Multi-Wavelength Study of Narrow-Line Seyfert 1 Galaxies
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the
Graduate School of The Ohio State University
By
Patrizia Romano
*****
The Ohio State University
2002
Dissertation Committee: Approved by
Professor Bradley M. Peterson, Adviser
Professor Richard W. Pogge Advisor Astronomy Graduate Program Dr. Smita Mathur
Dr.T.JaneTurner
UMI Number: 3076737
______UMI Microform 3076737 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ______
ProQuest Information and Learning Company 300 North Zeeb Road PO Box 1346 Ann Arbor, MI 48106-1346 ABSTRACT
We study the multiwavelength properties of Ark 564 and Ton S180, two
NLS1s that have been the object of extensive multiwavelength spectroscopic campaigns undertaken during the last four years to test the current hypothesis that Narrow-Line Seyfert 1 galaxies (NLS1s) have relatively lower black hole masses and higher accretion rates than normal broad-line Seyfert 1 galaxies
(BLS1s) of the same luminosity.
We present an analysis of the X-ray variability properties of Ark 564 and Ton S180, based on 35- and 12-day continuous observations with ASCA, respectively. The mean spectra are characterized by a very steep power-law continuum (Γ = 2.54 and 2.44, respectively), a strong soft excess at energies
< 2keV,andanFeKα line that has a large equivalent width and originates in ionized material. Their spectral energy distributions peak at significantly higher energies in these objects than in BLS1s (15–100 eV for Ton S180 and ∼ 50 eV in
Ark 564). We also find that, based on available UV, FUV and X-ray data on absorption lines, the UV and X-ray absorbers in Ark 564 are physically related, and possibly identical.
ii The multiwavelength properties of these two NLS1s are consistent with
the predictions of the prevailing model for the structure of NLS1s, i.e. of AGNs
with high accretion rate (M˙ ≈ 0.1–0.9) onto a relatively low-mass black hole
6−7 (M ≈ 10 M ).
iii To my parents and No’, who taught me to follow my dreams,
and to Stefano, who helped me realize them.
iv ACKNOWLEDGMENTS
I wish to thank my advisors, B.M. Peterson, R.W. Pogge, S. Mathur and
T.J. Turner, for teaching me and supporting me–each in their unique way–and
Anil Pradhan and Don Terndrup, for their encouragement along the years. Their help took different forms that ranged from sound scientific advice to reality check, from proof-reading a manuscript 10 times to fighting software, from a comforting pat on the back to a figurative smack on the side of the head, from supplying sugars to limiting caffeine intake.
Cheers go to my Tae Kwon Do gals. Lisa, my sah bum nim, who taught me about power; Deb, my comrade in arms, fellow dissertation struggler who made the time and who taught me about strength and persistence; Kath, tower of strength and true personal inspiration, who brings the grace of a dancer in martial arts; Melanie, my screaming buddy (and possibly the most energetic and ethusiastic woman to walk on this planet), who taught me to “just breathe”;
Glenna, strength in motion, who taught me about momentum–and the value of a well-placed kick. And all the others who were there for me through thick and thin. They kept me sane, gave me appropriate targets to hit, quiet places to be, and provided a friendly environment where I could let my hair down. Or up.
v They fasted with me, ate with me–they even made sure I was awake in time for my defense. Special tuna-scented thanks go to Oscar and Felix, rotisserie and head-butt furry friends.
Last but not least, I wish to thank my parents and my grandmother for supporting me through these long years away from home, and for encouraging me to follow my dreams–even when they involved sciatica and broken bones. Federica and Francesca, my oldtime friends and supporters, out of sight, but not far from my heart. And Stefano, whom I cherish as a brother, a friend, and a lifemate, for his waking me up in the morning and helping me fall asleep at night, and for being there, parted from me and never parted.
vi VITA
August 21, 1967 ...... Born– Venice,Italy
1993 ...... LaureainAstronomy,UniversityofPadova
1994 ...... SummerReaearchProgram,Netherlands Foundation for Research in Astronomy
1994 – 1995 ...... VisitingScholar,PhysicsDepartment, University of Ljubljana, Slovenia
1995 – 1996 ...... VisitingScholar,Department of Physics and Astronomy, University of Alabama
1996 – 1997 ...... GraduateFellow,TheOhioStateUniversity
1997 – 2002 ...... GraduateTeachingandResearch Associate, The Ohio State University
1999 ...... M.S.inAstronomy,TheOhioStateUniversity
PUBLICATIONS
Research Publications
1. P. Romano, T. Zwitter, M. Calvani, and J. Sulentic, “On the wings of broad H-alpha emission in active galactic nuclei”, Monthly Notices of the Royal Astronomical Society, 279, 165, (1996).
2. P. Romano, P. Marziani, and D. Dultzin-Hacyan, “Balmer Line Variations in
vii the Radio-Loud AGN PG 1512+370”, The Astrophysical Journal, 495, 222, (1998).
3. M. Bautista, M., P. Romano, and A.K. Pradhan, “Resonance Averaged Photoionization Cross Sections for Astrophysical Models”, The Astrophysical Journal Supplements, 118, 259, (1998).
4. B.M. Peterson, et al., “X-Ray and Optical Variability in NGC 4051 and the Nature of Narrow-Line Seyfert 1 Galaxies”, The Astrophysical Journal, 542, 161, (2001).
5. T.J. Turner, P. Romano, I.M. George, R. Edelson, S.J. Collier, S. Mathur, and B.M. Peterson, “Multiwavelength Monitoring of the Narrow-Line Seyfert 1 Galaxy Ark 564. I. ASCA Observations and the Variability of the X-Ray Spectral Components”, The Astrophysical Journal, 561, 131, (2001).
6. S. Collier, et al., “Multiwavelength Monitoring of the Narrow-Line Seyfert 1 Galaxy Ark 564. II. Ultraviolet Continuum and Emission-line Variabil- ity”, The Astrophysical Journal, 561, 146, (2001).
7. O. Shemmer, P. Romano, et al., “Multiwavelength Monitoring of the Narrow-Line Seyfert 1 Galaxy Ark 564. III. Optical Observations and the UV–Optical–X-Ray Connection”, The Astrophysical Journal, 561, 162, (2001).
8. P. Romano, T.J. Turner, S. Mathur, and I.M. George, “A 12-day ASCA Observation of the Narrow-Line Seyfert 1 Galaxy Ton S180: Time-Selected Spectroscopy”, The Astrophysical Journal, 564 , 162, (2002).
9. D.M. Crenshaw, et al., “Reddening, Emission-Line, and Intrinsic Ab- sorption Properties in the Narrow-Line Seyfert 1 Galaxy Ark 564”, The Astrophysical Journal, 566, 187, (2002).
10. T.J. Turner, P. Romano, et al., “The Spectral Energy Distribution of the Seyfert Galaxy Ton S180”, The Astrophysical Journal, 568, 120, (2002).
viii 11. P. Romano, S. Mathur, R.W. Pogge, B.M. Peterson, and J. Kuraszkiewicz, “FUSE Observations of the Narrow-Line Seyfert 1 Galaxy Arakelian 564”, The Astrophysical Journal, in press, (2002).
FIELDS OF STUDY
Major Field: Astronomy
ix Table of Contents
Abstract...... ii
Dedication...... iv
Acknowledgments...... v
Vita...... vii
ListofTables...... xvi
ListofFigures...... xviii
ListofAbbreviations...... xxiii
1 Introduction 1
1.1 Narrow-Line Seyfert 1 Galaxies, a Test Case for AGN Models . . . 4
1.2TheX-ray/UVWarmAbsorber...... 8
x 1.3SpectralEnergyDistributionofNLS1s...... 9
1.4TheFocusofThisStudy...... 11
1.5PublishedWork...... 14
2 Multiwavelength Monitoring of the Narrow-Line Seyfert 1 Galaxy
Ark 564. ASCA Observations and the Variability of the X-ray
Spectral Components 16
2.1Introduction...... 16
2.2ObservationsandDataReduction ...... 18
2.3 The Time Variability ...... 20
2.3.1 Fractional Variability Amplitude ...... 22
2.4TheMeanSpectrum...... 24
2.4.1 TheSoftComponent...... 26
2.4.2 The Fe Kα Regime...... 27
2.4.3 X-rayandUVAbsorption...... 31
2.5 Spectral Variability ...... 31
2.5.1 MethodandSelectionDetails...... 31
xi 2.5.2 Variability of the Continuum ...... 33
2.5.3 Variability of the Soft X-ray Hump ...... 34
2.5.4 Variability of the Fe Emission Line ...... 37
2.6RMSSpectra...... 40
2.7Cross-correlationResults ...... 41
2.8SummaryofObservationalResults...... 44
2.9DiscussionandConclusions...... 46
3 A 12–day ASCA Observation of the Narrow-Line Seyfert 1 Galaxy
Ton S180: Time-Selected Spectroscopy 67
3.1Introduction...... 67
3.2ObservationsandDataReduction ...... 68
3.3 The Time Variability ...... 70
3.3.1 Fractional Variability Amplitude ...... 73
3.4TheMeanSpectrum...... 74
3.4.1 TheSoftComponent...... 76
xii 3.4.2 The Fe Kα Regime...... 78
3.5 Spectral Variability ...... 80
3.5.1 MethodandSelectionDetails...... 80
3.5.2 Variability of the Continuum ...... 81
3.5.3 Variability of the Soft X-ray Hump ...... 82
3.5.4 Variability of the Fe Emission Line ...... 84
3.5.5 RMSSpectra...... 87
3.6SummaryofObservationalResults...... 88
3.7ComparisonwithArk564...... 89
3.8DiscussionandConclusions...... 90
4 FUSE Observations of Arakelian 564: Properties of the Warm
UV–X-ray Absorber 108
4.1Introduction...... 108
4.2ObservationsandDataReduction...... 109
4.3Dataanalysis...... 111
xiii 4.3.1 Intrinsic O VI EmissionModels...... 113
4.3.2 Intrinsic O VI AbsorptionMeasurements...... 115
4.3.3 C III columndensity...... 118
4.3.4 VelocityCentroids...... 119
4.4PhotoionizationModeling...... 119
4.4.1 Input continua ...... 120
4.4.2 Physical conditions of the UV/X-ray absorber ...... 123
4.5Discussion...... 124
4.6Summary...... 127
5 The Spectral Energy Distribution of the Seyfert Galaxy Ton S180
141
5.1Introduction...... 141
5.2ASCA...... 143
5.3Chandra...... 144
5.4 EUVE ...... 146
xiv 5.5FUSE...... 146
5.6HST...... 148
5.7 Ground-based Data ...... 150
5.8ExaminationoftheSED...... 151
5.9Discussion...... 153
5.9.1 InterbandIndices...... 153
5.9.2 TheFormoftheSED...... 154
5.9.3 TheEnergyBudgetofTonS180...... 159
5.9.4 TheStateoftheCircumnuclearGas...... 160
5.10Summary...... 164
6 Quasi-simultaneous Spectral Energy Distribution of the Narrow-
Line Seyfert 1 Galaxy Arakelian 564 180
6.1Introduction...... 180
6.2Observations...... 181
6.2.1 DataReduction...... 181
xv 6.2.2 Summary of results from the multi-waveband observations . 184
6.2.3 ReddeningCorrection...... 186
6.3TheSEDofArk564...... 187
6.4Discussion...... 188
6.5FutureWork...... 191
7 Conclusion 200
7.1Summary...... 200
7.2FutureWork...... 203
Bibliography 206
xvi List of Tables
2.1 X-Ray and UV Cross-Correlation Results for Ark 564 ...... 52
3.1 Ton S180 Spectral Fits in the 2–10 keV Band ...... 96
3.2 Comparison of the Properties of Ton S180 and Ark 564 ...... 97
3.2Continued...... 98
4.1 O VI Model Emission-Line Parameters in Ark 564 ...... 129
4.2 O VI Column Densities from Intrinsic Absorption ...... 130
4.3 Column Densities from Intrinsic Absorption ...... 131
5.1 Observing Log for Ton S180...... 165
5.2UVAbsorptionLines...... 166
5.3STISEmissionLines...... 167
xvii 5.4 uvby Photometry...... 168
5.5 J, H and Ks Photometry...... 168
5.6DatafromtheSED...... 169
5.7SpectralIndices...... 170
5.8Luminosities...... 171
5.9Characteristicdatafor2BLS1s...... 171
6.1ObservingLogforArakelian564...... 193
6.2DatafromtheSEDs...... 194
6.3Luminosities...... 195
xviii List of Figures
1.1 A schematic model for active galactic nuclei ...... 3
2.1 Light curves for the Ark 564 ASCA data...... 53
2.2 Data/Model Ratio where the model is a simple power law fit . . . . 54
2.3 Data/Model Ratio where the model is a power law plus soft hump
plus Kerr model for the Fe Kα line...... 55
2.4SISLightcurves...... 56
2.5Spectralandtimingparameters...... 57
2.6 Ratio plots obtained by fitting the best-fit model for the first
spectrumtothefollowing39spectra...... 58
2.7 The ∆χ2 =2.3, 4.61, 9.21 contour levels for the soft hump
normalizationvs.photonindex...... 59
xix 2.8SoftHumpstrengthversusPhotonIndex...... 60
2.9 Fe Kα profileinhighandlowstate...... 61
2.10 Contour levels for Fe K-shell line intensity vs. photon index in high
andlowstate...... 62
2.11 Time Variable Fe Kα profiles...... 63
2.12 Contour levels for Fe K shell line intensity vs. photon index in
intervals1and4 ...... 64
2.13MeanandRMSspectrum...... 65
2.14 Light curves and cross-correlation functions ...... 66
3.1 Light curves for the Ton S180 ASCA data...... 99
3.2 Same as Figure 3.1, for two “events” at JD ≈ 2451523 ...... 100
3.3Spectralandtimingparameters...... 101
3.4 Data/Model Ratio where the model is a simple power law fit . . . . 102
3.5SISLightcurves...... 103
3.6 Ratio plots obtained by fitting the best-fit model for the first
spectrumtothefollowing13spectra...... 104
xx 3.7 Soft hump normalization vs. photon index in high and low state . . 105
3.8 Fe K-shell line intensity vs. photon index and line profiles in high
vs.lowstate...... 106
3.9 The ratio of RMS/Mean spectrum for SIS1 and SIS0 ...... 107
4.1 FUSE spectrumofArk564...... 132
4.2 Full-resolution (0.07 A,˚ ∼ 10 pixels) spectrum of Ark 564, in the
Lyβ/O VI wavelengthregion...... 133
4.3 Normalized line profiles of the absorption system at za ≈ ze. ....134
4.4 Normalized line profile, covering factor, and optical depth as a
functionofradialvelocity...... 135
4.5 Comparison of the adopted SEDs for Ark 564 ...... 136
4.6 Photoionization curves at constant ionic column density on the
log U–log NH plane...... 137
4.7 Same as Fig. 4.6, but with a ionizing continuum with no intrinsic
reddening...... 138
4.8 Same as Fig. 4.6, but with a ionizing continuum with Galactic and
intrinsicreddening...... 139
xxi 4.9 Same as Fig. 4.6, but with a ionizing continuum described in Laor
et al. (1997a) and Zheng et al. (1997) ...... 140
5.1 The 12-day ASCA lightcurveofTonS180...... 172
5.2 ±1st order Chandra/LETGlightcurve...... 173
5.3 LETG and ASCA data/model ratio compared to the α =1.44
power-law...... 174
5.4 FUSE spectrum...... 175
5.5 HST STIS (1150–3150 A)data...... 176˚
5.6Themulti-wavebanddataforTonS180...... 177
5.7SpectralEnergyDistributionofTonS180...... 178
5.8ComparisonofSEDs...... 179
6.1 Quasi-simultaneous SED of Arakelian 564 ...... 196
6.2 FUV-Optical Observed Spectrum of Arakelian64 ...... 197
6.3 Reddening corrected, rest-frame SEDs of Ark 564 ...... 198
6.4 Comparison of BLS1, Sy1, and NLS1 SEDs ...... 199
xxii List of Abbreviations
AGN Active Galactic Nuclei
ASCA Advanced Satellite for Cosmology and Astrophysics
BBB big blue bump
BLR broad-line region
BELR broad emission-line region
BLS1 broad-line Seyfert 1 galaxies
ESO European Southern Observatory
EUVE Extreme Ultraviolet Explorer
EW equivalent width
FIR far infrared
FUSE Far Ultraviolet Spectroscopic Explorer
FUV far ultra-violet
xxiii FWHM full-width at half maximum
HST Hubble Space Telescope
IR infrared
IRAS Infrared Astronomical Satellite
IRAF Image Reduction and Analysis Facility
JD Julian Date
NED NASA/IPAC Extragalactic Database
NLR narrow-line region
NLS1 narrow-line Seyfert 1 galaxies
QSO quasi-stellar object
RL / RLQ radio loud / quasar
ROSAT R¨ontgen Satellite
RQ / RQQ radio quiet / quasar
RXTE Rossi X-ray Timing Explorer
SAA South Atlantic anomaly
SED spectral energy distribution
S/N signal-to-noise ratio
UT Universal Time
UV ultra-violet
XMM X-ray Multi-Mirror satellite
xxiv Chapter 1
Introduction
Active galactic nuclei (AGN) are bright sources at the centers of galaxies that emit an amount of energy so large that they outshine (by up to a factor of
100) the light produced by their entire host galaxy of billions of stars, effectively making them among the brightest objects in the Universe. The breadth of their electromagnetic spectrum is also extraordinary, as they show strong continuum emission from γ-rays to radio wavelengths, and conspicuous emission and absorption lines. These objects are fascinating because they produce very high luminosities in a small region comparable in size to our solar system, through physical processes even more efficient than the nuclear fusion that powers stars.
