated with optical luminosity (see Table 8.3) and IR luminosity (Table 8.2). There is also evidence for correlations of the Balmer line luminosities with both L 0.2keV and L 2keV (although there are fewer data pairs for the L 2 keV correlation because many sources for which I have optical spectra have only upper limits to their hard X-ray luminosity; out of 7 such data pairs, only 3 are secure sources). A similar dependence is found for the luminosity in the HeIA5876 line with soft X- ray, optical and IR continuum luminosities (correlations with L hard have not been made because there are only 4 data pairs).

An examination of the ratios of the broad line luminosities to Lo.2keV. reveals a strong dependence in these ratios on redshift and reflects the strong redshift dependence in Lo. 2fceV (see Section 8.3.1). There is no evidence for a similar redshift dependence for ratios with the optical and hard X-ray luminosities. A more representative sample of AGN is required to confirm the apparent correlation between broad line luminosity and Lo .2keV* There are no correlations between the MgIIA2798 luminosity and the optical luminosities. Although there is no correlation with Lo. 2fcevs there is a correlation with Laoft. Correlations with the hard component were not possible because there was only 1 data pair (due to the wavelength coverage of the optical spectra and the number of upper limits on the hard X-ray component).

Forbidden lines

The luminosity in the narrow, forbidden [OIII]A5007 line is correlated with L 2keV and Lo. 2fce V"• The [OIIIJA5007 luminosity is also correlated with Lopt but this is not 162 Chapter 8 as strong as for the Balmer lines (confirming the Stephens results) and HeIA5876. There are also weak correlations with the IR luminosities.

A comparison with predicted broad line ratios

USS AGN fine ratios have been compared with the models of Krolik and Kallman (1988) who calculated UV and optical line ratios for three different models; a ‘bare power law’ (with an index of 1.2 for energies <2 keV and 0.7 >2 keV), a ‘10 eV bump’ spectrum and an ‘80 eV bump’ spectrum. The two ‘bump’ spectra are the sum of the bare power-law and a simple accretion disk (eg. Shakura and Sunyaev 1973), which is geometrically thin and optically thick. Krolik and Kallman com­ pared these (27) models with the list of observed fine ratios compiled by Kwan and Krolik (1981) and found that the 10 eV bump model was the best fit. A HeIA5876/H/? ratio typical of the USS AGN (average ratio for the USS is 0.15) was reproduced by the 10 eV Krolik and Kallman model, which predicted ratios in the range ~0.14-0.20. The other Krolik and Kallman models predict ratios which are generally too low. The 10 eV model also agrees well with the HeIA5876/H/9 ratios for the samples of Osterbrock (1977; ratio=0.18) and Stephens (ratio=0.18). The USS AGN ratios of Ha/H/? and MgIIA2798/H/? are also similar to those predicted by the 10 eV model, as are the UV line ratios of E0132-411, which include L y a/H a ( 10).

[OIII]\5007/Balmer line ratios.

Grindlay et al. (1980) found a correlation between the [OIIIJ/H/? luminosity ratio and X-ray luminosity for radio-quiet, X-ray selected objects. Steiner (1981) con­ firmed this correlation in his low redshift (z<0.7) AGN but only for objects with Fell emission lines. Kriss and Canizares (1982) found no significant correlation in their data but point out that their sample is much smaller. There is evidence of a correlation of this ratio with R, the ratio of core to extended radio luminosity, which is an indicator of the angle of the radio axis to the fine of sight (Jackson and Browne 1991). While I could find no correlation between the [OIIIJ/H/? ratio and the separate X-ray luminosities for the USS sample, there is evidence for a corre­ lation with the FWHM of H/? and this is shown in Figure 8.13 with the Stephens sample for comparison (there is also a corresponding correlation with a ox - see Section 8.4.1.2). The Multiwavelength Analysis 163

1.5

S 0.5 §HH HH ^ ^ O 0.0 @ o ttf) O*. 5 -0-5 o o

- 1.0

-1.5 0 2000 4000 6000 8000 FWHM Hp

Figure 8.13 : The [OIII]A5007/H/? luminosity ratio plotted as a func­ tion of the H/9 FWHM for the secure and non-secure USS AGN. Ratios from measurements in this thesis are plotted as asterisks (secure USS in boxes) while measurements taken from Stephens (1989) are plotted as diamonds.

8A.2.2 Equivalent widths The EWs of the broad lines (Ha, H/3 and HeIA5876) in the USS AGN axe gener­ ally low compared to the Stephens sample and, since the optical luminosities are typical of X-ray selected AGN, this implies that the USS broad line luminosities are relatively low (see Figure 8.14). The [OIII] A5007 EWs for the USS AGN are typical of the Stephens sample.

Correlations with other parameters

There are correlations between the Balmer line equivalent widths of the USS AGN and the hard X-ray luminosity. No similar correlation was found in the Stephens narrow line AGN, but the hard X-ray luminosities in Stephens’ sample have not been corrected for any contaminating soft X-ray component. There are also weak correlations of the Balmer line EWs and the optical luminosity in the USS data. There are weak correlations between the Balmer line EWs and HeIA5876 EWs with Lo.2 keV, but note that the EWs of Ha and Hel have a similar degree of correlation with redshift. There are also correlations between the broad line EWs and a sx. The H a and HeIA5876 EWs are correlated with Li^m (both 96%) but only 164 Chapter 8

150 “ T "■ ■ "

: o [Oin]A5007 1*1 100 3 I o o IE 50 _ ° o o _ £ 0 0 O ° [I] pj ° O @ - V -,0 o °o o a «o 8 n r* fo JS

120 100

0 0.1 0.2 0.3 0.4 0.5 Redshift

Figure 8.14 : The equivalent widths of USS AGN emission lines (asterisks in boxes) plotted as a function of redshift and compared to the Stephens (1989) sample (diamonds).

weakly with Li.65Mm (86% and 84% respectively). Correlation probabilities be­ tween IR luminosity and H/? and [OIIIJA5007 are lower still, and again relatively weaker with Li.65^m(see Table 8.2). There are no correlations with a j r . Jackson and Browne (1991) have reported a strong anti-correlation between the [OIIIJA5007 equivalent width and R, the ratio of core to extended radio lu­ minosity, for their sample of radio-loud quasars and suggest that this is due to The Multiwavelength Analysis 165

anisotropic optical emission, for instance angular dependent emission in the case of an accretion disk model. There is no correlation of the [OIIIJA5007 EW with

Lopt or hotkey for the USS AGN although there is a correlation with li2 keV- There is no dependence of the [OIII] equivalent width on the Balmer line FWHM.

The Baldwin effect

The ‘Baldwin effect’ is a term originally used to describe the anti-correlation be­ tween the equivalent width of the CIVA1549 line and the continuum luminosity, L„, at that line centre (Baldwin 1977; Baldwin et a1. 1978). Kinney, Rivolo and Koratkar (1990), from IUE spectra of 101 quasars and 88 Seyferts, found that this effect extends over seven decades in luminosity and is also seen in the Lya line. I have tested the optical spectra of the USS AGN for the Baldwin effect and I find weak correlations for the Ha and H/? lines (coefficient=0.5; note that this is in the opposite sense to the Baldwin effect) and no significant correlation for HeIA5876, [OIIIJA5007 and MgIIA2798. There is a correlation between the Ha EW and redshift (coefficient=0.7) which may make a contribution to the EW(Ha)/L„ correlation, but there is none between the H/? EW and redshift.

8.4.3 Optical Fell emission line parameters 8.4.3.1 Fell luminosities I find very strong correlations of the Fell red and blue blend luminosities, L blue and Lred, with the optical luminosity (I also find these correlations in the Stephens sub-sample of narrow-line objects) and the optical power-law index. There are

correlations with Lo. 2fceV and a very weak correlation of L blue with L2keV (the associated correlation probability is 70%). There axe strong correlations between Lu ue and a ox and a 8X but these are based on only 5 data pairs. There is an anti­ correlation between L blue and a oa, but both quantities correlate similarly with redshift. Table 8.3 also lists a weak anti-correlation (-0.7) between the L&/ue/H/? ratio and the hard X-ray luminosity, with an associated correlation probability of 85%. This result is in agreement with the results of Joly (1987), whose model requires that the ionizing flux must be low to suppress the H (3 emission relative to Fell. However, this correlation is based on few data pairs (5; although the upper limits in general do not contradict the suggested correlation). xuo

Correlations with other emission lines

There is a weak anti-correlation of L blue and Lre<* with the FWHM of H a (coef­ ficients of -0.5 and -0.7 for L blue and Lred respectively). The Fell luminosities are strongly correlated with the broad line luminosities (coefficients ~0.9) but less so with the [OIIIJA5007 luminosity (coefficients ~0.7-0.8). The Lbiue/^red ratio is weakly correlated with the Hel luminosity (coefficient=0.7 for 8 data pairs). There are also anti-correlations between the Lrerf/H/9 and Lf,/ue/H/? ratios and the broad line EWs (~0.6-0.8). There is a weak correlation between the Fell luminosities and the FWHM of [OIIIJA5007 and, although I found no correlation between the [OIIIJA5007 FWHM and Fell luminosity for the Stephens (1989) sample as a whole, for the narrow-line Stephens objects there is a weak; correlation with L red (coefficient=0.7 based on 7 data pairs).

8A.3.2 Fell equivalent widths

Zheng and O’Brien (1990) found that the optical Fell EW is higher when the H/? FWHM is narrow and suggested that this is due to an aspect dependence which is consistent with the models of Collin-Souffrin et al. (1988; see Section 1.3.5.2 for a brief description). Jackson and Browne (1991) report a correlation between the Fell/[OIII] ratio and the ratio of core to extended radio luminosity, R and also attribute it to the aspect dependence of Collin-Souffrin et al.. I have found a dependence similar to that reported by Zheng and O’Brien (1990) between the FeIIrerf and FeII&/ue EWs and H/? FWHM for the Stephens sample (38 objects). In Figure 8.15 I plot H (3 FWHM against Fell equivalent width for the USS AGN and compare them with the Stephens sample. Upper limits for the USS AGN are indicated but these were not available for the Stephens AGN. The plot shows that for the FeIIred EW, the USS AGN lie within the Stephens range and have relatively high EWs (again, I point out that upper limits for the Stephens sample are not included). For FeII&/ue, several of the USS AGN have relatively low EWs at low FWHM when compared to the Stephens AGN but the remainder lie within the range of the Stephens sample. However, there may be systematic differences in the measurements of Fell between the USS and the Stephens samples, even though I have tried to follow the same procedure. For both samples, there is an absence of objects at high equivalent width and The Muitiwavelength Analysis 167

8000 8000

6000

CO a 4 0 0 0 4000

2000

0 0 20 40 60 80 100 80 10040 EW (FeIIbluc) EWCFellJ

Figure 8.15 : The B./3 FWHM plotted as a function of the equivalent widths of the red and blue Fell blends for the USS AGN (asterisks in boxes). Values taken from the Stephens sample are plotted for comparison (diamonds). Upper limits to the Fell equivalent widths are indicated by arrows. Secure and non-secure sources are included. high FWHM. There axe no significant correlations in the USS data between the Fell equivalent widths and the H/? and Ha FWHM. However, the range of FWHM for the USS is comparatively narrow; the Zheng and O’Brien (1990) sample extends to FWHM(H/?) of 11,000 km s-1 and Stephens to 8,000 km s-1, whereas for the USS the FWHM lie mostly below ~3,000 km s-1.

8.4.3.3 Fell emission and the soft X-ray excess. In a sample of 9 low-redshift (z<0.3) quasars, Wilkes, Elvis and McHardy (1987) found that the optical Fell emission was correlated with the index of a single power-law fitted to IPC data in the 0.1-3.5 keV range, in the sense that AGN with steeper soft X-ray slopes (ie. soft excess objects) showed stronger Fell emission (at 99.8% significance). Remillard and Schwartz (1987) found a similar relation from the optical spectroscopy and EXOSAT spectra of hard X-ray selected quasars, as did Kruper, Urry and Canizares (1990) in the Einstein spectra of 11 Seyfert galaxies. However, Zheng and O’Brien (1990), in their sample of 33 predominantly low-redshift QSOs, found no correlation between the soft X-ray slope and optical Fell EW. Considering the properties of the sample as a whole, the X-ray spectra of those 168 Chapter 8

USS AGN for which I have measured Fell parameters are best fit with an ‘average’ single power-law of index 6 (see Section 8.4.1.1). This is much steeper than that inferred for the HGLS (1.5) and for samples of X-ray bright AGN detected with Einstein in the 0.1-4.0 keV range (~1; Canizares and White 1989; Wilkes and Elvis 1987; Kruper, Urry and Canizares 1990). The Fell equivalent widths of the USS AGN are generally high when compared to the Stephens sample, which would be consistent with the Wilkes et al. effect. The a 8X may be considered as a measure of the slope of the composite X-ray spectrum for individual objects, but unfortunately there are few measurements of a 8X for those sources which have optical spectra (due to the high number of upper limits to the hard X-ray flux) and the a 8X has a strong redshift dependence

(due to the redshift dependence in Lo. 2 fcev)- With these caveats, I have found no correlations between the equivalent widths of the red and blue Fell blends and a 8X. There is a correlation between L&/ue and a 8X, although both of these quantities are also correlated with redshift. There are also correlations of Lred and L&/ue with Lo.2fceV*

8.4.4 Optical and IR continuum slopes 8.4.4.1 Optical - total spectrum The distribution of the power-law index measured from the optical spectra, a opt , is shown in Figure 8.16a. This index represents the slope fitted to the composite optical spectrum, ie. only obvious emission-line features have been removed before fitting the power-law; no corrections have been made for the Balmer continuum or blended Fell emission which might resemble a continuum (see Section 5.1.2.4). The diagram illustrates that the USS selects AGN with a wide range of op­ tical continuum slopes. There is a strong anti-correlation of a opt with the optical luminosity in the sense that brighter objects are bluer (Figure 8.16a). There are weaker anti-correlations of a opt with Lo. 2 JteV (Figure 8.16b) and L 2 keV- Both broad and narrow line luminosities are strongly anti-correlated with the optical power-law index (except MgIIA2798) and the anti-correlation is strongest for the broad lines (see Figure 8.16d and e). Also, the broad line EWs (except Mgll) are anti-correlated with a opt (especially Ha) but there is no significant anti-correlation for [OIIIJA5007.

The strong anti-correlations of a opt with Lopt, L 2500A? Balmer and HeIA5876 line luminosities may prove to be powerful indicators of the optical spectral pa- The Multiwavelength Analysis 169 rameters, especially if these dependences are seen in the lines and continuum of a single, varying source.

5 32 ' “ 1 1 - i 32 K 31 (b) 31 (c) ' # * > 30 * 30 X X 4 M X A X X 29 ** * X 29 X

Log X * * M K> 00 X * 28 Z 27 X * ' o 3 1c < 27 I ■ ■ ____ 26 .1 —^------1------. o |Uh z 2

1

OLL i i i l_L -0.5 0.5 1.5 2.5 -0.5 0.5 1.5 2.5 -0.5 0.5 1.5 2.5 a,'opt a,'opt a opt

Figure 8.16 : (a) The distribution of total optical power-law indices for the USS AGN. (b)-(e) Correlations of various line and continuum parameters with oc0pt f°r the USS AGN. (b) Optical luminosity at 2500A plotted against otopt. (c) Soft X-ray luminosity at 0.2 keV plotted against OLopt. (d) Jl/3 luminosity plotted against a 0p t. (e) [OIII]A5007 luminosity plotted against OL0pt-

8.4.4.2 Optical - nuclear spectrum

For the sample of AGN for which IR images were taken, the contribution to the total optical spectrum from the host galaxy has been removed to produce the spectrum of the nucleus. The method of galaxy subtraction, plots of the nuclear spectra and values for the index of a power-law fitted to the spectrum are given in Chapter 6. The median index of a power-law fitted to the nuclear optical spectra of the USS AGN is 0.6 (the mean is 0.56±0.16). In a sample of 718 bright quasars, Francis et al. (1991) find a median optical index of 0.32, while the median for the Neugebauer et al. (1987) sample is 0.2. This implies that the USS AGN optical spectra (in the IR sample) are relatively flat. Kriss (1988) found that quasars with log(Li/im)<30 have small or absent UV bumps, and are therefore relatively flat. There are only 10 measurements of the nuclear optical spectra for the USS AGN; all of these have Log(Li/im)<30.3. Furthermore, plots of multicomponent fits to the USS AGN IR-to-X-ray spectra (Chapter 6) show that a steep rise to the blue above the underlying power-law, indicative of a strong big blue bump, is seen in only 2 out of 10 AGN. Thus the optical slopes of the USS AGN appear to be consistent with other samples of AGN; although their soft X-ray components are unusually strong, their optical/UV big blue bumps are not.

For Lo.2 keV and L2500A? there was no significant change in the anti-correlation with the nuclear optical index compared with the total optical index. Correla­ tions with li2keV were not possible because there were too few data pairs. Anti­ correlations of the nuclear optical index with broad and narrow line luminosities were strong but they were even stronger with the total optical index. However, the opposite is true for the Fell blends whose luminosities were more strongly correlated with the nuclear index than the total (but note that only 8 and 5 data pairs were available for the blue and red blends respectively). There are weak anti-correlations between broad line EWs and nuclear optical index but again, correlations with the total index are stronger.

8.4.4.3 Infra-red - total spectrum

The index of a power-law fitted to the integrated IR spectrum shows a weak (83%) anti-correlation with a ox only.