The power source of AGNs is thought to be accretion onto a supermassive black hole. In this “black hole paradigm”, the basic structure of AGN can be summarized by the cartoon shown in Figure 1.1. A black hole with a mass of
6 9 10 –10 M (where M is the mass of the Sun) resides at the center of each AGN.
1 The matter in the vicinity of the black hole responds to the intense gravitational
field by forming an accretion disk and slowly spiralling into the black hole. The
accretion disk emits a continuum spectrum generally peaking in the optical
and ultraviolet (UV). A hot corona that surrounds the accretion disk produces
ahard(hν =2–10 keV) X-ray power-law continuum (photon flux at energy E
−Γ is PE ∝ E , where Γ is the photon index). Broad UV and optical emission
lines with characteristic full widths at half maximum (FWHM) in excess of a
few thousand km s−1 are emitted by the broad-line region (BLR, black “blobs”
in Figure 1.1). The large FWHM of the broad lines is thought to be due to
large-scale motions in the gravitational potential of the black hole. A thick, dusty
torus obscures the BLR from some lines of sight, causing AGNs to have different
observational properties depending on orientation. The clouds in the narrow-line
region (NLR, grey “blobs” in Figure 1.1), that lie well outside the obscuring
torus, also emit UV and optical lines whose widths are narrow (FWHM of a few
hundred km s−1). Finally, jets of relativistic material may be ejected from the
innermost regions and extend to distances up to 10 times larger than the size of
the host galaxy itself. The jets are most prominent at radio wavelengths where
they emit primarily synchrotron radiation. X-ray and UV absorption lines also
2 Fig. 1.1.— A schematic model for AGNs. Adapted from Urry & Padovani (1995).
indicate the presence of another region, the X/UV warm absorber, at a distance from the central black hole comparable to the BLR radius.
It is believed that, despite the apparent differences in the observational properties of AGNs, a few basic physical properties govern the processes that power AGNs, i.e., the mass of the black hole M, the accretion rate M˙ (the rate at which the matter falls into the black hole), and the properties of the host galaxy. It is also clear that in order to understand the structure of AGNs a multiwavelength study is necessary since different wavelengths probe different physical regions in AGNs.
3 1.1. Narrow-Line Seyfert 1 Galaxies, a Test Case for
AGN Models
Narrow-Line Seyfert 1 galaxies (NLS1s) are a subclass AGN with relatively narrow permitted optical emission lines (Hβ FWHM< 2000 km s−1)anda high Fe II/Hβ ratio (Osterbrock & Pogge 1985; Goodrich 1989). Historically, the ratio [O III]λ5007 / Hβ<3 has also been used as a defining criterion for the NLS1 objects, although it now appears that this is not a necessary one
(Rodriguez-Ardila et al. 2000). Such properties place these galaxies at the lower end of the line-width distribution for the Seyfert 1 population, thus distinguishing them from the bulk of Seyfert 1 galaxies (hereafter “broad-line Seyfert 1s” or
BLS1s). Furthermore, Boroson & Green (1992) found that NLS1s lie at one extreme end of the primary eigenvector of AGN properties, and conclude that the dominant source of differences in the observed optical properties of low-redshift
AGN is a fundamental physical parameter, such as M or M˙ .
In the X-ray regime, many NLS1s exhibit rapid and large-amplitude variability (Boller, Brandt, & Fink 1996, hereafter BBF96). Turner et al. (1999c) showed that X-ray variability is systematically more extreme in NLS1s than in BLS1s despite both having the same luminosity distribution. For a given luminosity, NLS1s show an order of magnitude greater variability (Turner et al.
1999c; Leighly 1999a). Some NLS1s show X-ray flares up to a factor of 100, on
4 timescales of days (e.g., Forster & Halpern 1996; Boller, Brandt, & Fabian 1997),
compared to the factors of a few seen in BLS1s. ROSAT 1 observations revealed
that the soft X-ray continuum slopes of NLS1s are systematically steeper than
−Γ those of BLS1s (BBF96), the photon index Γ (photon flux PE ∝ E ) sometimes
exceeding 3. This phenomenon was found to extend to higher energies using
ASCA2 observations (Brandt, Mathur, & Elvis 1997; Turner, George, & Nandra
1998; Leighly 1999b; Vaughan et al. 1999b), since the mean photon index is
2.19 ± 0.10 in NLS1s and 1.78 ± 0.11 in BLS1s (Leighly 1999b). The very strong
anticorrelation between Hβ FWHM and both the X-ray spectral slope in Seyferts
(BBF96) and in quasars (Laor et al. 1997a) and intrinsic variability amplitude
(Turner et al. 1999c) suggests that the remarkable X-ray properties of NLS1s
represent an extreme of Seyfert activity, possibly due to an extreme value of a
fundamental physical parameter related to the accretion process.
One increasingly popular hypothesis to explain the differences in X-ray
properties across the Seyfert population is that NLS1s have relatively low masses
for the central black hole compared to BLS1s with similar luminosities. Smaller
black-hole masses naturally explain both the narrowness of the optical emission
lines, which are generated in gas that has relatively small Keplerian velocities,
1ROSAT was a joint Germany/US/UK X-ray (0.1–2.4 keV) satellite, launched on 1990 June
1 and operated until 1999 February 12. 2ASCA (formerly named Astro-D) was a joint Japan/US X-ray (2–10 keV) satellite, launched on 1993 February 20 and operated until 2001 March 2.
5 and the extreme X-ray variability, since the primary emission would originate in a smaller region around the central engine (e.g., Laor et al. 1997a). Given that NLS1s have comparable luminosity to that of the BLS1s, Pounds, Done, &
Osborne (1995) suggested that they must be emitting at higher fractions of their
3 ˙ ˙ Eddington luminosity , hence higher fractional accretion rates (m ˙ = M/MEdd) are also required.
The closer the luminosity is to the Eddington limit (and the lower the black-hole mass), the greater the fraction of the energy emitted by the accretion disk in the soft X-rays (Ross, Fabian, & Mineshige 1992). Thus NLS1s might be expected to show disk components which peak at higher energies than for
BLS1s. Pounds, Done, & Osborne (1995) and Maraschi & Haardt (1997) noted that soft photons from the disk may Compton-cool hard X-rays from the corona, and cause the steep observed photon indices. In the case of a high accretion rate, the surface of the disk is also expected to become highly ionized (Matt, Fabian,
& Ross 1993); evidence of reflection from such an ionized disk (from Fe Kα)is found in six NLS1s: Comastri et al. (1998), Turner, George, & Nandra (1998) for Ton S180; Vaughan et al. (1999a), Comastri et al. (2001), Turner, George,
& Netzer (1999b) for Ark 564; Ballantyne, Iwasawa, & Fabian (2001) for the
3 LEdd =4πcmp GM/σe, where G is the Gravitational constant, c is the speed of light, mp is the proton mass, and σe is the Thomson cross-section, is the maximum luminosity of a source powered by steady-state spherical accretion (Frank, King & Raine 1992).
6 previous sources plus Mrk 335, NGC 4051, PG 1244+026, and PKS 0558-504.
However, lines from ionized Fe are not unique to the NLS1 population, since they sometimes occur in BLS1s as well (e.g., Guainazzi et al. 1998).
Alternative explanations for the difference in observed parameters across the
Seyfert population include the possibility that NLS1s have broad-line regions more distant from the nucleus than BLS1s and hence smaller Keplerian line widths
(Guilbert, Fabian, & McCray 1983), or that NLS1s are observed preferentially close to face on. However, Boroson & Green (1992) and Kuraszkiewicz et al.
(2000) disfavor the low-inclination model, while Nandra et al. (1997b) show that the inner regions of BLS1s also appear to be observed close to face-on. The former possibility can be tested by measuring the size of the broad-line region using reverberation mapping.
Reverberation mapping uses measurements of the correlations between emission-line and continuum flux variations across a broad energy band to determine the size-scales of the emitting regions (Peterson 1993). Line profiles give the kinematics of the emitting gas, and combining the size and velocity information yields an estimate of the central mass of an AGN (Wandel, Peterson &
Malkan 1999). Results obtained to date have indicated that for at least two NLS1s
6 7 8 for which this can be done, MBH ≈ 10 M as opposed to MBH ≈ 10 − 10 M
7 found for BLS1s of comparable luminosity and size (Kaspi et al. 2000b; Peterson et al. 2000).
1.2. The X-ray/UV Warm Absorber
More than half of the Seyfert 1 population shows optical/UV intrinsic absorption associated with their active nucleus (Crenshaw et al. 1999 and references therein). The strong UV absorption lines of Lyα,CIV,NV (and less frequently Si IV and Mg II) are found to be blueshifted, or at rest, with respect to the narrow emission lines of the same species, indicating the presence of a net radial outflow of the absorbing gas. A similar fraction of AGNs also show an associated ionized (“warm”) X-ray absorber (George et al. 1998; Reynolds 1997)
2 that is characterized by high ionization, U =0.1–10 (U = Q/4 πr nH c,where
Q is the number of Hydrogen ionizing photons emitted per second) and high
21 23 −2 total Hydrogen column density, NH =10 –10 cm . The signature of the X-ray warm absorber is the presence of O VII and O VIII absorption edges. During the last decade, evidence has accumulated that indicates that the same gas is responsible for both the UV and X-ray absorption spectra in Seyfert 1s (Mathur
1994a; Mathur et al. 1994b; Mathur, Elvis, & Wilkes 1995; Crenshaw et al. 1999;
Kriss et al. 2000; Monier et al. 2001; Brotherton et al. 2002). Although it is not always possible to model the X-ray/UV absorber as a single zone (especially
8 when the complex UV absorption is resolved into multiple velocity components characterized by a large range of column densities and ionization), common characteristics of these absorbers have emerged: they are composed of high ionization, low density, high column density gas that is outflowing and is located in or outside the broad emission-line region (BELR). It is therefore worthwhile to investigate the nature of this nuclear component in AGN as it represents an outflow (or wind) that can carry away a significant amount of kinetic energy at a mass-loss rate comparable to the accretion rate required to fuel the AGN proper
(Mathur, Elvis, & Wilkes 1995).
1.3. Spectral Energy Distribution of NLS1s
There has been a long-standing suggestion that the intrinsic spectrum of an
AGN is dominated by a so-called “Big-Blue-Bump” (BBB) of continuum emission, that peaks in the EUV–soft X-ray region, perhaps arising in the accretion disk.
Unfortunately, this emission emerges in a part of the electromagnetic spectrum that is strongly absorbed by atomic Hydrogen, and so effectively unseen. However, there are indications of the appearance of the low-energy tail of this component in the UV spectra of Seyfert 1 galaxies (Shields 1978; Malkan & Sargent 1982). The peak energy of the disk emission is predicted to be dependent on the accretion rate (Matt, Fabian, & Ross 1993). Thus the spectral-energy-distribution (SED)
9 of an AGN provides crucial information about accretion rates and conditions
close to the disk. However, determination of the X-ray–UV continuum in AGN
has been extremely difficult because of the severe attenuation of photons of these
energies by even small amounts of Galactic interstellar gas along the line-of-sight.
Some indication of the strength of the unseen continuum has been inferred from
the strengths of emission lines such as He IIλ1640 (e.g. Mathews & Ferland 1987).
Zheng et al. (1997) have suggested the form of the unseen X-ray–UV spectrum is
−α fv ∝ v with α = 2 between the Lyman limit (at 912 A)˚ and ∼ 0.5 keV. Laor et al. (1997a) combine this with a mean soft X-ray spectrum, based upon ROSAT observations of quasars, to dispute the existence of a large X-ray–UV bump.
Telfer et al. (2002) extend the work of Zheng et al. (1997) by including more data in the extreme UV band; those authors find the data to be adequately represented by a simple power-law with α =1.76 between 500 and 1200 A.˚ Korista, Ferland,
& Baldwin (1997) have discussed the problem that extrapolating the known soft
X-ray spectrum of AGN, there appear to be too few 54.4 eV photons to account for the strength of the observed He II lines. They consider the possibility that
the broad-line clouds see a harder continuum than the observer does, or that the
X-ray–UV spectrum has a double-peaked shape.
Examination of the SED of a NLS1, and comparison with that obtained
for BLS1s will offer insight into the relative accretion rates across the Seyfert
10 population. Measuring the SED requires observations taken simultaneously over
a long wavelength baseline stretching from infrared to hard X-ray wavelengths.
1.4. The Focus of This Study
This dissertation presents an investigation of the nature of NLS1 galaxies,
whose extreme continuum and emission line properties make them ideal for
testing the models of AGNs. We specifically test the current hypothesis that
NLS1s have relatively lower black hole masses and higher accretion rates than
BLS1s of the same luminosity. The objects of this study are Arakelian 564
and Tonantzintla S180, two NLS1s that have been the object of extensive
multiwavelength spectroscopic campaigns during the last 4 years.