8.4.4.4 Infra-red - nuclear spectrum

Starlight from the host galaxy has also been removed from the IR data to produce nuclear IR J, H and K fluxes and the index of a power-law fitted to these three data points is given in Chapter 6. The average index of the nuclear IR power-law for the USS AGN is 0.85±0.18, which is considerably lower than the average index for the Barvainis (1990) sample (~2), ie. the USS AGN slopes are relatively flat. However, Barvainis (1990) has shown that the IR power-law index depends on IR luminosity; it rises rapidly from low-luminosities and then levels out to an index of 2.7. The USS AGN luminosities are all low relative to the Barvainis AGN and their IR power-law indices also low; this is consistent with the Barvainis distribution The Multiwavelength Analysis 171 for radio-quiet quasars. There is evidence for a weak correlation of the nuclear IR index with ho.ikeV

(87% based on 11 data pairs); correlation probabilities for L 2500J4 and \j2keV are 79% (12 data pairs) and 61% (6 data pairs) respectively. For the optical line luminosities, while there are no significant correlations with the total IR index, there are correlations with the nuclear IR index for Ha (a probability of 92% for 9 data pairs) and H f3 (91% for 10 data pairs). For the narrow [OIIIJA5007 FWHM, there is a weak correlation with the nuclear IR index (85% for 9 data pairs).

8.4.4.5 IR to X-ray underlying power-law Again, for the 12 objects for which IR images have been taken, the index of a power-law connecting the IR to the hard X-rays ( o tiR -x ) has been calculated from three component fits (galaxy + soft X-ray excess -f IR to X-ray power-law) to the multi wavelength spectra (see Section 6.2). These fits and the power-law indices are given in Chapter 6. The average o jr - x for the USS AGN is 0.73, close to the ‘canonical’ X-ray slope (0.7). There are no significant correlations between o tiR -x and the continuum (IR, optical and X-ray) parameters. There are however, correlations between <*i r - x and the FWHM of H/? (a probability of 93% for 10 data pairs) and a weak corre­ lation for the Ho FWHM (80% for 9 data pairs). c?

Sum m ary

1. By comparing the H/? FWHM distribution of the USS with other X-ray se­ lected samples, I find that the permitted fines of the USS AGN are biased to narrow widths. 2. Approximately one third of the USS AGN hard X-ray luminosities are low relative to the X-ray selected comparison samples (the EMSS and HGLS), while the remainder are typical of the EMSS and HGLS. 3. The optical luminosities of the USS AGN are not distinguishable from those of the EMSS and HGLS. 4. The optical-to-X-ray ratio, a ox, is relatively high for a third of the USS AGN, suggesting that the hard X-ray luminosity is depressed relative to the opti­ cal. For the remaining two-thirds, a ox is typical of the EMSS and HGLS, indicating that the strong soft X-ray excess may be superposed on a ‘normal’ underlying optical-to-X-ray continuum. Also, there is tentative evidence of an anti-correlation between Balmer fine FWHM and a ox for the USS. 5. The equivalent widths of the permitted fines (Balmer fines and HeIA5876) are weak relative to the Stephens AGN and there is some evidence for a correlation of their strength with that of the hard X-ray component. 6. Optical Fell emission is generally seen in objects which have no significant hard X-ray flux. Furthermore, there is weak evidence for an anti-correlation between the FeII/H/? flux ratio and hard X-ray luminosity, favouring the Joly model for optical Fell production. Evidence for a dependence of Fell strength on the soft X-ray slope is inconclusive. 7. The narrow fine ([OIII]A4959,5007) strengths are typical of other X-ray se­ lected AGN. 8. The IR properties of the USS AGN are not distinguishable from those of the optically selected samples of Worrall and Kriss. 9. There is no evidence for an unusually strong big blue bump in the optical spectra, despite the presence of a strong ultra-soft X-ray excess. 10. There are very tight correlations between optical fine and continuum lumi­ nosities of the USS AGN with the slope of the optical continuum, in the sense that the steeper the rise to the blue, the stronger the fine and continuum luminosities. These correlations are strongest for the composite (ie. nuclear plus galaxy) continuum slope. CHAPTER 9

Discussion and Conclusions

The USS survey selects objects which have a high ratio of counts in the 0.16 keV to 0.56 keV energy band compared to the 0.56 keV to 1.08 keV energy band. This favours objects with a narrow range of spectral properties which, for AGN, is redshift dependent because the selection is made in the frame of the observer (see Section 8.3.1). Most of the AGN in the sample have redshifts below 0.5 and a soft component with a blackbody temperature near to a mean of 10 eV in the rest frame. Of necessity, the Galactic absorption column in the line of sight to the USS AGN must be low in order for us to see soft X-rays. The UV and optical reddening is therefore also small (Nh = 2 x 1020 cm-1 corresponds to Av=0.1). Further, the extinction in the host galaxy must also be low! USS AGN are selected according to their soft component flux, and exhibit a much larger range of hard X-ray to optical luminosity ratio than samples selected at higher X-ray energies. A third of the USS AGN have very low or no detected hard X-ray flux in the IPC energy range. An example is E0132^111 (Figure 7.3) whose hard X-ray flux is at least a factor of 30 lower than that of the nearby Seyfert Mkn 841 as a proportion of the optical, UV and soft X-ray continuum. The existence of objects like E0132-411 challenges the suggestion (eg. Avni and Tananbaum 1982) that all broad-line AGN, ie. all Seyfert Is and quasars, are hard X-ray loud.

“Verbosity leads to unclear , inarticulate things. ”

— DAN QUAYLE 174 Chapter 9

9.1 X-RAY AND OPTICAL CONTINUA The X-ray to optical continua of the USS AGN fall into two main categories (see Section 8.3.3.1). For approximately two-thirds of the AGN, the strong soft component is superposed on a ‘normal’ underlying hard X-ray continuum. The remaining objects have a significant deficit of X-rays, or no detected hard X- ray component, when compared to other X-ray selected AGN. In both cases, the optical luminosities of the USS AGN are typical of X-ray selected AGN for which USS-type soft X-ray excesses are not observed. (It is important not to rule out the possibility that intrinsic soft X-ray excesses, which are not observed due to absorption or inclination effects, may occur in other, and perhaps all, AGN.)

9.1.1 X-ray and optical continua in an accretion disk model For a thick accretion disk model, the optical portion of the disk spectrum is bright­ est when the disk is viewed face-on (Madau 1988; see Figure 1.3). The difference in the observed optical luminosity between a pole-on view and an edge-on view of a thick disk is half a decade when reflection of photons from the funnel walls is taken into account. Therefore, if the presence of a soft X-ray excess in the USS AGN indicates that these contain face-on thick accretion disks, we would expect them to appear optically bright when compared with a sample of randomly orien­ tated AGN. This would be in agreement with the results of Jackson and Browne (1991) who suggest that the optical emission is anisotropic and, for an accretion disk, would be strongest when the disk is seen face-on (see Section 8.4.2.2). For the thin accretion disk model (Sun and Malkan 1989), soft X-ray excesses at ~0.2 keV are seen in edge-on systems. The disk spectrum shifts to higher energies as the angle of inclination increases and plots of the Sun and Malkan models show that the optical emission observed from a thin disk viewed edge-on is more than two decades fainter than when seen face-on (see Figure 1.3). Neither of these trends are observed in the USS AGN; their optical luminosity distribution is typical of other X-ray selected AGN (see Figure 8.4).f

f Note however, that for the thick disk, if reflections from the funnel walls are neglected, then the model predicts a much smaller range of observed optical luminosities, approximately a quarter of a decade (Madau 1988). In this case, the predicted differences in optical luminosity between the USS AGN and other X-ray selected objects would be small. Discussion and Conclusions 175

9.1.2 X-ray and optical continua in the cool clouds model

Ferland and Rees (1988) have calculated the expected spectrum based on the Guilbert and Rees (1988) ‘cool clouds’ model. Two cool clouds spectra from Ferland and Rees are shown in Figure 9.1 compared with the multi wavelength spectrum of E0132-411. Their results show that more hard X-rays are absorbed by the clouds as the volume filling factor increases, and the big blue bump grows (see Figure 9.1). This steepens the slope of the optical continuum, but the optical flux in the range that we are considering (ie. from 2500A - 7000A) shows little change.

-10 Key -11 e = -3 e = -5

(if -12

3 -13

-14

-15 14 15 17 18 1916 Log v

Figure 9.1 : The multiwavelength spectrum of EO132-411 with models from Ferland and Rees (1988). The optical and UV spectra of E0132-411 are plotted together with the Einstein X-ray data in three energy bands. The X-ray data axe modelled (thick solid line) with two components, a soft blackbody with Teff=10eV and a hard power-law with an index of 0.7. Plotted for comparison, axe models from Ferland and Rees (1988) for a density N = 1014 cm-3 and filling factors, € = -5 (dotted line) and -3 (dashed line).

For this model then, we would expect AGN with intrinsic soft X-ray excesses to have optical luminosities which are typical of other AGN and optical power-law indices which are generally steeper. The former is observed for the USS AGN, the latter we are unable to test because there is no X-ray-selected data with which to compare them (but see Section 8.4.4 for a discussion of the USS AGN optical power-law indices). Ferland and Rees caution that this is a preliminary model which should not be directly compared with observations, although the 176 Chapter 9

basic principles, ie. that photoelectric absorption removes X-rays in the medium range and reradiates the energy in the optical to soft X-ray region, still hold.

9.2 THE PREDOMINANCE OF NARROW BALMER LINES A striking aspect of the USS AGN is the high proportion of narrow-line objects in the sample. This implies that the broad-line clouds in these systems have low velocities in the line of sight and suggests a link between this and the visibility of the soft X-ray component by which the USS sources are defined. We consider two possibilities; an inclination effect or a distant BLR. If the broad-line emitting clouds are associated with an accretion disk, or are otherwise confined to circulate in a plane, then an explanation for the narrow permitted line widths would be that we are viewing the system face-on, ie. the BLR clouds are circulating in the plane of the sky. Alternatively, assuming that the velocities of the clouds are associated with motion due to the gravitational field of a black hole, the cloud velocities would decrease as their distance from the black hole increased. Thus, the lines would also be predominantly narrow if the BLRs of the USS AGN were further away from the central black hole than for other AGN.

9.2.1 A face-on accretion disk... If the narrow lines are evidence of a face-on BLR and the BLR lies in the plane of an accretion disk, then the accretion disks in AGN with strong soft X-ray excesses would be viewed face-on. Thick disks (where the strong soft X-ray excess is viewed face-on; Madau 1988) would therefore be favoured over thin disks (where the soft excess is viewed in edge-on systems; Sun and Malkan 1989). Core-dominated radio quasars, which are believed to be extended double sources seen end-on, also have predominantly narrow permitted line widths (Wills and Browne 1986). If the radio axis lies perpendicular to the plane of a disk-shaped BLR, then this agrees with a face-on accretion disk geometry. There is tentative evidence that the permitted lines are narrower when the ratio of optical to hard X-ray luminosity, a ox, is high (see Section 8.4.1.2). In the context of BLR clouds associated with an accretion disk, this change in a ox would be an inclination effect, ie. the optical flux decreases relative to the hard X-ray flux as the angle of inclination increases. Such a relationship would be in agreement with the thick disk model (see Section 9.1.1) provided that the hard X-rays were emitted comparatively isotropically. Discussion and Conclusions 177

9.2.2 ...or a more distant BLR? Rather than being associated with a face-on BLR, the narrow lines may be evidence of a BLR which lies further out from the black hole. For NGC 5548, Krolik et al. (1991) have reported a relationship between the broad line FWHM and the time taken for the line to respond to variations in the continuum (a measure of the distance to that line-emitting region), and found that the FWHM of the emission lines decrease as the distance increases. A ‘distant BLR’ geometry could be tested by monitoring the lines and continuum of a varying USS AGN and measuring the distance to the BLR. In the case of a distant BLR, an a ox vs. broad line FWHM correlation becomes more intriguing, as the distance of the BLR from the central source would be linked to the ratio of the optical and hard X-ray luminosities.

9.3 THE STRENGTH OF THE PERMITTED LINES The Balmer line and HeIA5876 equivalent widths of USS AGN are generally low compared to other X-ray selected AGN and, since we know that the optical con­ tinuum luminosities are typical (see Figure 8.4), this implies that the broad line luminosities are low. Ha and B./3 are produced by photons in the 13.6-54.4 eV range (Krolik and Kallman 1988) as well as by hard X-rays; HeIA5876 is produced by 300-500 eV photons. Despite the presence of a strong soft X-ray flux, these lines are weaker than in AGN where a soft X-ray excess is not observed. If we assume that the Hel and Balmer photons are emitted isotropically, then these results indicate that the soft X-ray excess emission does not communicate with the Balmer and Hel line-emitting regions. The correlation (albeit tentative due to the small number of data pairs) between broad line equivalent width and the hard X-ray component luminosity implies that the hard X-ray flux does communicate with the Balmer and Hel line regions.

9.3.1 Weak permitted lines in a thick disk model... Soft X-rays from a thick disk are emitted preferentially in a cone of radiation along the axis of the disk (see Section 1.3.3.1). Assuming that the BLR is in the plane of the disk, it will lie in the soft X-ray shadow and there would indeed be no communication between the soft X-rays and the BLR. Therefore the line emission would depend on the observed hard X-ray flux, assuming that the latter is emitted isotropically. 178 Chapter 9

9.3.2 .. and the cool clouds model .. In the cool clouds model (see Section 1.3.4), both the hard and soft X-ray emis­ sion are presumed to be emitted isotropically. Therefore the line-emitting region must be shielded in some way from the ionizing flux and there must be no ab­ sorbing material along the line of sight to the observer (for the soft X-rays to remain visible). For instance, in the NGC 5548 model of the BLR, the LIL region, where the Balmer and Hel lines are produced, is flattened and lies behind the HIL region which absorbs the soft X-rays (see Section 1.3.5.3). Netzer’s model (Sec­ tion 1.3.5.1) assumes a flattened distribution of ‘standard’ BLR clouds, in which the soft X-rays are absorbed in the HIL region in the front of the clouds, before reaching the LIL region.

9.3.3 ...or a more distant BLR? If the BLR is flattened and lies a long way from the central ionizing source, the ionizing flux is more dilute and the Balmer and HeIA5876 lines would be relatively weak and narrow (shielding from the hard X-rays is not necessary in this case).

9.3.4 Anisotropic broad line emission Based on the results from the NGC 5548 campaign which showed that, for this object, the BLR lies closer to the central object than predicted by photoionization models (see Section 1.3.5.3), Ferland et ai. 1992 have considered the effect of the much stronger ionizing flux on the BLR. They find that most broad emission lines will be optically thick and line emission will then be emitted preferentially back towards the central ionizing source. While this is based on a spherical BLR, they do not rule out a disk-shaped distribution. Therefore, assuming that the BLR is disk-shaped and that the FWHM de­ scribes the viewing angle, then we would see relatively weak broad lines for a BLR seen face-on and narrow FWHM in the permitted lines. We would also see a decrease in the [OIIIJA5007/H/? flux ratio as the H/? FWHM increases (assuming that the [OIIIJA5007 line is emitted comparatively isotropically); this is seen for the USS and Stephens samples (see Section 8.4.2.1). A soft X-ray excess would also be observed if; (i). this BLR lies in the plane of a thick disk or (ii). if the line of sight to a ‘cool clouds’ region has a low absorbing column. Indeed, Krolik et aJ. (1991) deduced that the low-ionization zone of NGC 5548 (see Section 1.3.5.3) is disk-shaped, and lies edge-on to the line of sight (but note Discussion and Conclusions

that there is also a soft X-ray excess in this object, Branduardi-Raymont 1989; Turner and Pounds 1989; Nandra et al. 1991).

9.4 OPTICAL FEII EMISSION

The optical Fell lines are generally believed to be emitted in the presence of a strong X-ray flux (see Section 1.3.5.4) so we might expect the strength of the Fell emission to depend on the hard X-ray component. However, the luminosities of the optical Fell blends axe only weakly correlated with haxd X-ray luminosity and there are no significant correlations for the EWs (see Section 8.4.3). While the hard X-ray luminosities of the USS AGN are generally low (see Section 8.3.2), the FeIIred EWs are generally high (although this may not be true for the blue blend; see Section 8.4.3.2). There are secure USS AGN which have little or no hard X-ray flux yet strong Fell emission, eg. E0132—411 and E1423+201. In the model of Joly (1987), no haxd X-ray flux is required to produce optical Fell; indeed the model demands very weak or no ionizing radiation incident on the Fell-emitting region (see Section 1.3.5.4). This is necessary to suppress the H(3 emission relative to Fell. Only two of the USS AGN for which we have optical spectra, have measurable Fell emission and a strong hard X-ray flux, E1425+169 and E0844+377. These objects also have the lowest measured FeII/H/? ratio.

9.4.1 A dependence on H (3 F W H M

There is evidence that the Fell strength decreases as the H/? FWHM increases (see 8.4.3.2). If the observed H/? FWHM is a measure of the viewing angle, this implies that Fell emission is anisotropic and is emitted preferentially along the axis of the BLR. Alternatively, if the H/? FWHM is a measure of the distance to the BLR, then this suggests that the Fell emission is greater when the BLR is further away from the central source. We note that in the Joly (1987) model, the FeII/H/9 flux ratio is greatest when the incident radiation is weak or zero.

9.4.2 The relationship with the hard X-ray flux

We consider the nature of the Fell emission from the USS AGN in the context of hard X-ray dependent models of their production. For those sources with weak observed hard X-rays, the hard X-ray emission must be enhanced in the direction of the optical Fell region. Any enhanced hard X-ray emission must be in a direction which is not in the line of sight to the observer and it must be directed away from 180 Chapter 9 the region which produces the Balmer lines (assuming that these also depend on the hard X-ray flux and are emitted isotropically). The observed hard X-ray emission must correlate with the hard X-rays which reach the Balmer line region. A separation of the optical Fell and Balmer line regions disagrees with the standard model-type clouds where Balmer lines and optical Fell are produced in the same region, ie. in the back of the clouds. In the Collin-Souffrin et ai. (1988) model, again the Balmer lines and optical Fell are produced in the same region; the outer part of an accretion disk. If the HeIA5876 and Balmer photons axe emitted preferentially in the plane of the BLR (see Section 9.3.4) while the Fell photons are emitted along its axis, then the Fell and broad lines could be formed in the same region, (although enhanced X-ray emission in this zone is still necessary) and the ‘standard’ model and the Collin-Souffrin et ai. model could then apply. In this situation, we would expect to see an anti-correlation between the observed strengths of the Fell and the Balmer lines. However, I have looked at the H/3 and Fell EWs from the Stephens sample (which covers a wider range of Fell and Balmer line properties than the USS AGN) and I could find no evidence for a correlation of this kind. For an angular dependent ionizing continuum model where a strong EUV excess is viewed in face-on disks, Netzer (1987) found that the Balmer lines and HeIA5876 were not formed in the same region as optical Fell. Formation of the Balmer and Hel lines was strongest above the poles and Fell and MgIIA2798 were strongest in the plane. However, the evidence suggests that Fell emission is strongest above the poles (see Section 8.4.3.2 and above). This model may be inappropriate for comparison with the USS too; the ‘big blue bump’ component of the model did not extend sufficiently into the soft X-rays for it to qualify as a USS-type spectrum and the effect of a weak hard X-ray component was not considered.