Arakelian 564 (Ark 564, IRAS 22403+2927, MGC +05-53-012) is a bright,
nearby Narrow-Line Seyfert 1 (NLS1) galaxy, with z =0.02467 and V =14.6
43 magnitudes (mag; de Vaucouleurs et al. 1991), and L2–10 keV =2.4 × 10 ergs
s−1 (Chapter 2, Turner et al. 2001a, hereafter Paper I), hence an ideal test case for accretion disk models of AGN. Ark 564 was the object of one of the most intensive broad-band reverberation mapping programs undertaken to date, which aimed to determine the nature of the relationship between X-ray and UV–optical continuum variations and thus obtain an estimate of the BLR size and virial mass of the central source. This campaign provided the longest baseline for a
11 quasi-continuous X-ray study of any AGN to date (i.e. interrupted only by earth
occultation, south Atlantic anomaly (SAA) passage etc). Ark 564 was observed by
ASCA (2000 June 1 to July 6, Paper I; Pounds et al. 2001; Edelson et al. 2002),
Hubble Space Telescope (HST 4, 2000 May 9 to July 8, Collier et al. 2001, Paper
II; Crenshaw et al. 2002, Paper IV) and from many ground-based observatories as
part of an International AGN Watch5 project (1998 Nov to 2001 Jan, Shemmer
et al. 2001, Paper III). Ark 564 has shown a strong associated UV absorber
(Crenshaw et al. 1999, Paper II, Paper IV). There are indications that it also
possesses a warm X-ray absorber, as seen by the absorption lines of O VII and
O VIII detected in a Chandra6 spectrum (Matsumoto, Leighly, & Marshall 2001)7.
Tonantzintla S180 (Ton S180, PHL 912) is a bright (MB = −23.1 mag)
NLS1, with redshift z=0.06198 (Wisotzki et al. 1995), and a low Galactic column
20 −2 density along the line-of-sight (NH =1.52 × 10 cm ; Stark et al. 1992). The
source is at the extreme end of the Seyfert range of line widths with FWHM Hα
and Hβ ∼ 900 km s−1, making it a good choice for isolating the fundamental
parameter which determines the classification of a Seyfert galaxy. We selected
Ton S180 for study as it is bright in the soft X-ray regime and has low line-of-sight
4HST is a joint European Space Agency (ESA) and the National Aeronautics and Space
Administration (NASA) 2.4-meter reflecting telescope that was deployed on 1990 April 25. 5All publicly available data and complete references to published AGN Watch papers can be
found at http://www.astronomy.ohio-state.edu/∼agnwatch. 6The Chandra X-ray Observatory is a NASA satellite launched on 1999 July 23. 7 http://www.pha.jhu.edu/groups/astro/workshop2001/papers/.
12 and intrinsic extinction, allowing a view of the bare continuum form. Ton S180
was observed by ASCA on 1999 December 3 to 15, during a multi-wavelength monitoring campaign that included observations from HST,theRossi X-ray
Timing Explorer8 (RXTE), Chandra,theExtreme Ultraviolet Explorer9 (EUVE),
the Far Ultraviolet Spectroscopic Explorer10 (FUSE), in addition to optical–IR
observations obtained from ground-based observatories.
In Chapter 2 and 3 we present an analysis of the X-ray variability properties
of the Narrow-Line Seyfert 1 galaxies Ark 564 and Ton S180, based on 35- and
12-day continuous observations with ASCA, respectively.
In Chapter 4 we present a 63 ks FUSE observation of Ark 564. We use the
column densities of O VI and C III in conjunction with the published column
densities of species observed in the UV and X-ray bands to derive constraints on
the physical parameters of the absorber through photoionization modeling.
In Chapter 5 and 6 we present the spectral energy distributions of Ton S180
and Ark 564 (the latter is preliminary), and discuss the main differences when
compared with the SEDs of BLS1s and radio-quiet AGNs.
8 RXTE is a NASA X-ray satellite which was launched in 1995. 9EUVE is an extreme UV NASA satellite that was launched on 1992 June 7 and operated
until 2001 January 31. 10FUSE is a FUV satellite launched on 1999 June 24.
13 Chapter 7 summarizes our results, draws our conclusions, and highlights the avenues of future research.
1.5. Published Work
Chapter 2 has been previously published as T.J. Turner, P. Romano, I.M.
George, R. Edelson, S.J. Collier, S. Mathur, and B.M. Peterson, “Multiwavelength
Monitoring of the Narrow-Line Seyfert 1 Galaxy Ark 564. I. ASCA Observations and the Variability of the X-Ray Spectral Components”, 2001, The Astrophysical
Journal, 561, 131
Chapter 3 has been previously published as P. Romano, T.J. Turner, S.
Mathur, and I.M. George, “A 12-day ASCA Observation of the Narrow-Line
Seyfert 1 Galaxy Ton S180: Time-Selected Spectroscopy”, 2002, The Astrophysical
Journal, 564 , 162
Chapter 4 will be published as P. Romano, S. Mathur, R.W. Pogge, B.M.
Peterson, and J. Kuraszkiewicz, “FUSE Observations of the Narrow-Line Seyfert
1 Galaxy Arakelian 564”, 2002, The Astrophysical Journal, in press
Chapter 5 has been previously published as T.J. Turner, P. Romano, et al.,
“The Spectral Energy Distribution of the Seyfert Galaxy Ton S180”, 2002, The
Astrophysical Journal, 568, 120
14 Reformatting for the dissertation included renumbering of tables and figures to ensure continuity, and moving bibliographic material to the general list of references, as well as shortening the introductions, to avoid unnecessary repetition.
15 Chapter 2
Multiwavelength Monitoring of the Narrow-Line Seyfert 1 Galaxy Ark 564. ASCA Observations and the Variability of the X-ray Spectral Components
2.1. Introduction
Ark 564 (IRAS 22403+2927, MGC +05-53-012) is a bright, nearby NLS1 with z =0.0247, (Huchra, Vogeley & Geller 1999); V =14.6 mag, (de Vaucouleurs
43 −1 et al. 1991), and L2–10 keV ≈ 2.4 × 10 ergs s (from these data); hence an ideal test case for accretion disk models of AGN. ROSAT observations (Brandt et al.
1994) showed a complex soft X-ray spectrum with spectral features around ∼ 1 keV. However, the PSPC data could not discriminate between absorption or emission as the origin of the complex form of the spectrum. Subsequent ASCA observations (∼ 50 ks, Turner, George, & Netzer 1999b; Vaughan et al. 1999a;
16 Leighly 1999b) confirmed the complexity of the soft spectrum and showed a strong
iron line at rest energy of about 7 keV, indicative of reflection from highly ionized
material. This conclusion was supported by Vaughan et al. (1999b) who detected
an edge at 8.5 keV in combined ASCA and RXTE data, attributable to reflection from a strongly irradiated disk. BeppoSAX 1 data (Comastri et al. 2001) are also
indicative of reflection from a highly ionized, optically-thick accretion disk.
Ark 564 was observed by ASCA on 2000 June 1 to July 6, during a multiwavelength monitoring campaign that included observations from HST between May 9 and July 8 (Collier et al. 2001, Paper II), RXTE between June 1 and July 1 (Pounds et al. 2001; Edelson et al. 2002), and the X-ray Multi Mirror
Telescope2 (XMM ) June 17, and from many ground-based observatories as part of the International AGN Watch project (Shemmer et al. 2001, Paper III).
In this Chapter we present the results from the ∼ 1MsASCAobservation of Ark 564. In §2.2 we present the data; in §2.3 we discuss the time variability of the source; §2.4 we analyze the mean spectrum; in §2.5 we discuss time-resolved spectroscopy; §2.6 presents an investigation of the RMS spectrum; §2.7 looks at the correlations between the X-ray parameters, and with the UV flux at 1365 A;˚
1 BeppoSAX was an Italian X-ray (0.1–300 keV) satellite that was operated between 1996
April 30 and 2002 April 30. 2XMM is an ESA X-ray satellite that was launched on 1999 December 10.
17 §2.8 presents a summary of our observational results; finally, in §2.9 we discuss the X-ray results.
2.2. Observations and Data Reduction
ASCA is equipped with four focal-plane detectors that are operated simultaneously, namely, two CCDs (the Solid-state Imaging Spectrometers
SIS-0 and SIS-1, 0.4–10 keV, Burke et al. 1991) and two gas-scintillation proportional-counters (Gas Imaging Spectrometers GIS-2 and GIS-3, 0.7–10 keV,
Ohashi et al. 1996, and references therein). ASCA obtained a total duration of ∼ 2.98 Ms on Ark 564, starting from JD = 2451697.024 (for the screened data). The data were reduced using standard techniques as described in Nandra et al. (1997a); in particular, the methods and screening criteria utilized by the
Tartarus3 database (Turner et al. 1999c) were used. Data screening yielded an effectiveexposuretimeof∼ 1.11 Ms for the SISs and ∼ 1.29 Ms for the GISs.
The mean SIS-0 count rate was 1.894 ±0.001 cts s−1 (0.6–10 keV band).
Since launch in 1993, the ASCA SIS detectors have been experiencing a
degradation in efficiency at the lower energies, which is probably due to increased
dark current levels and decreased charge transfer efficiency (CTE), producing
SIS spectra which diverge from each other and from the GIS data. This loss
3http://tartarus.gsfc.nasa.gov.
18 in efficiency is not well understood and therefore not corrected for by any of
the software (in particular, CORRECTRDD does not solve the problem), so that the discrepancy between SIS-1 and GISs can be as much as 40% below 0.6 keV for data taken in January 20004 Data from the last phase of ASCA operations,
AO-8, revealed that there has been a non-linear evolution of the SIS CTE. The
Ark 564 data were first calibrated using linear extrapolations of SIS gain from the
last determination of CTE values on 1997 March 11. That calibration produced
unacceptably large inconsistencies between the SIS and GIS instruments up to
a few keV. Thus the previous solution of excluding SIS data below 0.6 keV is
inadequate. We recalculated the SIS gain using the interim solution released on
2001 February 13 (using CTE file sisph2pi 130201.fits). This interim solution
reduced the instrument discrepancies significantly, but still necessitated the
exclusion of some sections of the SIS data. For this reason, we proceed as follows.
We have modeld the features which appear in both GIS and SIS detectors and
ignored residuals which are in one instrument or instrument pair only (and are
likely artifacts of calibration problems). We then used our simple parameterization
of the data to examine the time-variability of the source spectrum.
The divergence of the SIS detectors at low energies can be partly corrected
for in the spectral analysis. One approach (Yaqoob et al. 2000)5,istoquantify
4 http://heasarc.gsfc.nasa.gov/docs/asca/watchout.html. 5http://lheawww.gsfc.nasa.gov/∼yaqoob/ccd/nhparam.html.
19 the apparent loss in SIS efficiency as a function of time throughout the mission;
an empirical correction can be applied by parameterizing the efficiency loss with
a time-dependent absorption (“excess NH”). The correction for SIS-0 follows the
−8 7 20 linear relationship, NH(SIS0) = 3.635857508 × 10 ( T − 3.0174828 × 10 )10
cm−2,whereT is the average of start and stop times of the observation, based on seconds since launch. There is no simple analytical relationship that satisfies the SIS-1 excess NH, but it is usually found that a slightly larger column can be
applied to the SIS1 data to bring it into line with SIS0. Naturally, one has to be
extremely conservative in interpreting spectral features derived from fits in the
lower-energy region, in particular below 1 keV, once this empirical correction has
8 20 been applied. For our observation, where T =2.36 × 10 s, NH(SIS0) = 7.5 × 10
−2 21 −2 cm and we adopted NH(SIS1) = 1.05 × 10 cm .
A more conservative approach consists of only considering data above 1 keV
for both SIS and GIS, where the disagreement is less dramatic. However, as we
are very interested in the variability of the soft hump spectral component, in this
Chapter, we will use the excess NH correction for our fitting.
2.3. The Time Variability
Light curves were extracted using bin sizes of 256 s, and 5760 s in the
full-band (0.7-10 keV) for the SIS, a soft-band (0.7–1.3 keV) for the SIS data,
20 and the hard-band (2–10 keV) for both GIS and SIS data. The use of 0.7 keV as a lower limit for the SIS data was due to the high setting for the SIS lower level discriminator towards the end of the observation. In all cases we combined the data from the SIS and GIS detector pairs. The exposure requirements for the combined curves were that the bins be fully exposed in each instrument for the
256 s curves and at least 10 % exposed for the 5760s curves. The observed count rates correspond to a mean 2–10 keV flux of 2 × 10−11 erg cm−2 s−1 and 2–10 keV
43 −1 −1 −1 luminosity 2.4 × 10 ergs s (assuming H0 =75kms Mpc ,q0 =0.5). This mean flux level is ∼ 20 % brighter than that observed during a previous ASCA observation in 1996 December 23 (Turner, George, & Netzer 1999b). Figure 2.1 shows the combined 0.7–1.3 keV SIS soft-band and GIS hard-band light curves in
5760 s bins. The background level in the source cell is about 4 % of the source count rate, and not plotted or subtracted. Figure 2.1 also shows the softness ratio, defined as the ratio between the count rates in the 0.7–1.3 and 2–10 keV bands. The spectrum hardens during the ASCA observation.
The light curves binned to 5760 s show trough-to-peak variations in flux by a factor of ∼ 10 in the 0.7–1.3 keV band (SIS), ∼ 7 in the 2–10 keV band (GIS).
Examination of light curves binned to 256 s reveals even larger amplitudes due to fast flickering. The maximum amplitude of variability in these (256 s) curves is a factor ∼ 16 for the SIS data in the 0.7-1.3 keV band and ∼ 14 for the GIS data in the 2–10 keV band. In their Figure 3, Edelson et al. (2002) show the 0.7–1.3
21 keV and 2–10 keV light curves, in 256 s bins, for a “flare” at JD ≈ 2451710
(centered around 1153 ks after the start of the observation). The SIS and GIS data show rate variations of a factor of 1.4 and 1.8 (respectively) in ∆t = 1280 s, corresponding to a variation in luminosity of ∆L =1.9 × 1043 ergs s−1.The close-up examination of that flare shows that the hard X-ray event is sharper than that in the soft band, and that there is a significant change in softness ratio on a timescale of ∼ 1000 s (Edelson et al. 2002). Close examination of the light curve reveals other similar examples of rapid changes in spectral shape and also some flares which are not accompanied by a strong spectral change (Edelson et al. 2002). The behavior of the source is obviously complex with some events characteristically different to others.
Here we concentrate on the search for variations in spectral parameters across the 35 day baseline of the observation to understand the nature of the spectral variability.
2.3.1. Fractional Variability Amplitude
F σ The fractional variability amplitude var and its error Fvar are defined in
Edelson et al. (2002), as
S2 −σ2 S2 err 1 1 F ,σF . var = 2 var = 2 (2.1) X Fvar 2N X
22 S2 σ2 where is the total variance of the light curve, err is the mean squared error,
and X is the mean count rate.
The quantity Fvar was calculated approximately every day using even
sampling and constructing light curves with 256 s bins. We examined the quantity
in the soft (0.7–1.3 keV) band and the hard (2–10 keV) band. These bands were
chosen to be as widely separated as possible in the energy-bandpass, but having
good signal-to-noise (S/N) in each curve. Fvar is variable across the observation,
but there appears to be no correlation between this quantity and the flux-state of
the source, probably due to a random nature of the light curve. Fvar appears very
well correlated between the soft and hard bands. As we will demonstrate in §2.4,
in the mean spectrum the power-law continuum provides about 75 % of the flux in
the 0.7–1.3 keV band. The rapid variations in the power-law component dominate
the variability so the two bands have a correlated component of variability,
making Fvar appear similar, to first order, in the soft and hard bands.