9.4.3 Optical Fell and the soft X-ray excess The presence of a soft X-ray excess has been linked to the strength of the optical Fell emission via evidence that the steepness of the soft X-ray slope is correlated with the optical Fell strength (see Section 8.4.3.3). Unfortunately, analysis of the USS AGN parameters is inconclusive regarding a possible dependence of optical Fell strength on the soft X-ray excess (mostly due to the strong redshift effect in Lq .2 keV - see Section 8.4.3.3) but as a dependence on the hard X-rays would Discussion and Conclusions 181 require the conditions discussed in Section 9.4.2, we consider a link with the soft X-ray excess as an alternative.

9.4.3.1 F ell emission from a thick disk... There are indications (see Section 8.4.3.2 and Section 9.4.2) that optical Fell emis­ sion is anisotropic and is emitted preferentially along the axis of an accretion disk. This would be consistent with Fell emission from the inner regions of a thick disk or from within the soft X-ray cone itself. It also removes the need for hard X-ray beaming. If the soft X-ray cone and the radio jet are aligned, this hypothesis would be in agreement with the model of Joly (1991), where optical Fell emission is produced in the interaction layer between the radio jets and their surrounding medium. However, the Fell emitting region must be shielded from the ionizing continuum in this case.

9.4.3.2 ...and from cool clouds In the cool clouds model, if the optical Fell emission was related to the strength of the soft X-ray excess, we would expect that the Fell EWs of the USS AGN would be relatively high compared to other X-ray selected AGN (since the optical luminosities would be normal; see Section 9.1.2). Evidence of this effect for the USS AGN is presented in Section 8.4.3.2. We would also expect to see a correlation of Fell strength with the slope of a single power-law fitted to soft-to-medium X- ray spectra (assuming that this is a measure of the hard X-ray flux absorbed). A dependence of this kind has been reported by previous authors (see Section S.4.3.3). When we consider the sample of USS AGN for which we have optical spectra (secure and non-secure), we find that the Fell strengths are relatively strong and the ‘average’ index of a single power-law fitted to their X-ray spectra is steep (6; see Section 8.4.3.3). Since we assume that the cool clouds spectrum is isotropic, it would be difficult to justify anisotropic Fell emission in this situation, if the H/? FWHM describes the angle of inclination of the BLR. This may be solved by assuming that the H (3 FWHM is a measure of the distance to the BLR. For the Joly model then, the Fell EW would increase with the distance as the ionizing continuum becomes more dilute, resulting in an anti-correlation between Fell strength and H/2 FWHM. 182 Chapter 9

9.5 INFRA-RED PROPERTIES

The near-IR properties of the USS AGN are not distinguishable from those of the optically-selected comparison samples (Worrall 1987; Kriss 1988). Their IR luminosities are generally low compared to these samples, but ratios of IR with optical and with X-ray luminosity show that they are not atypical of other low- luminosity AGN.

9.5.1 Dust emission

The IR component is believed to be due to radiation from dust which lies beyond the BLR with the near-IR emitted from grains which lie closest to the central source (see Section 1.2.5). There are two reasons why a strong soft X-ray excess might have no effect on the near-IR emission; either the dust does not lie in the path of the soft X-rays or there is no interaction between them. For a thick disk, if the dust lies in the obscuring torus (between the BLR and NLR - see Section 1.3.6) and the torus is co-planar with the disk, then the soft X-rays will not reach the dust. In the cool clouds model, the soft X-ray emission is assumed to be isotropic so we would expect the dust to lie in the soft X-ray path. However, Voit (1991) considered how X-ray spectra with different shapes and lumi­ nosities affected the emergent dust spectrum. Plots of the predicted dust spectra show no significant differences between a purely soft (0.2 keV bremsstrahlung) and hard (F,/ oc fc'-1) incident spectra, for a given total X-ray flux.

9.5.2 Soft excesses, narrow lines and warm IRAS Seyferts

The link between warm IRAS Seyfert Is (de Grijp et ai. 1985) and narrow-line objects (Ward 1988) - and the tempting connection with ultra-soft X-ray excesses - is an intriguing one, especially in the light of Voit’s models which indicate that there should be no dependence of the IR on the X-ray spectral shape. However, the IR/USS connection may be incidental and the true correlation may be between the narrow lines and a warm IRAS flux (see Ward 1988). A measurement of the mid-far IR fluxes in the USS AGN would identify where these possible trends lie. Uiscussion ana Lsonclusions loo

9.6 THE PICTURE OF A USS AGN In the light of these results, I present two possible geometries of an ultra-soft X-ray emitting AGN.

9.6.1 A face-on thick accretion disk We may be looking directly into the face of a thick accretion disk, along the soft X-ray cone, and perpendicular to the plane where the disk itself, the BLR and perhaps the dusty molecular torus which lies between the BLR and the NLR, all lie. The X-ray flux is isotropically emitted from the central source whereas the optical continuum is emitted preferentially along the axis of the disk. The permitted lines are narrow because we look onto the BLR face-on, and weak because they are emitted preferentially towards the central source which is perpendicular to our line of sight, or because the hard X-ray flux is weak. The near-IR dust may be spherically distributed or disk-shaped. Non-USS AGN may be thick disk systems seen at larger viewing angles, ie. outside the soft X-ray cone (and therefore soft X-rays may be directed away from the line of sight). Note however that this model predicts that face-on disks should be optically bright: this is not apparent when comparing the USS AGN to other X-ray selected objects.

9.6.2 Cool clouds around the central source Ultra-soft X-ray excess AGN may be a special case where the central energy source is surrounded by cool clouds with a relatively high volume filling factor. There is no accretion disk. The disk-shaped BLR lies further away from the central ionizing continuum than in other AGN, thus the permitted lines are narrow and weak. The continuum emitted by the clouds (optical to X-rays) is isotropic and the continuum shape may be related to the distance to the BLR. Again, the near-IR dust may be spherically distributed or disk-shaped. The clouds in non-USS AGN may have a smaller filling factor and, assuming isotropic continuum emission, there would be no soft X-ray excess beamed away from the line of sight.

These two models are not mutually exclusive, ie. there are some aspects which may be interchangeable. They are intended to be illustrations of the kind of structure that previous work, and the results from this thesis, point towards for active galaxies where a strong, ultra-soft X-ray excess is observed. 184 Chapter 9

9.7 CONCLUSION All of the AGN in this sample have a very strong flux of X-rays in the softest range of the Einstein IPC detector, ie. within 0.16 keV to 0.56 keV, compared to the flux in the medium range; 0.56 keV to 1.08 keV. Therefore we know that the amount of absorbing material along the line of sight to the AGN in our galaxy and the host galaxy must be low. The sample is dominated by narrow-line objects, implying a link between the presence of an observed soft X-ray excess and the relatively low velocity of the line-emitting material in the line of sight. If the FWHM describe the inclination of the BLR, this link favours thick accretion disks over thin, where a strong soft X-ray flux is seen in face-on systems. The Guilbert and Rees cool clouds are another promising mechanism for producing a soft X-ray excess. Alternatively, the low velocity of the permitted lines may be due to a more distant BLR. The separation from the central source to the BLR can be measured directly from the time-lag between line and continuum flux variations. Measurements for objects with a range of FWHM would confirm whether the line widths depend on the distance to the BLR. The hard X-ray flux of the USS is generally low compared to other X-ray selected AGN, and there is tentative evidence for a correlation with the permitted line strengths, which are also relatively low. Their optical luminosities are not distinguishable from those of other X-ray selected AGN. The optical Fell emission is strong when the hard X-ray flux is weak, contradicting hard X-ray dependent models for the production of the optical Fell blends and favouring the Joly model where no ionizing continuum is required. Many other aspects revealed by the analysis of these unusual AGN are in­ triguing, eg. the anti-correlation between a ox and Balmer line FWHM and the strong correlations between a opt and other optical parameters. Unfortunately, the redshift dependence in L 80ft may be masking evidence of its relationship with optical properties of the USS AGN, eg. the strength of the optical Fell emission. The analysis of a sample of AGN with a range of soft component strengths is re­ quired to search for dependences on the ultra-soft X-ray flux. Observations made with the ROSAT satellite will have superior spectral resolution and much greater sensitivity. With the support of a complete and thorough optical programme, a ROSAT sample of AGN at the lowest columns will provide a wealth of information for characterizing AGN and constraining current models. 185

R eferen ces

Allen, C. W. 1973, in “A s trophy si cal Quantities 3 rd• Ed.”, London: Athlone Press.