Fvar calculated across the whole dataset is 34.84 % ± 0.46 for the 0.7–1.3
keV band, and 33.19 % ± 0.40 for the 2–10 keV band. This quantity measures
deviations compared to the mean, integrated over the entire month. The slightly
greater Fvar for the soft band is probably due to the relatively strong variation in the soft component occurring over a month timescale, as we will show in §2.5.3.
23 2.4. The Mean Spectrum
For the spectral analysis the source counts were grouped with a minimum of
20 counts per energy bin. After examining the spectral fits separately to quantify
the cross-calibration problems (as described in §2.2), the data were fit allowing
the relative normalization of the four instruments to be free to allow for small
differences in calibration of the absolute flux, and differences in the fraction of
encircled counts contained within the SIS and GIS extraction cells. The spectral
fits have been performed with the XSPEC V11.0.1 package, using the response
matrices released in 1997 for the GIS, and response files generated using HEAsoft
v5.0.4 for the SIS.
The spectral continuum slope was first determined by fitting a power-law
20 −2 model modified by Galactic absorption (NH =6.4×10 cm ,Dickey&Lockman
1990), and corrected for the SIS low-energy problem as described in §2.2. For this
fit we used data in the bandpass 2–5 plus 7.5–10 keV data in the rest-frame (∼
1.8–4.9 keV observer’s frame). Furthermore, we excluded SIS data in the 1.7–2.5 keV regime and above 7.32 keV (both observer’s frame) due to problems in the calibration that show up in data of such high S/N. The power-law fit yielded
Γ=2.538 ± 0.005 and χ2 = 1403 for 1225 degrees of freedom (dof). This and subsequent errors represent, unless otherwise specified, the 90 % confidence level.
24 The data/model ratio is shown in Figure 2.2, along with the good data overlaid relative to this continuum model, and also the bad (excluded) data, shown as a different point-style. A strong soft excess is evident which appears asahumpofemission,risingabovethepower-lawcontinuumbelow2keV,then
flattening off below 1 keV, as observed in an earlier ASCA observation (Turner,
George, & Netzer 1999b); hereafter we refer to this component as the soft hump.
It is also interesting to note that the shape of the soft hump is evident in the
PSPC spectrum. We constructed a data/model spectral ratio for the archival
ROSAT PSPC data from 1993. The data were compared to a power-law model of the mean slope noted above. The PSPC data between 1.5–2.0 keV were used to normalize this component (there being insufficient bandpass to determine the hard X-ray slope) and the rest of the spectral data were then overlaid. The ratio is shown in the inset panel of Figure 2.2, demonstrating that the soft X-ray spectrum turns up again between 0.4–0.5 keV. (Even if the power-law index is different at the epoch of the PSPC observation, we could not introduce the structure observed in the soft X-ray spectral shape.) Also evident in the main panel is an excess of emission close to 7 keV, which we know to be due to an unmodeled Fe Kα line. The overlay of the excluded bad points serves to show where the strongest problems are, due to the detector aging and calibration issues described in §2.2.
25 2.4.1. The Soft Component
The status of the calibration, the degradation of the SIS energy-resolution towards the end of the mission, and the small bandpass of data available to examine the soft hump component conspire to make it impossible to achieve an unambiguous parameterization of the hump shape using these data. This led us to use a very simple parameterization of the hump in order to simply examine its
flux variability.
A previous detailed study of Ark 564 (Turner, George, & Netzer 1999b) detected the soft hump and ruled out origins solely due to the effects of emission and/or absorption from photoionized gas. Models based upon emission from thermal gas were consistent with the spectral data, but posed problems in terms of physical consistency with the picture of an AGN (Turner, George, & Netzer
1999b). In light of the Chandra results for the NLS1s Ton S180 (Turner et al.
2001b) and NGC 4051 (Collinge et al. 2001) that the hump is apparently a smooth continuum component, a variability study such as this has turned out to be a better technique to understand the nature of the soft hump than high resolution spectroscopy, as shown below in §2.5.3.
The lack of constraints on the form of the soft hump lead us to use a simple
Gaussian parameterization of the component. This adequately models the shape and flux of the excess in the ASCA data, and allows us to study the variability
26 of the component using time-resolved spectroscopy. Using SIS data in the range
0.75–1.7 and 2.5–4.88 keV simultaneously with GIS data in the range 1.0–4.88
and 7.32–9.76 keV (observer’s frame), the Gaussian model which best fits the
excess has a peak energy E =0.57 ± 0.02 keV, width σ =0.36 ± 0.01 keV and
n . +0.12 × −2 −1 −2 normalization =125−0.17 10 photons s cm corresponding to a mean
+11 equivalent width (EW)= 110−15 eV.
2.4.2. The Fe Kα Regime
AsignificantFeKα emission line is evident in the ASCA spectrum
(Figure 2.2). The line profile is asymmetric with a marked red wing, as observed
in Seyfert 1 galaxies (Nandra et al. 1997b), but with a peak close to 7 keV.
These basic properties of the line will be robust to refinements to the instrument
calibrations.
We utilized SIS data in the range 2.5–7.32 keV simultaneously with the GIS
data in the range 1.8–9.76 keV. A narrow Gaussian component is an inadequate
model for the line profile, but allows us to determine that the line peak is at an
observed energy of 7.1 keV. The asymmetry of the line led us to fit the data with
a disk-line model profile (Fabian et al. 1989). The model assumes a Schwarzschild
geometry and we assumed an emissivity law r−q for the illumination pattern of
the accretion disk, where r is the radial distance from the black hole. We assume
27 the line originates between 6 and 1000 gravitational radii (Rg) and we constrained
the rest-energy of the line to lie between 6.4 and 7 keV (as this first test is for Fe
Kα and this range represents the rest-energies possible for this line, depending
on ionization-state of the reprocessing gas). The inclination of the system is
defined such that i = 0 is a disk oriented face-on to the observer. This model
gave χ2 = 6584 for 2481 dof for a fit including the full range of good data. The
E . +0.00p rest-energy of the line was =700−0.13 keV, i.e. the energy pegged at the upper
limit allowed in the fit. The inclination was i =26± 2 degrees, emissivity index
was q =5.7 ± 0.9 and normalization n =4.80 ± 0.50 × 10−5 photons s−1 cm−2.
+29 . +0.006 The equivalent width was EW= 351−37 eV. The index was Γ = 2 541−0.004 in
this model, i.e. consistent with that obtained by fitting for the continuum alone.
While the fit is statistically poor, the contributions to χ2 are dominated by a few
areas where the data from the four instruments diverge, due to problems with the
calibration for data from this epoch. No systematic residuals remain which are
present in all instruments.
Next we tested a model for the line profile assuming a Kerr metric, as
implemented by Laor (1991), for a maximally rotating black hole which will
have the most intense gravitational effects. In this case we fixed the innermost
radius as the last stable orbit for a Kerr hole, and the outer radius at the
maximum value allowed by the model, 400 Rg. The Kerr model provides a
fit-statistic χ2 = 6573 for 2481 dof, an improvement (at 95 % confidence) over
28 E . +0.01p the Schwarzschild model. The rest-energy of the line was =699−0.13 keV,
i +11 q . ± . inclination was =17−17 degrees, emissivity index was =325 0 14 and normalization n =8.04 ± 0.50 × 10−5 photons s−1 cm−2. The equivalent width
± . +0.019 was EW= 653 85 eV. The photon index was Γ = 2 583−0.003.
Analysis of previous short ASCA observations revealed some ambiguity as to whether the Fe Kα line arose from ionized material, or a disk highly inclined to the line-of-sight (Turner, George, & Nandra 1998), therefore we explicitly tested for an origin in neutral material. If the line energy is fixed at 6.4 keV, but all other
χ2 / dof parameters are allowed to vary, then the fit-statistics are Schwarz = 6653 2482
χ2 / dof for the Schwarzschild model, and Kerr = 6607 2482 for the Kerr model.
These fits are significantly worse than those which show a line energy close to 7 keV, thus we conclude that the Fe Kα (Lyα) emission in Ark 564 does originate in highly ionized material, dominated by emission from Fexxvi (H-like) ions.
The line energies derived are indicative that H-like Fe dominates more than
He-like Fe. However, we consider this to be an area which requires revisiting with the forthcoming improvements to GIS and SIS calibration, and drawing
firm conclusions on the relative importance of He-like and H-like ions would be premature with the calibration used here.
The residuals to the Kerr fit are shown in Figure 2.3. There is a significant excess of emission suggestive of an additional emission line just above 8 keV
29 (this shows up in both the GISs whose data we used, and examination of the
disguarded SIS data in this range also revealed the line component). Using a
E . +0.10 narrow Gaussian model to fit this excess revealed a line energy =815−0.12 keV,
+54 with equivalent width 63−38 eV. The addition of this line to the model improves
the fit by ∆χ2 = 41. This excess is identified as a combination of the Kβ line of
Fexxvi (Lyβ)andtheKα line of Ni (Ni Lyα; see discussion). The addition of
this line-blend to our models does not change significantly the implied parameters
of the rest of the fit. This line blend has a width consistent with that of the Fe
Kα line, although line widths should be revisited with the final ASCA calibration.
Allowing an additional narrow line component within rest energy-range 6.4–7.0
keV did not improve upon the fit obtained with the Kerr model, ∆χ2 < 1. The 90
% upper limit on a narrow line with rest-energy between 6.4–7.0 keV is ∼ 80 eV.
In light of the evidence for ionized material, we also tested for the presence
of an Fe K edge. The fit did not improve on addition of this model component
(∆χ2 = 0). In order to make a direct comparison with Vaughan et al. (1999a) we
calculated the 90 % confidence upper limit on an edge at 8.76 keV (from He-like
Fe), which is τ<0.09. In light of the strong emission features it is difficult to understand the apparent absence of an Fe edge. We suspect the edge is present, and that the improved calibration of ASCA may later make it easier to precisely model this complex region of the spectrum.
30 2.4.3. X-ray and UV Absorption
The spectral-energy-distribution of Ark 564 is relatively depressed in the optical and UV regime leading Walter & Fink (1993) to suggest that the unusual
UV (λ1375) to 2 keV flux ratio in Ark 564 could be due to absorption of the
UV continuum. Crenshaw et al. (2002) use the observed Heii λ1640/4686 ratio and the continuum shape in the UV/optical regime to derive a reddening excess of E(B-V) = 0.17 mag (Galactic =0.03 mag, intrinsic=0.14 mag) as the total reddening for Ark 564. For a Galactic dust-to-gas ratio this implies a column
∼ 9 × 1020 cm−2 of absorption in the X-ray band. We found this column of gas to be consistent with the X-ray data, and the presence of some absorber would lead to attenuation in the soft X-ray regime, and could thus explain the significant curvature (flattening) of the soft hump component below 1 keV (§2.4.1). However, again we stress that the absolute form of the soft hump is not the primary goal of this paper, rather the insights obtained from the variability of the component.
2.5. Spectral Variability
2.5.1. Method and Selection Details
In order to construct a complete picture of the multi-waveband variability of an AGN, one must consider the variations of X-ray spectral parameters. Simple
31 flux-flux correlations can miss important clues to the emission and reprocessing
mechanisms at work. In order to study the spectral evolution of Ark 564, we
created 40 time-selected spectra across the 35 day observation. We sampled
throughout the light curve following flares and dips using Xselect V2.0.The resulting average baseline for each time-selected spectrum was ∼ 75 ks, and
the average on-source exposure time was ∼ 25 ks. (We note that the spectral
variations which we will demonstrate and discuss were also evident when the
data were sampled evenly, although such a sampling seemed to average out some
interesting fluctuations). Figure 2.4 shows our choice of intervals as vertical
dashed lines plotted over the combined SIS light curve. We set the background,
ancillary response, and response matrix files to be those of the mean spectrum,
as the background spectrum and flux did not vary significantly during the
observation, and we wished to attain the best possible S/N in the subtracted
data. We also grouped the source counts with a minimum of 20 counts per energy
bin as before. All fits were performed fixing the scaling factors for instrument
normalization among the 4 datasets (SIS and GIS) and the corrections for the
SIS low-energy problem (see §2.2) to the best-fit values from the mean spectrum.
We consistently modified our models with Galactic absorption by a column
20 −2 NH =6.4 × 10 cm . The time stamps of the light curves shown in this section
are in JD-2450000 and refer to the mid-point of the observation. Figure 2.5 shows
the results from this analysis, described in full below.
32 2.5.2. Variability of the Continuum
For each of the 40 time-selected spectra we fitted the SIS data in the
range 2.5–4.88 keV simultaneously with the GIS data in the ranges 1.8–4.88 and
7.32–9.76 keV (all observers frame) using a simple power-law model (the same
data exclusions as for the mean fit to the continuum slope, §2.4). Figure 2.5 shows the light curves for the (model) continuum flux and the best-fit photon index
Γ. Significant (see §2.7) but small variations are observed in Γ, which has a full range 2.45–2.72, (i.e. ∆Γ = 0.27) across the 35 days. Some significant changes are apparent down to timescales of ∼ 1 day (Figure 2.5). Fitting the photon indices to a constant model yields χ2 =67/39 dof. In addition to changes in slope, we
note that the power-law component dominates the 2–10 keV band, thus rapid flux
variations observed down to timescales of ∼ 1000 s in that band are attributable to flux changes in the power-law continuum.
As an illustrative way of examining these spectral variations we constructed a plot to highlight the different aspects of the spectral evolution. Figure 2.6 shows ratio plots obtained by comparing the best-fit model for the first spectrum to the following 39 spectra. No fitting was performed. The plot is constructed to illustrate the variations of the spectrum and flux compared to the first day of data. Strong variations in both flux and continuum slope are present, as well as a strongly varying soft hump. We note that the soft hump is always evident above
33 the power-law continuum, although it appears to change in absolute strength and
relative to the power law. The data show a hint of variations in the Fe Kα line, which we investigate in more detail in §2.5.4.
2.5.3. Variability of the Soft X-ray Hump
To examine the variability of the soft hump we utilized SIS data in the range
0.75–1.7 and 2.5–4.88 keV simultaneously with GIS data in the range 1.0–4.88
and 7.32–9.76 keV (observer’s frame). The lower limit of the SIS was based on
the level of agreement achieved between the two CCDs using our methods of
correction, as described in §2.2. The model was a simple power-law plus a broad
Gaussian for the soft hump, with Gaussian peak and width fixed at the values noted in §2.4.1, while normalization of the soft hump, plus that of the continuum power law and the power-law slope were all left free. To avoid an overly complex model (which can result in false or local minima being found) we excluded data in the Fe Kα regime (4.88-7.32 keV) for these fits. Figure 2.5 shows the time series for the normalization of the soft hump. The soft hump shows a marked decrease in flux across the 35-day observation. Fitting a constant model to the flux of the soft hump yields χ2 = 870/39 dof. The hump shows a flux range of a factor
6.44 ± 3.30 while the 2–10 keV flux (when binned the same way) has a range of a factor 3.97 ± 0.06. An alternative measure to the trough-to-peak was obtained
34 by averaging the first and last three of the 40 points for each component, this
method avoids the numbers being dominated by a single strong event. This shows
that the power-law falls by a factor 1.68 ± 0.01 and the soft hump by a factor
2.81 ± 0.24 from the start to the end of the ASCA observation. The different
amplitudes of variability of the power-law and hump explain the gross change in
softness ratio across the observation (Figure 2.1). By eye, it is clear that some of
the flares in the hard X-ray flux are also evident in the soft hump. We discuss the
detailed correlation between X-ray parameters in §2.7.
To confirm this result, we split the data into “hard” and “soft” states based
upon the softness ratio. We chose T < 106 s for our soft-state spectrum, and T
> 2.23 × 106 s for our hard-state spectrum where T is the time from the start of the observation (see Figure 2.1). Fits were performed on each state, in the same way as for the 40 individual spectra. Contour plots were generated for the normalization of the soft hump versus the photon index, Γ. Figure 2.7 shows the contours for both states, confirming a significant decrease in the strength of the soft hump across the observation. The clear separation of contours illustrates that this is not confused with changes in the photon index, which are too small to explain the observed spectral changes anyway. Γ is also plotted against the strength of the soft hump component, showing no evidence for a correlation between the two (Figure 2.8). Another concern is that one might expect to see an anti-correlation between the fluxes of the hump and power-law if they were
35 difficult to separate in the spectral fit. However, Figure 2.8 demonstrates that the high-state data have a systematically higher strength for the soft hump than the low-state. This suggests that the two spectral components are well-separated in the fitting process, and that the two components have a close physical connection.
Figure 2.6 indicates that the hump undergoes changes in shape on timescales as short as to days, as evident (for example) by comparison of panels for time-cuts
13 and 35. However, the changes are small relative to the variations in flux of the hump, thus our model with fixed shape for the hump parameterizes most of the
flux in the hump at each epoch. Thus, the evidence for some variation in hump shape does not compromise our approach in fixing the shape when testing for flux variability.
We performed a final test, to check our result was not due to variations in the SIS soft response on timescales of weeks. While the SIS degradation is thought to be progressing slowly, the accelerated deterioration evident in data taken during calendar year 2000 led us to look to the GIS for confirmation of the variability in the soft hump. Thus we excluded all SIS data below 1 keV, and used the GIS data down to 0.8 keV to determine the hump flux at each of the
40 epochs. This test strongly confirmed our result, that the soft hump flux falls during the course of this observation, and that this cannot be attributed to any instrumental effects in the ASCA data.
36 2.5.4. Variability of the Fe Emission Line
In order to investigate the variability of the Fe emission line, we utilized SIS data in the range 2.5–7.32 keV simultaneously with the GIS data in the range
1.8–9.76 keV. The model was a simple power-law plus the best-fitting profile to the Fexxvi Kα line using the Kerr geometry, and the Gaussian model to the
Fe Kβ/Ni Kα blend, the latter was linked to the flux of the Kα line, using the observed ratio from the mean spectrum. The shape parameters of the Kerr (Kα) line were assumed to be constant. As a crude test of this assumption, we split the data using an intensity division equivalent to an SIS0 count rate of 2 cts s−1, that yielded high- and low-state spectra with similar S/N in each. We then fit for the mean continuum slope as described in §2.4 and overlaid the data in the
4.88-7.32 keV band relative to the local power-law slope in each case. The two line profiles are shown as data/model ratios in Figure 2.9. The profiles appear similar, although there is some evidence for a slightly higher equivalent width in the low state. Using the high and low-state data we calculated contours of line flux versus
Γ. Figure 2.10 shows overlapping contours, thus this division of the data reveals no evidence for significant flux variability. We return to the variability of the line
EW later.
Returning to the 40 time-selected spectra, we fit those data with a model having free parameters of photon index, the normalization of the power law, and
37 the normalization of the line complex (using the shape parameters from the Kerr
fit to the mean profile), no model component was included for the soft hump, as
the soft-band data were excluded for this test. Figure 2.5 shows the time series
for the Kα line. Fits to a constant model yield χ2 =36/39 dof. Thus the line does not show significant changes in flux when sampled on this timescale.
As an alternative test we split the data into 8 intervals, to sample an intermediate timescale of several days per integration. The intervals each cover several of the original 40 time periods as follows: 1–6; 7–10; 11–17; 18–21; 22–26;
27–29; 30–35; 36–40, and were chosen to follow apparent trends in line flux from
Figure 2.5. A model was constructed using the mean Fe line profile complex from §2.4.2 and now assuming the line flux also is constant. The soft-band data were excluded, as for all the Fe line fits performed to date, and so the only free parameters in the fit were photon index and normalization of the power law. If the line is a constant flux component on top of the power-law throughout the observation, then by fitting for the local continuum slope and flux in each interval, one should be able to obtain an acceptable fit for each of the eight intervals.
Figure 2.11 shows the data/model for each fit. While some of the spectra show consistency with the mean line flux, several do not. For example, in interval 1, the line flux is significantly lower than the mean, while in interval 2, and 4, it is higher. The shortest timescale on which one can see some indication for line variability is in comparison of the first two intervals (Figure 2.11), taking the start
38 time of the first interval, and stop time of the second then this constrains the line
to arise in a region < 700, 000 light seconds, or ∼ 1 light week in size. Taking
the intervals with strongest contrast in residuals (1 and 4), we again calculated
contours of line strength versus Γ. Figure 2.12 shows these contours, which
provide supporting evidence for line variability down to timescales of several days.
As an example, the difference in fit statistic for a line of fixed versus free strength
for interval 2 is ∆χ2 = 14, indicating the variability in line flux to be significant at > 99 % confidence.
This result is supported by comparison of the Fe Kα line in the soft-state versus hard-state data (defined in §2.5.3); a significant difference in line flux and EW is evident between those two states. For the soft-state we found
+103 n . +1.5 × −5 −1 EWsoft = 491−95 eV, with line normalization soft =62−1.2 10 photons s
−2 +97 n . +1.1 × −5 −1 cm ; for the hard-state EWhard = 887−114 eV, hard =96−1.2 10 photons s
cm−2. This variation cannot be an artifact of confusion with a change in photon
index, as the soft and hard states are defined by the strength of the soft hump, and
apparently unrelated to Γ. This result confirms that the line flux and equivalent
width both vary during this ASCA observation. There cannot be a simple
correlation between soft hump flux and Fe line strength, if that was the case, we
would expect to see a gradual decrease in line strength (in Figure 2.11) as the
hump strength went down dramatically across the observation; a correlated drop
in line strength is not observed. We note the appearance of some absorption-type
39 features at some epochs (e.g. interval 2), we do not investigate these apparent minor changes to line profile, as such small effects are best investigated when the
final ASCA calibration is made available.
2.6. RMS Spectra
In optical/UV spectroscopy of AGN, it is easy to obtain a series of spectra of sufficiently high S/N to perform time-resolved spectroscopic analysis. In this case, a useful way to isolate variable features is to construct a root-mean-square
(rms) spectrum. This ASCA long look has allowed us to use our 40 individual time-selected spectra described in §2.5 in analogous way. For each detector the 40 spectra have been rebinned so that each bin has at least a 5-sigma detection (but up to a maximum of 10 adjacent bins were combined to achieve a S/N ratio of 5 within a bin) then they were degraded to the resolution of the worst spectrum
(i.e. the spectrum with worst S/N determined the actual bin widths for them all).
We created mean and rms spectra by calculating the simple mean and rms flux in each bin. The choice of a simple mean as opposed to a weighted mean was dictated by the fact that we did not want to weigh in favor of high state spectra.
The top panel in Figure 2.13 shows the mean spectrum, the middle panel, the rms spectrum. The bottom panel shows the rms spectrum after the power-law continuum is subtracted (hence only the energy bands used in the fits are plotted).
40 This shows that the most pronounced variability occurs in the soft-X band,
consistent with our results of §2.3.1, and that the variations in the Fe line are of low significance with this division of the data into 40 intervals.
2.7. Cross-correlation Results
The X-ray light curves in different energy bands (Figure 2.1) exhibit visually
similar characteristics, suggesting a short time delay between the variations in
each curve. In order to quantify any correlations, we undertook a cross-correlation
analysis using the interpolation cross-correlation function (ICCF) method of
Gaskell & Sparke (1986) and Gaskell & Peterson (1987) as implemented by White
& Peterson (1994).
We first considered a simple flux–flux correlation. We calculated the CCFs
of the total hard-flux in the 2–10 keV band with respect to the total soft-flux in
the 0.7–1.3 keV band (binned to 5760 s). The maximum value of the correlation
r . τ . +0.003 coefficient is max =0942 and the ICCF centroid cent =0012−0.011 d, less than
0.02 d at 95 % confidence. The CCF is sampled at a resolution of 0.005 day,
and the centroids are computed using all points near the peak of the CCFs with
values greater than 0.8 rmax.The1-σ uncertainties quoted for τcent are based on
the model-independent Monte Carlo method described by Peterson et al. (1998).
Given the ∼ 500 points in our light curves, there is a 0.1 % chance of exceeding
41 rmax ≈ 0.3 from uncorrelated samples. The power-law component provides ∼
75 % of the flux in the 0.7–1.3 keV band (in the mean spectrum) and so rapid variability in the power-law flux is dominating the soft/hard flux correlation. To understand the physics, we also need to examine the spectroscopically-separated components.
The cross-correlations between X-ray spectral parameters are summarized in Table 2.1. We calculated the CCFs of the light curves listed in Column (1) of Table 2.1 relative to the light curves shown in the first line, namely, the hard
X-ray continuum, the soft X-ray hump, and the photon index Γ. The maximum value of the correlation coefficient rmax and the ICCF centroid τcent relative to
the hard X-ray continuum are given in Columns (2) and (3); the ones relative to
the X-ray soft hump are given in Columns (4) and (5); and the ones relative to
the photon index are given in Columns (6) and (7), respectively. The CCFs are
defined such that a positive lag means that the light curves listed in Line (1) of
Table 2.1 are leading the light curves listed in Column (1). The number of points
in the X-ray light curves is 40; the number of points in the overlapping portion of
the X-ray and 1365 A˚ light curves is 34. The CCFs are sampled at a resolution
of 0.1 day, and the centroids are computed using all points near the peak of the
CCFs with values greater than 0.8 rmax.The1-σ uncertainties quoted for τcent
are based on the model-independent Monte Carlo method described by Peterson
et al. (1998). For our 40 points there is a 0.1 % chance of exceeding rmax =0.5
42 from uncorrelated samples. Figure 2.5 shows the light curves and the CCFs calculated relative to the hard X-ray continuum, and the hard X-ray continuum autocorrelation function (ACF).
We see a strong correlation with no significant lag between the 2–10 keV
flux and soft hump flux. The soft hump flux is distinct from the total soft-band
flux (which is the sum of power law plus soft hump counts), as it was obtained by spectral fitting. The use of quantities derived from spectral fitting limits the finest time resolution to approximately a day. The correlation is good because events on timescales of approximately a day occur in both the hump and power-law components. Also, while the soft hump shows a ∼ 50 % stronger decline over the
35 day observation than that which occurs in the 2–10 keV flux, both components show a gradual correlated decline. We also showed that the power-law flux varies down to timescales of ∼ 1000 s and the change in softness ratio indicates this occurs without a comparable change in the soft hump (Edelson et al. 2002). Thus the strength of the correlation between the spectral components depends on the timescale of sampling.
The final panel of Figure 2.5 compares the UV light curve at 1365 Awith˚ the hard X-ray flux. The correlation is strong, and the lag of the UV relative to the hard X-rays is consistent with ∼ 0, with an upper limit of 1.0 d at 95 % confidence. The Fe Kα line does not show significant correlations with any X-ray
43 quantity (and thus is not shown in the Table or Figure) as the error on each of the 40 measurements is very large. It is also important to note that Γ is not well correlated with the UV flux or the soft hump flux (Table 2.1), ruling out simple models where a steepening of the photon index can explain observed properties in the UV and soft X-ray regimes.
The strong correlation between the hard X-ray and UV curves leads us to make a direct comparison of the soft hump flux and UV light curve. The CCF of 1365 A˚ relative to the soft X-ray hump is shown in Figure 2.14. There is a good correlation between soft hump and UV 1365 A˚ flux, with the UV–soft hump lag consistent with 0 days and less than 1.0 d at 95 % confidence. Most notably, the bright X-ray flare around day 1710 is evident in the hard X-ray, soft hump and UV light curves. The difference between the correlation coefficients obtained for the hard-flux–1365 A˚ versus hump–1365 A˚ is not significant, so we cannot distinguish which is the primary link.
2.8. Summary of Observational Results
1. Ark 564 shows flux variations by factors up to 16 in the 0.7–1.3 keV band,
14 in the 2–10 keV band when sampled on 256 s timescales over a 35-day
baseline.
44 2. Fractional variability amplitude changes with time with no clear correlations
with flux or spectral parameters.
3. The mean photon index is Γ ≈ 2.54 in the hard band, with variations
of ∆Γ = 0.27 and significant changes observed down to a timescale of
approximately a day.
4. A separate soft hump component detected below 1 keV is found to be
variable down to timescales of approximately a day, ranging by a factor of
∼ 6 in normalization over the observation, but always being present. Some
changes in hump shape are evident down to timescales of days.
5. The photon index does not appear to be significantly correlated with any of
the other X-ray parameters and is not confused with soft hump strength.
6. The powerlaw and hump are not well correlated on timescales of 1000 s.
Sharp variations occur in the power-law component, but examination of
the softness ratio during those times (Edelson et al. 2002) indicates that
comparable rapid changes do not occur in the soft hump flux.
7. The soft hump shows ∼ 50 % larger variation in amplitude than the
power-law component over a baseline of weeks, when sampled approximately
daily. This difference causes changes in the softness ratio across the
observation.
45 8. The UV 1365 A˚ flux is well correlated with the soft hump and hard X-ray
flux. Correlated events (of different amplitudes) appear in all three light
curves on timescales of approximately a day. We detect no significant lag
between the UV and X-ray bands.
9. Fe Kα emission is detected in addition to a line representing a blend of Fe
Kβ emission and Ni Kα, all from H-like ions, confirming that the reprocessor
is highly ionized.
10. The flux and EW of the Fe Kα line vary, on timescales as short as a week.
2.9. Discussion and Conclusions
Ark 564 shows large amplitude and rapid X-ray variability, with a trough-to- peak range of a factor of ∼ 16 sampled in the 0.7–1.3 keV band using 256 s bins over this 35 day ASCA observation. The 2–10 keV band also shows rapid and large amplitude variability with a maximum range a factor of 14 (using 256 s bins) and factors of several change over a few thousand seconds. While the fractional variability changes with time, this does not appear to be well correlated with flux.
Furthermore, the fractional variability does not appear to be correlated with any
X-ray, UV or optical parameter at zero lag (see Paper III for the X-ray–optical correlations, Shemmer et al. 2001). It is difficult to understand the apparent lag of the peak in flickering behavior (Fvar lagsby3daysrelativetothebigX-ray
46 flares). The BLS1 NGC 7469 also shows variations in fractional variability which
do not appear to correlate with X-ray spectral parameters of UV flux in any
clear way. Ark 564 does not show the strong anti-correlation between fractional
variability and UV flux indicated for NGC 7469 (Nandra et al. 2000).