Allington-Smith, J. R. et al. 1989, M.N.R.A.S., 238, 603.

Antonucci, R. and Barvainis, R. 1988, Ap. J. Lett., 332, L13.

Antonucci, R. R. J. and Miller, J. S. 1985, Ap. J., 297, 621.

Arnaud, K. A., Branduardi-Raymont, G., Culhane, J. L., Fabian, A. C., Haz­ ard, C., McGlynn, T. A., Shafer, R. A., Tennant, A. F. and Ward, M. J. 1985, M.N.R.A.S., 217, 105.

Avni, Y., Soltan, A., Tananbaum, H. and Zamorani, G. 1980, Ap. J., 238, 800.

Avni, Y. and Tananbaum, H. 1986, Ap. J., 305, 83.

Baldwin, J. A. 1977, Ap. J., 214, 679.

Baldwin, J. A., Burke, W. L., Gaskell, C. M. and Wampler, E. J. 1978, Nature, 273, 431.

Bart hel, 1989, Ap. J., 336, 606.

Barvainis, R. 1987, Ap. J., 320, 537.

Barvainis, R. 1990, Ap. J., 353, 419.

Beckwith, S., Evans, N. J., Becklin, E. E. and Neugebauer, G. 1976, Ap. J., 208, 390.

Berriman, G. 1990, Ap. J., 354, 148.

Blandford, R. D. 1984, in “Proc. of the Manchester Conference on Active Galactic Nuclei”, ed. J. Dyson, Manchester: University Press, pg. 281.

Blandford, R. D. and Payne, D. G. 1982, M.N.R.A.S., 199, 883.

Blandford, R. D. and Znajek, R. L. 1977, M.N.R.A.S., 179, 433.

Blitz, L., Magnani, L. and Mundy, L. 1984, Ap. J. Lett., 282, L29.

Blumenthal, G. R., Keel, W. C. and Miller, J. S. 1982, Ap. J., 257, 499.

Boulanger, F. and Perault, M. 1988, Ap. J., 330, 964.

Branduardi-Raymont, G. 1989, in “Active Galactic Nuclei”, ed. D. E. Osterbrock and J. S. Miller, Dordrecht: Kluwer, pg. 177. 186

Branduardi-Raymont, G., Mason, K. O., Murdin, P. G. and Martin, C. 1985, M.N.R.A.S., 216, 1043. Browne, I. W. A. and Wright, A. E. 1985, M.N.R.A.S., 213, 97. Canizares, C. R. and White, J. L. 1989, Ap. J., 339, 27. Chanan, G. A., Margon, B. and Downes, R. A. 1981, Ap. J.Lett., 243, L5. Cheng, F. H., Gaskell, C. M., and Koratkar, A. P. 1991, Ap. J., 370, 487. Clavel et a1. 1991, Ap. J., 366, 64. Clavel, J., Wamsteker, W. and Glass, I. S. 1989, Ap. J., 337, 236. Collin-Souffrin, S., Dyson, J. E., McDowell, J. C. and Perry, J. J. 1988, M.N.R.A.S., 232, 539. Comastri, A., Setti, G., Zamorani, G., Elvis, M., Giommi, P., Wilkes, B. J. and McDowell, J. C. 1992, Ap. J., 384, 62. Cordova, F. A. and Kartje, J. F. 1989, in “Proc. 23rd ESLAB Symposium, Two Topics in X-ray Astronomy”, ed. J. Hunt and B. Battrick, ESA SP-296, 2: AGN and the X-ray Background, pg. 843. Cordova, F. A., Kartje, J. F., Thomson, R. J., Mason, K. O., Puchnarewicz, E. M. and Hamden, F. R. (C92) 1992, Ap. J. Suppl., (in press). Courvoisier, T. J.-L. and Robson, E. I. 1991, Scientific American, 264, 24. Cutri, R. M., Wisniewski, W. Z., Rieke, G. H. and Lebofsky, M. J. 1985, Ap. J., 296, 423. de Grijp, M. H. K., Miley, G. K., Lub, J. and de Jong, T. 1985, Nature, 314, 240. de Robertis, M. M. and Osterbrock, D. E. 1986, Ap. J., 301, 727. de Ruiter,H. R. and Lub, J. 1986, Astr. Ap. Supp., 63, 59. Done, C., Madejski, G. M., Mushotzky, R. F., Turner, T. J., Koyama, K. and Kunieda, H. 1992, in “Proc. 2nd Annual October Astrophysics Conference in Maryland, Testing the AGN Paradigm”, (in press). Doxsey, R., Bradt, H., McClintock, J., Petro, L., Remillard, R., Ricker, G., Schwartz, D. and Wood, K. 1983, Ap. J. Lett., 264, L43. Dumont, A. M. and Collin-Souffrin, S. 1990, Astr. Ap., 229, 313. Edelson, R. A. and Malkan, M. A. 1986, Ap. J., 308, 509. Edelson, R. A. and Malkan, M. A. 1987, Ap. J., 323, 516. 187

Elvis, M., Green, R. F, Bechtold, J., Schmidt, M., Neugebauer, G., Soifer, B. T., Matthews, K. and Fabbiano, G. 1986, Ap. J., 230, 540. Elvis, M., Lockman, F. J. and Wilkes, B. J. 1989, Astron. J., 97, 777 (ELW). Fabbiano, G. 1986, P.A.S.P., 98, 525. Fabbiano, G., Elvis, M., Carleton, N., Willner, S. P., Lawrence, A. and Ward, M. J. 1986, in “Quasars”, ed. G. Swarup and V. K. Kapaki, Dordrecht: Reidel, pg. 85. Fabian, A. C., Guilbert, P. W., Arnaud, K. A., Shafer, R. A., Tennant, A. F. and Ward, M. J. 1986, M.N.R.A.S., 218, 457. Feigelson, E. D. et al. 1986, Ap. J., 302, 337. Ferland, G. J., Korista, K. T. and Peterson, B. M. 1990, Ap. J. Lett., 363, L21. Ferland, G. J., Peterson, B. M., Horne, K., Welsh, W. F., and Nahar, S. N. 1992, Ap. J., 387, 95. Ferland, G. J. and Rees, M. J. 1988, Ap. J., 332, 141. Foley, A. R. and Barthel, P. D. 1990, Astr. Ap., 228, 17. Francis, P. J., Hewett, P. C., Foltz, C. B., Chaffee, F. H., Weymann, R. J. and Morris, S. L. 1991, Ap. J., 373, 465. Garrett, M. A., Walsh, D. and Carswell, R. F. 1992, M.N.R.A.S., 254, 27P. Gioia, I. M., Maccacaro, T., Schild, R. E., Stocke, J. T., Morris, S. L. and Henry, J. P. 1990, Ap. J. Supp., 72, 567 (EMSS). Giommi, P., Tagliaferri, G., Beuermann, K., Thomas, H-C., Graser, U., Brand­ uardi-Raymont, G., Mason, K. O., Mittaz, J. P. D., Murdin, P., Smale, A., Pooley, G., Tuohy, I. and Brissenden, R. (HGLS) 1991, Ap. J., 378, 77. Goodrich, R. W. 1989, Ap. J., 342, 224. Green, R. F., Schmidt, M., and Liebert, J. 1986, Ap. J., 310, 291. Grindlay, J. E., Steiner, J. E., Forman, W. R., Canizares, C. R. and McClin- tock, J. E. 1980, Ap. J.Lett., 239, L43. Guilbert, P. W., Fabian, A. C. and Rees, M. J. 1983, M.N.R.A.S., 205, 593. Guilbert, P. W. and Rees, M. J. 1988, M.N.R.A.S., 233, 475. Haniff, C. A., Wilson, A. S. and Ward, M. J. 1988, Ap. J., 334, 104. Hewitt, A. and Burbidge, G. 1987, Ap. J.Supp., 63, 1 (HB). Horne, K. 1986, P.A.S.P., 98, 609. Hutchings, J. B., Crampton, D., and Campbell, B. 1984, Ap. J., 280, 41 (HCC). Hutchings, J. B., Crampton, D., Campbell, B.,Gower, A. C. and Morris, S. C. 1982, Ap. J., 262, 48 (HCCGM). Jackson, N. and Browne, I. W. A. 1991, M.N.R.A.S., 250, 422. Jackson, N., Browne, I. W. A., Murphy, D. W. and Saikia, D. J. 1989, Nature, 338, 485. Johansson, S. 1986, in “Proc. 94th Colloquium of the I.A.U.”, ed. R. Viotti, A. Vittone, and M. Friedjung., Dordrecht: Reidel, pg. 13. Johnstone, R. M., Fabian, A. C. and Nulsen, P. E. J. 1987, M.N.R.A.S., 224, 75. Joly, M. 1987, Astr. Ap., 184, 33. Joly, M. 1991, Astr. Ap., 242, 49. Kinney, K. L., Rivolo, A. R. and Koratkar, A. P. 1990, Ap. J., 357, 338. Kriss, G. A. 1982, in “Ph.D. Thesis”, M.I.T.. Kriss, G. A. 1988, Ap. J., 324, 809. Kriss, G. A. and Canizares, C. R. 1982, Ap. J., 261, 51. Kriss, G. A. and Canizares, C. R. 1985, Ap. J., 297, 177. Kriss, G. A., Canizares, C. R. and Ricker, G. R. 1980, Ap. J., 242, 492. Krolik, J. H. et a1. 1991, Ap. J., 371, 541. Krolik, J. H. and Begelman, M. C. 1986, Ap. J. Lett., 308, L55. Krolik, J. H. and Begelman, M. C. 1988, Ap. J., 329, 702. Krolik, J. H and Kallman, T. R. 1988, Ap. J., 324, 714. Kruper, J. S., Urry, C. M. and Canizares, C. R. 1990, Ap. J. Supp., 74, 347. Kwan, J. Y. and Krolik, J. H. 1981, Ap. J., 250, 478. Landau, R. et ai. 1986, Ap. J., 308, 78. Lightman, A. P. and Zdziarski, A. A. 1987, Ap. J., 335, 57. Liszt, H. S. 1983, Ap. J., 275, 163. Lockman, F. J., Jahoda, K. and McCammon, D. 1986, Ap. J., 302, 432. Maccacaro, T., Gioia, I., Burg, R., McLean, B. and Stocke, J. 1992, in preparation (EMSS). MacDonald, D. and Thorne, K. S. 1982, M.N.R.A.S., 198, 345. Machalski, J. and Condon, J. J. 1983, Astron. J., 88, 1591. Madau, P. 1988, Ap. J., 327, 116. Malkan, M. A. 1984, in “X-ray and UV emission from Active Galactic Nuclei”, ed. W. Brinkman and S. Trumper, Munich: Max-Planck-Institute fur Ex- traterrestriche Physik, pg. 121. Malkan, M. A. and Sargent, W. L. W. 1982, Ap. J., 254, 22. Maraschi, L., Chiapetti, L., Falomo, R., Garilli, B., Malkan, M., Tagliaferri, G., Tanzi, E. G. and Treves, A. 1991, Astr. Ap., 247, 25.