The photon index Γ ≈ 2.54 is steep for a NLS1 galaxy compared to measurements of other such sources (Brandt, Mathur, & Elvis 1997; Turner,
George, & Nandra 1998; Leighly 1999b; Vaughan et al. 1999b). The photon index varies between 2.45–2.72, comparable to the variability in Γ observed in the BLS1
NGC 7469 (∆Γ = 0.32, Nandra et al. 2000). This may indicate that the disk corona shows a similar degree of fluctuation in temperature and/or optical depth in BLS1s and NLS1, although the absolute conditions appear to be very different.
In contrast to NGC 7469, there appears to be no correlation between photon index and UV flux (Figure 2.5). In addition, the extrapolation of the hard X-ray power law overpredicts the UV and optical fluxes (Walter & Fink 1993). Thus the hard X-ray power law does not appear to simply extrapolate into the UV band in
Ark 564, and the component must terminate or turn over at wavelengths shorter than 1365 A.˚
A distinct soft hump is observed in addition to the power-law continuum.
The component rises above the power-law below 2 keV, and shows a distinct
spectral flattening below 1 keV, rising again below 0.5 keV. This hump was
47 first detected in an earlier ASCA observation of Ark 564 (Turner, George, &
Netzer 1999b). The soft hump shows flares which are correlated with those in the power-law component, down to timescales of days (the shortest timescale we can reliably probe the hump spectroscopically using these data), the flares are not generally of the same amplitude in the soft and hard bands. There is a distinction in the soft hump and power-law variability over a timescale of weeks, with the former varying by a factor of ∼ 6 across the 35-day baseline of the observation compared to a factor 4 in the power-law (when sampled approximately daily).
This difference in amplitudes of variation causes changes in the softness ratio across the observation. In this case, the overall larger measure of Fvar in the soft band is most likely due to the relatively strong changes in hump-flux on timescales of weeks.
We note that the pronounced variability of the soft hump immediately rules out an origin as starburst emission. Chandra results from another NLS1, TonS180
(Turner et al. 2001b), have shown no absorption lines, indicating the shape of the hump in that case is unlikely to be due to a combination of absorption effects and a steepening continuum form. Ark 564 is more likely to show a soft component which is modified in shape by absorption, as the UV data indicate the presence of a column ∼ 9 × 1020cm−2 (Crenshaw et al. 2002). However, the fact that the ASCA spectra of Ark 564 and TonS180 look very similar has led us to consider the soft hump in Ark 564 likely to be a continuum component
48 or broadened reprocessed component from the accretion disk, as suggested for
TonS180. Collinge et al. (2001) find a soft hump in NGC 4051 of similar form
to that in Ark 564 and TonS180 and this may also be attributable to a smooth
continuum-like component. This good correlation between the UV flux and the
soft hump and the upper limit to the lag of ∼ 1 d indicates a contribution to both bands by the same physical component, perhaps emission from the accretion disk. The UV/soft hump correlation and the apparent association of the soft hump with NLS1s appears consistent with the idea that the disk peaks at a higher temperature in NLS1s than BLS1s, as expected if we are viewing thermal emission from a disk which is more highly ionized in the NLS1 case (Ross, Fabian,
& Mineshige 1992). Unfortunately these data do not allow us to distinguish whether the soft hump is due to reflected radiation from an ionized disk, thermal emission from the disk or a combination of both. The changes in the shape of the hump may indicate that it is the superposition of several different components of emission/reflection, which possess differing timescales of variability.
In general, we find disk-corona models to be challenged by the lack of correlation of Γ with the strength of the soft hump, presumably the seed photon source, which should cool the corona and cause a steepening in the power law as it increases in strength. One possibility is that the corona cooling is saturated, i.e. that even a significant reduction in the soft hump leaves a large enough soft-photon reservoir that the resulting spectrum is always strongly cooled.
49 An Fe Kα line of high equivalent width is detected and clearly attributable to highly ionized material. These are strong observational results which are robust to the current inaccuracies in calibration. Analysis using the current calibration suggests a dominance of Fexxvi ions in Ark 564 (yielding Lyα), although a blend of emission from several states is most likely. An additional line of equivalent
+54 width 63−38 eV is detected, close to a rest energy of 8.2 keV. Lines consistent with this energy are the Kβ line from H-like Fe (Lyβ)andtheKα line from
H-like Ni (Lyα). Assuming solar abundances then the predicted equivalent width of Fe Kβ is 77 eV, and that of Ni Kα is 35 eV, based on the strength of the
Fe Kα line. Thus we attribute the line at 8.2 keV to a blend of these two, and the total observed equivalent width is consistent with such a sum. A previous short ASCA observation of Ark 564 showed ambiguity between ionization-state of the reprocessor and inclination angle of the disk system (Turner, George, &
Netzer 1999b), however, these data show that an ionized reprocessor is clearly present in this source. The discovery of ionized reprocessors in some BLS1s (e.g.,
Guainazzi et al. 1998) indicates that the luminosity of the central source may be as important as accretion rate in creating an ionized surface to the disk.
The equivalent widths for the Fe lines are huge (with EW= 653 ± 85 eV for
Lyα) compared to those observed in Seyfert 1 galaxies (e.g. Nandra et al. 1997a).
Similarly large values have been noted in previous ASCA observations of this and other NLS1s (Turner, George, & Nandra 1998; Comastri et al. 1998). The large
50 EWs may be partly explained by the relatively large line fluxes expected from
He-like and H-like ions compared to neutral material. Another possibility is that
NLS1s have high Fe abundance. Noting the high EWs of Fe Kα and the strong
Feii emission in NLS1s Turner, George, & Netzer (1999b) suggested that high Fe abundance might be a general property of NLS1s; furthermore, high metallicities are expected in some evolutionary models for NLS1s (Mathur 2000a). The high
EW cannot be due to a large solid angle for the reprocessor or a hidden hard
X-ray component, as, in either of these cases we would observe a detectable
flattening to high energies due to the presence of the associated Compton hump.
The ASCA data show variations in the Fe Kα flux by a factor ∼ 2.4 on timescales
< 7 × 105 light seconds, indicating the bulk of the iron line originates within
approximately a light week of the nucleus.
51 F2−10 keV Soft Hump Γ a a a rmax τcent rmax τcent rmax τcent (1) (2) (3) (4) (5) (6) (7) +0.10 +0.04 ··· ··· Γ 0.389 -0.05−0.35 0.058 0.00−0.59 +0.20 ··· b ··· b ··· ··· Soft Hump 0.675 0.00−0.10 F . +0.31 +1.10 +2.25 var (0.7–1.3 keV) 0.584 2 94−0.15 0.232 3.00−0.15 0.153 -0.95−0.84 F . +0.30 +2.26 +3.89 var (2–10 keV) 0.557 2 85−0.10 0.405 3.05−0.10 0.239 -0.00−5.54 +0.26 +0.35 . +0.30 Fλ (1365 A)˚ 0.683 0.39−0.84 0.633 0.35−0.44 0.112 0 70−0.30
Table 2.1: X-Ray and UV Cross-Correlation Results. Notes: (a) 1-σ uncertainties. (b) The Soft Hump ACF is shown in Figure 2.14.
52 Fig. 2.1.— Light curves for the ASCA data taken between 2000 June 1 and July 6, in cts s−1 and in 5760 s bins. The top panel is the SIS soft band (0.7–1.3 keV) light curve; the middle panel the GIS hard band (2–10 keV) and the bottom panel is the ratio of 0.7–1.3/2–10 keV. The background level in the source cell is about 4 % of the source count rate, and not plotted. The times are reported both in seconds from the start of exposure (top axis) and in JD-2450000 (bottom axis). The horizontal lines show the periods referred to as the soft and hard states.
53 Fig. 2.2.— Data/Model where the model is a simple power law fit to the 2–5 plus 7.5–10 keV data (rest-frame). The rest of the ASCA data are then overlaid, revealing a strong soft hump and Fe emission line. The data used in our spectral analysis are shown as filled circles, the data disguarded due to problems with the current ASCA calibration are overlaid, as open circles. The inset panel shows the ROSAT PSPC data compared to the continuum power-law, demonstrating the shape of the soft hump to lower energies.
54 Fig. 2.3.— Data/Model where the model is a power law plus soft hump plus Kerr model for the Fe Kα line. The data show the presence of an additional component, due to a blend of Fe Kβ and Ni Kα emission at a rest-energy ∼ 8.2 keV.
55 Fig. 2.4.— Combined SIS 0.7–1.3 keV light curve in cts s−1 and in 256 s bins. The background level in the source cell is about 4 % of the source count rate, and not plotted. The vertical dashed lines delineate our 40 time intervals within which spectra were extracted, as described in §2.5.
56 Fig. 2.5.— Spectral and timing parameters obtained from fits to the individual time-resolved spectra (left-hand column) and CCFs (right-hand column, discussed in §2.7). From the top, the light curves are the model continuum flux in the hard band, the photon index Γ, the soft hump normalization, the Kα normalization, the fractional variability Fvar in the soft and hard bands, the continuum flux at 1365 A˚ from Collier et al. (2001). The CCFs are calculated relative to the hard X-ray continuum (top, left), and the top panel on the right is the hard X-ray continuum ACF.
57 58
Fig. 2.6.— Ratio plots obtained by fitting the best-fit model for the first spectrum to the following 39 spectra. Fig. 2.7.— The ∆χ2 =2.3, 4.61, 9.21 contour levels for the soft hump normalization (in units photons s−1 cm−2) vs. photon index Γ. The dashed contours correspond to the soft state, the full contours to the hard state, and the best-fit values are indicated by crosses.
59 Fig. 2.8.— The strength of the soft hump plotted against photon index Γ, with high-state (open circles) and low states (filled circles) overlaid.
60 Fig. 2.9.— Data from the Fe K regime compared to the continuum model, with high-state (open circles) and low states (filled circles) overlaid.
61 Fig. 2.10.— The ∆χ2 =2.3, 4.61, 9.21 contour levels for Fe K-shell line intensity (in units photons s−1 cm−2) vs. photon index Γ. The dashed contours correspond to the low state, the full contours to the high state, and the best-fit values are indicated by crosses.
62 Fig. 2.11.— Data from the Fe K regime compared to the continuum model plus fixed line profile from the mean spectrum, sampled every few days. The systematically high or low residuals indicate line variability, see §2.5.4 for more details.
63 Fig. 2.12.— The ∆χ2 =2.3, 4.61, 9.21 contour levels for Fe K shell line intensity (in units photons s−1 cm−2) vs. photon index Γ. The dashed contours correspond to interval 1 the full contours to interval 4, as described in §2.5.4.
64 Fig. 2.13.— The top panel shows the mean spectrum obtained from our 40 time- selected spectra (in units of counts s−1 keV−1, errorbars in the vertical direction are included); the middle panel shows the rms spectrum, that isolates the variable parts of the spectrum; the bottom panel shows the rms spectrum after the power- law continuum is subtracted. Circles refer to the SIS-0 spectrum; triangles to the SIS-1; squares to the GIS-2; pentagons to the GIS-3.
65 Fig. 2.14.— Light curves (left-hand column) and CCFs (right-hand column). From the top, the light curves are the X-ray soft hump normalization, and the UV continuum at 1365 A.˚ The X-ray soft hump normalization is in units of photons s−1 cm−2, and the UV fluxes in units of 10−15 ergs s−1 cm−2 A˚−1. The CCF is calculated relative to the X-ray soft hump (top, left), and the top panel on the right is the X-ray soft hump ACF. Positive peaks mean that the X-ray soft hump is leading. UV data from Collier et al. (2001).
66 Chapter 3
A 12–day ASCA Observation of the Narrow-Line Seyfert 1 Galaxy Ton S180: Time-Selected Spectroscopy
3.1. Introduction
Tonantzintla (Ton) S180 (PHL 912) is a bright NLS1 with a low Galactic
20 −2 column density along the line-of-sight (NH =1.52 × 10 cm ; Stark et al. 1992).
This source has FWHM Hα and Hβ ∼ 900 km s−1 and a redshift z=0.06198
(Wisotzki et al. 1995). Ton S180 has a relatively high luminosity, with absolute magnitude MB = −23.1 mag.
TonS180wasobservedbyASCA on 1999 December 3 to 15, during a multi-wavelength monitoring campaign that included observations from HST,
RXTE, Chandra, EUVE, FUSE, in addition to optical–IR observations obtained from ground-based observatories. The results of the long-baseline timing project
67 using RXTE, EUVE and ASCA are reported in Edelson et al. (2002). The HST
and FUSE data were obtained contemporaneously with Chandra and ASCA and
were undertaken to determine the spectral-energy-distribution of the source, as
reported in Turner et al. (2002). The Chandra spectral result is reported by
Turner et al. (2001b). In this Chapter we present the results from the ∼ 400 ks
ASCA observation of Ton S180. This long observation has allowed us to study the variability of the spectral components on different timescales, from 12 days to ∼ 1 day. This is particularly interesting since the results of a 35-day long observational campaign on Ark 564 indicate that superimposed on a fast-varying continuum component that dominates the spectrum is a slower-varying soft excess emission (Turner et al. 2001a; Edelson et al. 2002). In §3.2 we describe our observations and data reduction. In §3.3 we discuss the time variability of the source. In §3.4 we analyze the mean spectrum and in §3.5 we discuss time-resolved spectroscopy. In §3.6 we present a summary of our observational results, and in §3.7 a comparison with the properties of Ark 564. Finally, in §3.8 we discuss the results.
3.2. Observations and Data Reduction
The focal-plane instruments on board ASCA comprised two CCDs (the
Solid-state Imaging Spectrometers SIS0 and SIS1, 0.4–10 keV, Burke et al. 1991)
68 and two gas-scintillation proportional-counters (Gas Imaging Spectrometers GIS2
and GIS3, 0.7–10 keV, Ohashi et al. 1996, and references therein); these were
operated simultaneously. ASCA observed Ton S180 for a total of ∼ 1 Ms, starting
on JD 2451516.051 (for the screened data, see below). The data were reduced
using standard techniques as used for the Tartarus database (Turner et al. 1999c).
Data screening yielded an effective exposure time of 338 ks for SIS0, 368 ks for the
SIS1, and 405 ks for both the GISs. The mean SIS0 count rate was 0.586 ± 0.001
cts s−1 (0.5-10 keV band).
Increased dark current levels and decreased charge transfer efficiency (CTE)
have become evident in the SIS detectors since ∼ 1993, causing a divergence
of SIS and GIS spectra. The degradation is not completely understood and
at the time of writing not corrected for by any of the software (specifically,
CORRECTRDD does not correct for the effect). The instruments can diverge by
as much as 40% for energies < 0.6 keV for data taken in 2000 January. The
Ton S180 data were calibrated using the latest calibration file released on
2001 March 29 (sisph2pi 290301.fits). The divergence of the SIS detectors at low energies can be compensated for in the spectral analysis. Yaqoob et al. (2000) provide a quantification of the apparent loss in SIS efficiency as a function of mission-elapsed time. The efficiency loss can be parameterized as a time-dependent absorption term (“excess NH”). The correction for SIS0 follows a
7 12 −2 linear relationship, NH(SIS0) = (T − 3.0174828 × 10 ) × 3.635857508 × 10 cm ,
69 where T is the average of start and stop times of the observation, measured in seconds since launch. The SIS1 excess absorption term does not follow the simple linear form found for SIS0 but it is usually found that a slightly larger absorption column can be applied to the SIS1 data to bring it into line with SIS0. For our
8 20 −2 observations, where T =2.19 × 10 s, NH(SIS0) = 6.9 × 10 cm and we adopted
21 −2 NH(SIS1) = 1.0 × 10 cm . Application of the excess NH correction is important for this analysis, as it allows us to use the valuable data at energies < 1keVto examine the spectral variability of an interesting spectral component.
3.3. The Time Variability
Light curves were extracted using bin sizes of 256 s and 5760 s in the full-band (0.7-10 keV) for the SIS, the soft-band (0.7–1.3 keV) for the SIS data, and the hard-band (2–10 keV) for both GIS and SIS data. We adopted 0.7 keV as a lower limit for the SIS data, as appropriate for the setting of the SIS lower level discriminator for these observations. Light curves were constructed combining data from the SIS and GIS detector pairs. The exposure requirements for the combined light curves ensured the bins be fully exposed in each instrument for the
256 s curves and at least 10 % exposed for the 5760 s curves. The observed count
−12 −1 −2 rates correspond to a mean 2–10 keV flux of F2−10 =6.5 × 10 ergs s cm
43 −1 −1 −1 and 2–10 keV luminosity L2−10 =4.9 × 10 ergs s (H0 =75kms Mpc ,
70 q0 =0.5). This mean flux level is ∼ 20 % brighter than that observed during a
previous ASCA observation on 1996 July 10 (Turner, George, & Nandra 1998).
Figure 3.1 shows the combined 0.7–1.3 keV SIS soft-band and GIS hard-band
light curves in 5760 s bins. The background levels in the source cells are about 4%
and 10% of the SIS and GIS source count rates, respectively, and not plotted, or
subtracted. Figure 3.1 also shows the softness ratio, defined as the ratio between
the count rates in the 0.7–1.3 and 2–10 keV bands. There is significant hardening
of the spectrum during the observation, and the softness ratio changes by ∼ 20 %.
The light curves integrated to 5760 s show trough-to-peak flux variations
by a factor of ∼ 2.7 in the 0.7–1.3 keV band (SIS), ∼ 2.4 in the 2–10 keV band
(GIS). The light curves sampled on 256 s reveal even larger amplitudes due to
fast flickering, with a maximum amplitude of variability of a factor ∼ 3.5 for the
SIS data in the 0.7–1.3 keV band and ∼ 3.9 for the GIS data in the 2–10 keV band. Figure 3.2 shows the 0.7–1.3 and 2–10 keV light curves in 256 s bins for three rapid “events” centered around 603, 608, and 614 ks from the start of the observation (JD 2451523.03, 2451523.08, and 2451523.16). In the first event the soft and hard-band data show a variations (trough-to-peak) of factors of
Rmax(SIS) = 1.32 ± 0.13 and Rmax(GIS) = 1.80 ± 0.38, respectively, in ∆t = 1972 s
(the errors on Rmax are obtained propagating the errors in the light-curve points).
In the second event Rmax(SIS) = 1.34 ± 0.13 and Rmax(GIS) = 1.75 ± 0.37,
respectively, in ∆t = 1972 s. In the third event, the variation is a factor of
71 Rmax(SIS) = 1.33 ± 0.13 for the soft while the hard is consistent with a constant.
These events correspond to a variation in luminosity of ∆L =2.4 ± 0.9 × 1043
ergs s−1 (as calculated from SIS data, from the first event; GIS data yield
∆L =2.6 ± 0.9 × 1043 ergs s−1). In the first and second events the hard X-ray
flux variation is sharper than that in the soft band, and there is a change in the
softness ratio, in the sense that when the flux increases the spectrum hardens,
with a timescale of ∼ 1000 s (Figure 3.2, bottom panel). However, in the third
event a flux change is not accompanied by a strong spectral change. Further
examination of the light curves reveals other examples of diverse variability
behavior with no obvious general trend. This is a behavior previously reported in
ROSAT observations (Fink et al. 1997).
The soft and hard X-ray light curves in Figure 3.1 exhibit similar
characteristics, suggesting a short time delay between the variations in each curve.
To quantify any correlations we undertook a cross-correlation analysis using
the interpolation cross-correlation function (ICCF) method of Gaskell & Sparke
(1986) and Gaskell & Peterson (1987) as implemented by White & Peterson
(1994). We calculated the CCFs of the total hard-flux in the 2–10 keV band with
respect to the total soft-flux in the 0.7–1.3 keV band, which has the highest S/N.
The CCFs are sampled at a resolution of 0.05 day, and the centroids are computed
using all points near the peak of the CCFs with values greater than 80% of
the maximum value of the correlation coefficient, rmax.The1-σ uncertainties
72 quoted for the ICCF centroid, τcent, are based on the model-independent Monte
Carlo method described by Peterson et al. (1998). We obtain rmax =0.748 and
τ . +0.024 cent =0025−0.002 days. Given the 180 points in our light curves, the probability of exceeding rmax ≈ 0.3 from uncorrelated samples is 0.1 %, and the lag is less than 0.07 d at 95 % confidence. Since the power-law component provides ∼ 72 % of the flux in the mean spectrum 0.7–1.3 keV band, as we will show in §3.4.1, and the power-law component dominates the 2–10 keV band, this correlation is probably dominated by the rapid variability in the power-law flux.
3.3.1. Fractional Variability Amplitude
First we calculated Fvar (defined in Eq. 2.1) across the baseline of the entire observation. This quantity measures deviations compared to the mean, integrated over the entire 12 days. We measured this quantity in the soft (0.7–1.3 keV) and hard (2–10 keV) bands, using our light curves with 256 s bins. These bands were chosen to sample energy ranges containing different spectral components (see
§3.4) while still maintaining good S/N. Fvar thus calculated is 19.12 ± 0.58 % in the 0.7–1.3 keV band, and 17.26 ± 0.65 % in the hard-band, using SIS data for both tests.
To examine the evolution of timing properties of Ton S180 further, we split the data into 14 evenly-sampled sections across the 12 day ASCA observation.
73 Even sampling is important because Fvar depends strongly on the duration of the
data-train. The baseline for each time-selected section of data was 75 ks with
average on-source exposure time 25 ks. We then calculated Fvar over each of these
one-day intervals (again with 256 s bins in each “daily” light curve). This test
showed that Fvar is significantly variable showing correlated changes in the soft
and hard bands, on a day-to-day basis (figure 3.3).
As we will demonstrate in §3.4.1, in the mean spectrum the power-law continuum provides about 72 % of the flux in the 0.7–1.3 keV band, and the power-law variations contribute significantly to Fvar in both the soft and hard
bands, explaining the gross similarity between the quantity in those two bands.
The slightly greater Fvar for the soft band is probably due to a combination of
changes in the relative flux levels of two spectral components (power law and
hump), and slope changes in the power-law, as we will show in §3.5.3.
3.4. The Mean Spectrum
Source counts were binned with a minimum of 20 counts per energy bin for
the spectral analysis. The data from the four instruments were fit simultaneously,
with the relative normalizations free to allow for small differences in calibration of
the absolute flux and in the fraction of encircled counts encompassed by the SIS
versus GIS extraction cells. Spectral fits were performed using XSPEC V11.0.1
74 and response files generated with HEAsoft v5.0.4. Ton S180 is detected at better than 3-σ level in the SIS data for energies ∼< 7.5 keV and in the GIS data for energies ∼< 9.5 keV (observed frame, in the unbinned spectrum).
−Γ The photon index Γ (defined by photon flux PE ∝ E ) was determined by
fitting a power-law model attenuated by Galactic absorption with an additional correction applied to SIS0 and SIS1 to compensate for the low-energy degradation as described in §3.2. For this fit we used data in the bandpass 1.8–4.71 keV
(observer’s frame). The power-law fit yielded Γ = 2.44 ± 0.02 and χ2 = 720 for 683 degrees of freedom (dof). Inclusion of the data in the 7.06–9.42 keV bandpass produces consistent results. This and subsequent errors represent, unless otherwise specified, the 90 % confidence level.
The data/model ratio is shown in Figure 3.4, top panel. A strong soft excess is evident, appearing as a hump of emission rising above the power-law continuum at energies < 2 keV. This feature was also observed in the ROSAT
PSPC observations (Fink et al. 1997, inset in top panel of Figure 3.4), BeppoSAX observations (Comastri et al. 1998), and the earlier ASCA observation (sequence number 74081000; Turner, George, & Nandra 1998). Hereafter we refer to this component as the “soft hump”. Also evident in the main panel is an excess of emission close to 7 keV, which is due to an unmodeled Fe Kα emission line.
75 Minor calibration problems are visible in such high S/N data at ≈ 1.7–
3.0 keV. They are not so serious as to warrant exclusion from the fit. We note, however, that they contribute significantly to the χ2 in the soft component fits
(§3.4.1) and to a lesser extent to the Kα fits (§3.4.2).
3.4.1. The Soft Component
We confirm the presence of the soft X-ray emission component previously observed by Fink et al. (1997), Comastri et al. (1998), and Turner, George, &
Nandra (1998). At the time of writing, the status of the SIS calibration limits the accuracy to which the absolute form of the soft spectral hump can be determined.
Thus we adopted a simple parameterization of the soft hump which allowed us to perform a sensitive examination of the flux variability of the component. The simultaneous Chandra LETG data for Ton S180 (Turner et al. 2001b) and the results for NGC 4051 (Collinge et al. 2001) show that the soft hump is a smooth continuum component, as opposed to a blend of unresolved spectral features. The variability study afforded by the ASCA data is therefore the optimum available tool for exploring the nature of the soft hump. This was also shown to be the case for Ark 564 (Chapter 2; Turner et al. 2001a).
We used SIS data in the range 0.7–4.71 keV simultaneously with GIS data in the range 1.0–4.71 keV (observer’s frame). The lower energy limit of the SIS
76 data was based on the level of agreement achieved between the two CCDs using
our methods of correction (§3.2). The upper energy limit of both SIS and GIS effectively excluded the Fe Kα regime for these fits, in order to avoid an overly complex model (which can result in false or local minima being found). A steep power-law component is an inadequate representation of this excess, as its form shows some curvature. The Gaussian model which best fits the soft hump has
E . +0.17 an energy of the peak =017−0.17p keV (the energy pegged at the lower limit),
. +0.05 . +0.37 × −2 full width at half maximum FWHM = 1 01−0.12 keV and flux 1 50−0.61 10
−1 −2 +23 photons s cm corresponding to a mean equivalent width EW = 94−38 eV. For
this fit χ2 = 1069 for 896 dof, but there is a contribution to the χ2 from the ≈
2.5–3.0 keV region due to calibration problems (see §3.4). Significantly inferior (to
better than 99.9 % confidence; χ2 = 1096 for 897 dof) is a parameterization of the
soft hump as a blackbody model; the best fit that yielded a rest-frame temperature
kT +2 . +0.03 × −4 −1 −2 = 153−3 eV, flux 1 07−0.04 10 photons s cm , and absorption corrected
43 −1 luminosity in the 0.7–1.3 keV band L0.7−1.3 =1.6 × 10 ergs s .
Using a Gaussian parameterization of the soft hump, we determined that the
power-law continuum contributes to ∼ 72 % of the flux in the soft (0.7–1.3 keV)
band of the mean spectrum. We also retrieved the 1996 ASCA spectrum (Turner,
George, & Nandra 1998) from the Tartarus database (Turner et al. 1999c), and
found consistent contributions of the different spectral components to the soft
band over the same energy ranges as we used for our 12-day observation.
77 3.4.2. The Fe Kα Regime
Comastri et al. (1998) and Turner, George, & Nandra (1998) found evidence
for line emission at ∼ 7 keV, indicative of an origin from material containing
ionized iron, consistent with the iron line from H-like iron at 6.94 keV. Hence,
having found an adequate parameterization of the continuum shape (in the
1.8–4.71 keV band), we included the 4.71–7.45 keV SIS data and the 4.71–9.42 keV
GIS data in the analysis and examined the data/model ratio (versus the best-fit
continuum model, top panel of Figure 3.4). The mean line profile is broad and
asymmetric, similar to that observed in the previous ASCA observation (Turner,
George, & Nandra 1998).
We utilized the SIS data in the 1.8–7.45 keV range simultaneously with the
GIS data in the 1.8–9.42 keV range. Table 3.1 summarizes our results: Column
(1) lists the models, Column (2) the photon index, Column (3) the rest-frame
energy of the fitted Fe line, Column (4) its equivalent width, and Column (5) χ2
and degrees of freedom relative to the fit. PL indicates the power-law continuum,
BG the broad Gaussian, and NG the narrow Gaussian (see details below).
The asymmetry of the profile prompted us to fit the iron Kα line using the diskline model profile of Fabian et al. (1989). This model assumes a
Schwarzschild metric, with an emissivity law r−q for the illumination pattern of
the accretion disk, where r is the radial distance from the black hole. We adopted
78 q =2.5 based upon the results of Nandra et al. (1997b). We also assume that
2 the line emission originates within 1000 gravitational radii (Rg = GM/c ). The
inclination of the system is defined such that i = 0 is a disk oriented face-on
to the observer. Fits using this model yielded χ2 = 1353 for 1368 dof.The
E . +0.27 r +10 R rest-energy of the line was =640−0.00p keV, the inner radius was =6−6p g,
i +22 . +0.54 × −5 −1 −2 the inclination was =35−35p degrees, and flux 2 06−0.37 10 photons s cm .
+120 The equivalent width was EW = 461−84 eV.
We also tested a model for the line profile assuming a Kerr metric for a
maximally rotating black hole as implemented by Laor (1991). This will have
the most intense gravitational effects. We fixed the emissivity index as for the
Schwarzschild case and the outer radius at the maximum value allowed by the
model, 400 Rg. Using the same energy restriction as for the Schwarzschild model,
the Kerr model provides a fit-statistic χ2 = 1352 for 1368 dof. The rest-energy
E . +0.16 i +14 of the line was =655−0.15p keV, inclination was =23−23p degrees, and flux
. +0.53 × −5 −1 −2 +123 2 22−0.47 10 photons s cm . The equivalent width was EW = 517−111 eV.
The line was better fit as the sum of a broad and a narrow redshifted
Gaussian profile, with χ2 = 1328 for 1359 dof (a broad or narrow Gaussian
alone are a significantly worse fit at > 99 % confidence; see Table 3.1). The rest
E . +0.08 energy of the narrow line (fixed at 10 eV width) was N =681−0.12 keV, with EW
+33 E . +0.25 +225 =90−31 eV. The broad component gave B =658−0.28 keV and EW = 512−136 eV.
79 The residuals of the 2-Gaussian fit are shown in the bottom panel of Figure 3.4.
Some excess emission is evident in the 8–9 keV region. However, the S/N in this energy range is so low (the detection level is ∼ 3 σ) that no meaningful spectral
fit can be performed.
We do not detect any narrow (fixed at 10 eV width) Kα emission at rest energy 6.4 keV, in addition to the Schwarzschild, Kerr, single broad Gaussian, or
2-Gaussian models.
3.5. Spectral Variability
3.5.1. Method and Selection Details
To examine the spectral evolution of Ton S180, we created 14 time-selected spectra across the 12 day ASCA observation by sampling throughout the light curve following flares and dips with Xselect V2.0. The resulting average baseline for each time-selected spectrum was 75 ks, for an average on-source exposure time
25 ks. Our choice of intervals is shown in Figure 3.5. The “events” described in
§3.3 (Figure 3.2) occur during time bin 9. Background, ancillary response, and response matrix files were set to be those of the mean spectrum. The background spectrum and flux did not vary significantly during the observation and use of the mean spectra yielded the best possible S/N for the time-resolved spectroscopy.