Margon, B., Downes, R. A. and Chanan, G. A. 1985, Ap. J. Suppl., 59, 23 (MDC). Mathews, W.G. and Ferland, G. J. 1987, Ap. J., 323, 456. Mittaz, J.P.D. 1991, in “Ph.D. Thesis”, University of London. Moles, M., Garcia-Pelayo, J., Masegosa, J., Aparicio, A. and Quintana, J. M. 1985, Astr. Ap., 152, 271. Morrison, R. and McCammon, D. 1983, Ap. J., 270, 119. Mukai, K. 1990, P.A.S.P., 102, 212.

Mushotzky, R. F. 1984, in “Advances in Space Research”, 3, 10-13, pg. 157. Nandra, K., Pounds, K. A., Stewart, G. C., George, I. M., Hayashida, K., Makino, F. and Ohashi, T. 1991, M.N.R.A.S., 248, 760. Netzer, H. 1987, M.N.R.A.S., 225, 55.

Neugebauer, G., Green, R. F., Matthews, K., Schmidt, M., Soifer, B. T. and Bennett, J. 1987, Ap. J. Supp.,- 63, 615.

Neugebauer, G., Oke, J. B., Becklin, E. E. and Matthews, K. 1979, Ap. J., 230, 79.

Neugebauer, G., Soifer, B. T., Matthews, K., and Elias, J. H. 1989, Astron. J., 97, 957.

Norman, C. and Miley, G. 1984, Astr. Ap., 141, 85. O’Brien, P. T., Gondhalekar, P. M. and Wilson, R. 1988, M.N.R.A.S., 233, 801. Osterbrock, D.E. 1987, Lecture Notes in Physics, Vol. 307, . Osterbrock, D. E. 1977, Ap. J., 215, 733. 190

Osterbrock, D. E. and Mathews, W. G. 1986, Ann. Rev. Astron. Astrophys., 24, 171. Osterbrock, D. E. and Pogge, R. W. 1985, Ap. J., 297, 166.

Osterbrock, D. E. and Shuder, J. M. 1982, Ap. J. Supp., 49, 149. Padovani, P. 1989, Astr. Ap., 209, 27.

Padovani, P. and Rafanelli, P. 1988, Astr. Ap., 205, 53.

Penston, M. V. 1988, M.N.R.A.S., 233, 601.

Penston, M. V. and Perez, E. 1984, M.N.R.A.S., 211, 33P.

Perry, J. ., and Dyson, J. E. 1985, M.N.R.A.S., 213, 665.

Peterson, B. M. et ai. 1991, Ap. J., 368, 119.

Peterson, B. M., Foltz, C. B. and Byard, P. L. 1981, Ap. J., 251, 4.

Peterson, B. M., Foltz, C. B., Byard, P. L. and Wagner, R. M. 1982, Ap. J. Supp, 49, 469. Peterson, B. M., Foltz, C. B., Crenshaw, D. M., Meyers, K. A. and Byard, P. L. 1984, Ap. J., 279, 529. Phillips, M. M. 1977, Ap. J., 215, 746.

Phinney, E. S. 1981, in “Plasma Astrophysics”, ed. T. Guyenne, ESA SP-161, pg. 337.

Piro, L., Massaro, E., Perola, G. C. and Molteni, D. 1988, Ap. J. Lett, 325, L25. Pounds, K. A., Turner, T. J. and Warwick, R. S. 1986, M.N.R.A.S., 221, 7P.

Pringle, J. E. 1981, Ann. Rev. Astron. Astroph., 22, 471.

Rauch, K. P. and Blandford, R. D. 1991, Ap. J. Lett, 381, L39.

Rees, M. J. 1984, Ann. Rev. Astron. Astroph., 22, 471.

Rees, M. J. 1987, M.N.R.A.S., 228, 47.

Rees, M. J., Netzer, H. and Ferland, G. J. 1989, Ap. J., 347, 640.

Remilliard, R. and Schwartz, D. A. 1987, Bull. Am. Ast. Soc., 18, 915.

Richstone, D. O. and Schmidt, M. 1980, Ap. J., 235, 361.

Rowan-Robinson, M., Hughes, J., Jones, M., Leech, K., Vedi, K. and Walker, D. W. 1991, M.N.R.A.S., 249, 729. 191

Saikia, D. J., Junor, W., Cornwell, T. J., Muxlow, T. W. B., and Shastri, P. 1990, M.N.R.A.S., 245, 408. Schmidt, M. 1968, Ap. J., 151, 393. Schmidt, M. and Green, R. F. 1983, Ap. J., 269, 352. Shakura, N. I. and Sunyaev, R. A. 1973, Astr. Ap., 24, 337. Shapiro, S. L., Lightman, A. P. and Eardley, D. M. 1976, Ap. J., 204, 187. Shields, G. A. 1978, Nature, 272, 706. Shuder,J. M. and Osterbrock,D. E. 1981, Ap. J., 260, 437. Sikora M. 1981, M.N.R.A.S., 197, 257. Sitko, M. 1989, Bull. A. A. S., 20, 967. Stark, A. A., Gammie, C. F., Wilson, R. F., Ball, J., Linke, R.A., Heiles, C. and Hurwitz, M. 1992, Ap. J. Suppl., 79, 77. Steiner, J. E. 1981, Ap. J., 250, 469. Stephens, S. A. 1989, Astron. J., 97, 10. Stocke, J. T., Liebert J., Gioia, I. M., Griffiths, R., Maccacaro, T., Danziger, I., Kunth, D. and Lub, J. 1983, Ap. J., 273, 458 (EMSS). Stocke, J. T., Morris, S. L., Fleming, T. A., Gioia, I. M., Maccacaro, T., Schild, R., Wolter,A. and H.J. Patrick 1991, Ap. J. Suppl., 76, 813. Sun, W.-H. and Malkan, M. A. 1989, Ap. J., 346, 68. Svensson, R. 1987, M.N.R.A.S., 227, 403. Svensson, R. 1990, in “Physical Processes in Hot Cosmic Plasmas”, ed. W. Brinckmann et al., Kluwer Academic, pg. 357. Tananbaum, H., Avni, Y., Branduardi, G., Elvis, M., Fabbiano, G., Feigelson, E., Giacconi, R., Henry. J. P., Pye, J. P., Soltan, A. and Zamorani, G. 1979, Ap. J. Lett., 234, L9. Tananbaum, H., Avni, Y., Green, R. F., Schmidt, M. and Zamorani, G. 1986, Ap. J., 305, 57. Turner, T. J. and Pounds, K. A. 1989, M.N.R.A.S., 240, 833. Ulvestad, J. S. and Wilson, A. S. 1984, Ap. J., 278, 544. Urry, C. M. 1988, in “Multiwavelength Astrophysics”, ed. F. A. Cordova, Cam­ bridge: Cambridge University Press, pg. 279. 192

Urry, C. M., Arnaud, K., Edelson, R. A., Kruper, J. S. and Mushotzky, R. F. 1989, in “Proc. 23rd ESLAB Symposium, Two Topics in X-ray Astronomy”, ed. J. Hunt and B. Battrick, 2: AGN and the X-ray Background, pg. 789. Urry, C. M., Kondo, Y., Hackney, K. R. H. and Hackney, R. L. 1988, in “A Decade of TJV Astronomy with the IUE Satellite ”, ESA SP-281, 2, pg. 293. Vagnetti, F., Giallongo, E. and Cavaliere, A. 1991, Ap. J., 368, 366. Voit, M. G. 1991, Ap. J., 379, 122. Walsh, D., Carswell, R. F. and Weymann, R. J. 1979, Nature, 279, 381. Wambsganss, J„ Paczyriski, B. and Schneider, P. 1990, Ap. J., 358, L33. Wandel, A. and Petrosian, V. 1988, Ap. J. Lett, 329, L ll. Wandel, A. and Urry, C. M. 1991, Ap. J., 367, 78. Ward, M. J. 1988, M.N.R.A.S., 231, IP . Ward, M. J. 1988, in “Active Galactic Nuclei, Proc. of a Conference held at the Georgia State University, Atlanta, Georgia.”, ed. H. R. Miller and P. J. Wiita, Berlin Heidelberg: Springer-Verlag, pg. 275. Ward, M., Elvis, M., Fabbiano, G., Carleton, N. P., Willner, S. P. and Lawrence, A. 1987, Ap. J., 315, 74. Wilkes, B.J. and Elvis, M. 1987, Ap. J., 323, 423. Wilkes, B.J., Elvis, M. and McHardy, I. 1987, Ap. J., 321, L23. Wilkes, B. J., Masnou, J-L., Elvis, M., McDowell, J. and Arnaud, K. 1989, in “Proc. 23rd ESLAB Symposium, Two Topics in X-ray Astronomy”, ed. J. Hunt and B. Battrick, ESA SP-296, pg. 1081. Willacy, K. 1990, in “Ph.D. Thesis”, University of Manchester. Wills, B. J., Netzer, H. and Wills, D. 1985, Ap. J., 288, 94. Wills, B. J. and Browne, I. W. A. 1986, Ap. J., 302, 56. Wills. B. J. and Wills, D. 1980, Ap. J., 238, 1. Wisniewski, W. Z., Sitko, M. L. and Sitko, A. K. 1986, M.N.R.A.S., 219, 299. Wolstencroft, R. D., Gilmore, G. and Williams, P. M. 1982, M.N.R.A.S., 201, 479. Worrall, D. M. 1987, Ap. J., 318, 188. Zamorani. G. et al. 1981, Ap. J., 245, 357. Zdziarski, A. A. 1986, Ap. J., 305, 45. Zdziarski, A. A. 1991, in “in Relativistic Hadrons in Cosmic Compact Objects”, ed. A. A. Zdziarski and M. Sikora, Heidelberg: Springer, pg. 27. Zdziarski, A. A. 1992, in “Beams and Jets in Extragalactic Radio Sources, (Conf. Proceedings)”, (in press). Zdziarski, A. A. and Coppi, P. S. 1991, Ap. J., 376, 480. Zdziarski, A. A., Ghisellini, G., George, I. M., Svensson, R., Fabian, A. C. and Done, C. 1990, Ap. J. Lett, 363, LI. Zheng, W. 1991, Ap. J. Lett, 382, L55. Zheng, W. and O’Brien, P. T. 1990, Ap. J., 353, 433. 194 195 Acknowledgements

I would like to thank the following (in alphabetical order!) for their contributions to this thesis: Graziella Branduardi-Raymont and Phil Smith acquired and reduced the 21 cm Nh i measurements at Jodrell Bank. Also to Graziella for her support throughout my studentship. France Cordova and John Kartje provided the IPC data. Steve Howell measured CCD magnitudes for some of the AGN and provided the quotes from a certain US Vice-President. Chris Jomaron helped with the reduction of the INT CCD images. Mark Jones provided maps and measurements of the 100/zm cirrus for all of the USS AGN. Paul Mur din took the INT CCD images Richard Mushotzky pondered over the optical results and made some useful sug­ gestions Gail Reichert took an IUE spectrum of E0132—411 at NASA/GSFC, she has pa­ tiently answered all my transatlantic questions, looked after me during my visits to the US and is responsible for my large biceps. Jane Salton made the all-sky N# map. The optical ID observing team, including Jeremy Allington-Smith, Graziella, Phil Charles, Keith Mason, Jon Mittaz, Koji Mukai, Paul Murdin and Alan Smale, who took the optical spectra. The Science and Engineering Research Council provided the financial support for my studentship.

And of course, a great debt of thanks I owe to my supervisor, Keith Mason. He has been the driving force throughout, arranging for all of the observational data to be taken, stimulating and suggesting many ideas, and providing me with a demanding, fascinating and extensive project. This thesis, such as it is, in style and content, would have been a poor offering without him. He also tells me that the going rate for saying nice things about your supervisor is still only £10.

And finally, to all of my family, especially my parents Maggie and Derek Moore, who have been absolute bricks! iyo

The final word...

“Target prices? How that works? I know quite a bit about farm policy. I come from Indiana, which is a farm state. Deficiency pay­ ments - which are the key - that is what gets money into the farmer’s hands. We got loan, uh, rates, we got target, uh, prices, uh, I have worked very closely with my senior colleague, (Indiana Sen.) Richard Lugar, making sure that the farmers of Indiana are taken care of. ”

— DAN QUAYLE C orrigend a

pg. 38 In terms of Cl and C2, the expression ‘(Rl+1<7)<0.6’ is equivalent to:

^2 - aci/C2 > 1-6

where C l I ( O d \ 2 (OC2\2 (TC1/C2 - C2 x Y y C l J + v C2 / pg. 109 In Figure 6.4, the fluxes have not been corrected for the (1+z) loss in energy. This should be taken into account if these fluxes are used to calculate the rest-frame luminosity of the AGN. Applying this correction results in only an offset to the flux (which is constant with frequency) and does not affect the results of the fitting procedure, which involved measuring the slope of the various continua only and not absolute fluxes. Publications by the author Relevant to this thesis: Puchnarewicz, E. M., Mason, K. O., Cordova, F. A., Kartje, J., Brand­ uardi-Raymont, G., Mittaz, J. P. D., Murdin, P. G., and Allington-Smith, J. 1992, Optical Properties of Active Galaxies with Ultra-Soft X-ray Spectra, M.N.R.A.S., 256, 589. Puchnarewicz, E. M., Mason, K. O., Cordova, F. A., and Kartje, J. 1992, Optical Properties of Ultra-Soft X-ray AGN, in “Testing the AGN Paradigm, Proceedings of the 2nd Maryland Conference”, ed. S. S. Holt, S. G. Neff and C. M. Urry, American Institute of Physics, New York, pg. 204. Cordova, F. A., Kartje, J., Thompson, R. J., Mason, K. O., Puchnarewicz, E. M., and Harnden, F. R. 1992, Ultra-Soft X-ray AGN, Ap. J. Suppl., (in press). Others: Puchnarewicz, E. M., Mason, K. O., Murdin, P. G., and Wickramasinghe, D. T. 1990, Low State Spectrocopy of V834 Cen (E1405-451), M.N.R.A.S., 244, 20P. Howell, S. B., Szkody, P., Kreidl, T. J., Mason, K. O., and Puchnarewicz, E. M. 1990, CCD Time Resolved Photometry of Faint Cataclysmic Variables III, P.A.S.P., 102, 758. Reichert, G. A., Branduardi-Raymont, G., Filippenko, A. V., Mason, K. O., Puchnarewicz, E. M., and Wu, C.-C. 1992, Spatially resolved Ultraviolet spectroscopy of the LINER Galaxy NGC 3998, Ap. J., 387, 536. Branduardi-Raymont, G., Mittaz, J. P. D., Mason, K. O., Puchnarewicz, E. M., Murdin, P. G., Allington-Smith, J. R., Giommi, P., Tagliaferri, G., and Angelini, L. 1990, The Optical Characteristics of a Sample of Soft X-ray Selected Active Galctic Nuclei, in “Proceedings of 23rd• ESLAB Symposium”, ESA, pg. 713. Reichert, G. A., Puchnarewicz, E. M., and Mason, K. O. 1991, IUE and Einstein Observations of the LINER galaxy NGC 4579, Proc. Int. Symp. cEvolution in Astrophysics’, Toulouse, France, ESA SP-310, 535. Reichert, G. A., Puchnarewicz, E. M., Filippenko, A. V., Mason, K. O., Bran­ duardi-Raymont, G., and Wu, C-C. 1991, in “Relationships Between Active Galactic Nuclei and Starburst Galaxies”, ed. Filippenko, A., V., San Fran­ cisco: Astronomical Society of the Pacific(in press).