80 Again, spectra were binned to achieve a minimum of 20 counts per energy bin.
We performed all fits by fixing the scaling factors for instrument normalization
based upon the fits to the mean spectrum. Corrections for the SIS low-energy
problem were fixed at the same values used for the mean spectrum. All models
were modified with a Galactic absorption. Time assignments for the spectra
refer to the mid-point of the observation, in JD − 2450000. The results of our analysis are shown in Figure 3.3 in the form of time series curves for the various parameters. These are described in detail in the following sections.
3.5.2. Variability of the Continuum
We fit each of the 14 time-selected spectra in the 1.8–4.71 keV range (both
SIS and GIS; observer’s frame) with a simple power-law model. The light curves
for the (model) continuum flux and the best-fit photon index Γ are shown in
Figure 3.3. The photon index Γ, thus sampled on a ∼ 1 day timescale, ranges in 2.38–2.62 (i.e., ∆Γ = 0.24) across the 12 days, but these variations are not significant (fitting the photon indices to a constant model yields χ2 = 10 for
13 dof). Therefore, the rapid flux variability evident on timescales as short as
∼ 1000 s in the hard band (Figure 3.2, §3.3) is not due to changes in Γ, but is
truely flux changes in that band. However, on longer timescales of about a week
significant changes in Γ occur, as we will show in §3.5.3.
81 Figure 3.6 shows ratio plots obtained by comparing the best-fit model for the
first spectrum to the following 13 spectra, without performing any fit. The plot illustrates the variations of the spectral shape and flux compared to the first day of data. Strong variations in the flux and, to a lesser degree, in the continuum slope are evident. The soft hump is always present superimposed on the power-law continuum, although it appears to change in absolute strength, relative to the power-law continuum and possibly in shape (see, for example time-cuts 2 and 13) on a timescale of approximately a day. Variations in the shape of the soft hump, however, are a minor effect compared to the flux variations of the soft hump.
3.5.3. Variability of the Soft X-ray Hump
We examined the variability of the soft hump using SIS data in the range
0.7–4.71 keV simultaneously with GIS data in the 1.0–4.71 keV range (observer’s frame), as we did for the mean spectrum in §3.4.1. The model was a simple power-law plus a broad Gaussian for the soft hump, with the Gaussian peak and width fixed at the values noted in §3.4.1. The flux of the soft hump, the flux of the continuum power law, and the power-law slope were free parameters. The time series for the flux of the soft hump is shown in Figure 3.3. Fitting a constant model to this flux yields χ2 = 47 for 13 dof. The soft hump flux varies by a factor
82 of Rmax =2.33 ± 0.45 while the 2–10 keV flux (when binned the same way) varies by a factor of Rmax =1.65 ± 0.02.
Chapter 2 (Turner et al. 2001a) showed that in Ark 564 the different amplitudes of variability of the soft hump and power-law (6.44 ± 3.30, compared to 3.97 ± 0.06) explain the gross change in softness ratio across the observation.
In Ton S180, the amplitude of variation of the soft hump is not so pronounced, although it is clear from the light curves (Figure 3.3) that most of the flares in the hard X-ray flux are also present in the soft hump. To clarify the situation in
Ton S180, we split the data into “hard” and “soft” states based upon the softness ratio. We chose T < 6.3 × 105 s for our soft-state spectrum, and T > 6.3 × 105 s for our hard-state spectrum where T is the time from the start of the observation (see
Figure 3.1). Fits were performed on each state, as we did for the mean spectrum
(§3.4.1), and confidence contour plots were generated for the flux of the soft hump versus the photon index, assuming the soft hump shape to be approximately constant during the observation. The resulting Γ–soft hump flux contours are well separated (Figure 3.7, top) in both directions. Hence, the confidence contour plot indicates that the source varies significantly in photon index and strength of the soft hump on a timescale of about one week. Therefore both effects are contributing to the change in softness ratio observed across the baseline of the observation. The spectral changes observed between the soft and hard states
83 result in a change of the EW of the soft hump from EWsoft = 100 ± 5eVto
EW +6 hard =84−7 eV.
Since we adopt a fixed soft hump shape in our fits of the 14 time-selected spectra, we are confident our fitting method can separate the fluxes of soft hump and power-law continuum. Furthermore, the bottom panel in Figure 3.7 shows
Γ plotted against the strength of the soft hump component and demonstrates a strong correlation (Spearman rank coefficient rs =0.68, probability of chance occurrenceis8×10−3), as is expected from disk-corona models, as we will see in
§3.8. If the strength of the soft hump component and the photon index were difficult to separate in the spectral fit, we would expect an anticorrelation instead.
Figure 3.7 also shows that a steeper photon index appears to be associated with a relatively strong soft hump. Finally, as noted in §3.5.2, the variations in the shape of the Gaussian soft hump are small so that our model for the soft hump parameterizes most of the flux in each time-selected spectrum, and does not weaken our approach in fixing the shape when testing for flux variability.
3.5.4. Variability of the Fe Emission Line
As we have seen in §3.5.2, Figure 3.6 suggests that there are small variations in the strength of the Fe emission line. To investigate this possibility, we utilized
SIS data in the 1.8–7.45 keV range simultaneously with the GIS data in the
84 1.8–9.42 keV range, as we did for the mean spectrum in §3.4.2. The model was a simple power-law continuum plus different models for the Kα line, with fixed shape parameters. We initially tested the assumption of a fixed shape by splitting the data with an intensity division equivalent to an SIS0 count rate of 0.6 cts s−1.
This yielded high- and low-state spectra with similar S/N. We also calculated confidence contours of line flux versus Γ. Figure 3.8a shows overlapping contours, thus this division of the data, corresponding to a timescale of about 1 week, reveals no evidence for significant flux variability. Using the high and low-state data we then fit for the mean continuum slope as described in §3.4 and overlaid the data in the Kα band relative to the local power-law slope in each case. The two line profiles are shown as data/model ratios in figure 3.8b and they appear indistinguishable. The same test, performed with the soft- and hard-state division discussed in §3.5.3, leads us to conclude that no changes were observed either in
flux or in shape of the line. Figure 3.8c,d show the confidence contour and the line profile, respectively, for the soft- and hard-state spectra. We further tested the possibility that the shape of the line might be variable, on a ∼ 1 day timescale by overplotting the line profiles for spectra that looked qualitatively different. No significant changes in the line profile are observed.
Once assured that variations of the line shape, if present, are minor, we returned to our 14 time-selected spectra, where we considered the following models for Kα line: (1) a laor (Kerr) model; (2) a 2-Gaussian model (§3.4.2)
85 with the two fluxes allowed to vary independently; (3) a 2-Gaussian model in which either the broad Gaussian flux or (4) the narrow Gaussian is flux fixed to the best-fit value from the mean spectrum; and (5) a 2-Gaussian model in which fluxes have fixed ratio derived from fits to the mean spectrum. All fits were performed with photon index and the flux of the power law continuum left as free parameters. The shape of the Kα line was kept fixed to the best fit values obtained for the mean spectrum. No model component was included for the soft hump, as we excluded the soft-band data for these fits. Figure 3.3 shows the time series for the Kα line for the Kerr model, yielding results similar to those obtained from models (2) through (5). Fits to a constant model yield χ2 = 3 for 13 dof, thus the line does not show significant changes in flux when sampled on ∼ 1day timescale. Analogous conclusions can be drawn for all other models considered.
Finally, we considered the spectra extracted in bins 4 and 13, which have the highest and the lowest measured Kα flux, respectively (Figure 3.3). We calculated confidence contours of line flux versus Γ and overplotted the line profiles as we did above. Figure 3.8e shows the contour plots which demonstrate that the flux variation is insignificant and there is a large overlap of the contours. Figure 3.8f shows the line profiles, whose large equivalent widths are consistent at 90 %
+219 +252 confidence (EW4 = 889−368 eV, EW13 = 411−249 eV). Therefore, we do not detect significant variations in EW or flux of the Kα line. With the S/N available using
86 these data the line flux would have had to vary by at least a factor of 2 for us to detect it at the 99 % confidence level (cf. Figure 3.8e).
We also applied a laor model to the 1996 ASCA observation (Turner,
George, & Nandra 1998), within the same energy ranges as we used here and with the corrections for the SIS low-energy degradation appropriate to 1996 (see §3.2), and obtained peak energy, EW, and inclination for the Kα line consistent with what we found in §3.4.2.
3.5.5. RMS Spectra
We used the technique described in 2.6 to create rms and mean spectra. For each detector the 14 spectra have been rebinned so that each bin has at least a
5-sigma detection (but up to a maximum of 10 adjacent bins were combined to achieve a S/N ratio of 5 within a bin) then they were degraded to the resolution of the worst spectrum. Figure 3.9 shows the ratio rms/mean spectrum for SIS1 and SIS0, using data with energy > 0.7keV.Themeanvalueis ∼ 20 % and there is no systematic trend for different energies. This is consistent with the results from §3.4.1 showing that the power-law continuum component contributes most of the flux in the hard band and ∼ 72 % of the flux in the soft band. The possible weak trend for energies < 2 keV of increased variability for decreasing energy, may suggest that the soft hump is partially driving the variability in the soft
87 band. The “spike-like” features at ∼ 1.7–2.5 keV are probably due to calibration uncertainties.
3.6. Summary of Observational Results
1. On a 12-day baseline, the X-ray flux of Ton S180 presents trough-to-peak
variations by a factor 3.5 in the 0.7–1.3 keV band, and 3.9 in the 2–10 keV
band when sampled using 256 s bins.
2. The mean photon index, calculated from the continuum fit of the mean
spectrum, is Γ = 2.44 ± 0.02. Time-resolved spectroscopy reveals significant
changes on timescales of ∼ 1 week.
3. We confirm the presence of a separate “soft hump” component at energies
< 2 keV. This component shows flux variations down to timescales of ∼ 1
day, ranging by a factor of 2.3 in normalization over the 12 days of our
observations. Some flux changes appear to match changes in the 2–10 keV
flux on this timescale although the soft hump shows a drop in EW from
100 to 84 eV across the observation. Minor changes in the shape of the soft
hump are apparent on timescales of a few days.
4. In our approximately daily sampling the photon index is correlated with the
flux of the soft X-ray hump.
88 5. Variations in both the photon index and soft hump strength contribute to
the change in softness ratio observed across the ASCA observation.
6. The softness ratio reveals spectral variability down to timescales of ∼ 1000 s
in addition to the slow decline across the observation. The power-law and
soft hump show divergent behavior on very short (1000 s) or long (2 weeks)
timescales.
7. The Fe Kα line emission is detected with a narrow component peaking at
∼ 6.8 keV, indicating an origin in ionized material. A broad line component
(EW ∼ 500 eV) is also evident. We do not detect significant variations of
theFeKα line strength or equivalent width, on timescales of ∼ 1 day–12
days.
3.7. Comparison with Ark 564
Comparison of our results for Ton S180 with those obtained for Ark 564 in
Chapter 2 (Turner et al. 2001a)) reveals broad similarity in the overall shape of the X-ray spectra (steep power-law continuum, strong soft excess, and ionized, large EW Fe Kα), and variability characteristics. Given these similarities, that may be indicative of the characteristics of the NLS1 as a class, we summarize the properties of these two NLS1s in Table 3.2. Column (1) lists the properties;
89 Column (2) and (3) the values for Ton S180 and Ark 564, respectively; Column
(4) and (5) the references for Ton S180 and Ark 564, respectively.
3.8. Discussion and Conclusions
Ton S180 reaffirms the rapid and large-amplitude X-ray variations which are a characteristic of the NLS1 class. During our 12-day observation, the light curves sampled on 256 s timescales show trough-to-peak flux variations by a factor of ∼ 3.5 in the soft band (0.7-1.3 keV), and up to a factor of ∼ 4inthe hard band (2–10 keV). Further examination of these light curves shows variations up to a factor of ∼ 2 occurring on timescales of ∼ 1000 s which are sometimes accompanied by spectral changes. There is a strong correlation (rmax =0.748) between the soft- and hard-band light curves, with a 95 % confidence upper limit on the lag of hard with respect to soft of τcent < 0.07 d. The strength of this correlation, the short lag, and the fact that ∼ 72 % of the flux in the soft band is contributed by the power-law continuum while the hard flux is driven by the power-law, indicate that this correlation is dominated by the variations of the power-law continuum itself.
We used these timing data to obtain an estimate of some fundamental parameters for Ton S180. Our estimate of the accretion efficiency is η ∼> 7%, marginally above the limit for the efficiency in the Schwarzschild geometry,
90 perhaps favoring a Kerr geometry. However, large uncertainties are associated with this estimate (see Brandt et al. 1999). Light-crossing time arguments yield a
> 6 radius R for the emitting region, R ∼< 12 RS for MBH ∼ 8 × 10 M . The 1000 s timescale for variability observed in Ton S180 indicates that neither thermal nor viscous/radial drift phenomena can be held responsible for the observed fast variability (Frank, King & Raine 1992). The values of these parameters are consistent with what is generally found for other Seyfert 1s. Haardt, Maraschi, &
Ghisellini (1994) discuss a model where soft thermal photons from the accretion disk are Compton-upscattered in localized blobs of coronal plasma, constrained by magnetic loops from the disk. The resulting X-rays from the blobs then illuminate the disk and produce the so-called Compton Hump and Fe Kα emission line.
The amplitudes and timescales of the rapid variations observed in Ton S180 are consistent with those expected as a result of stochastic noise in the number of reprocessing blobs, which depend on the formation and reconnection of magnetic loops.
The continuum fit to the mean spectrum yields a photon index Γ = 2.44±0.02.
Our time-selected spectral fits give Γ =2.47 ± 0.07 (1-σ error) with variations from 2.38 to 2.62. The steep index and the range (∆Γ = 0.24) are comparable with that observed in Ark 564 (Γ = 2.54, ∆Γ = 0.27, Chapter 2; Turner et al.
2001a), though the variations on a ∼ 1 day timescale are significant in the case of Ark 564 and not for Ton S180 (where we detect variations of Γ in ∼ 1 week
91 timescale). The range of indices is also consistent with that found for the BLS1
NGC 7469 (∆Γ = 0.32, Nandra et al. 2000). This may indicate a fundamental similarity in the process which causes slope fluctuations, over a range of different
AGN.
We confirm the presence of the soft X-ray emission component at energies
< 2 keV previously observed by Fink et al. (1997), Comastri et al. (1998), and
Turner, George, & Nandra (1998). This component rises above the power-law continuum ∼< 2 keV, and has recently been shown to be a smooth continuum component rather than a blend of features, not only in Ton S180 (Turner et al. 2001b), but also in the NLS1 NGC 4051 (Collinge et al. 2001). Therefore, our time-selected spectroscopic analysis is particularly well suited to study its variability properties in particular and understand the nature of the soft hump in NLS1s in general. The soft hump shows flux variations on timescales as short as ∼ 1 day (the shortest timescales our time-selected spectroscopy can study), ranging by a factor of 2.3 in flux over the observation, but always being present.
We note that the fast variability observed in the soft hump of Ton S180 rules out an origin of the soft emission in large-scale components, such as circumnuclear starburst (as also concluded for the soft hump in Ark 564, Chapter 2; Turner et al. 2001a). The hard flux, when similarly binned on timescales of a day, has a range of a factor 1.65. The softness ratio shows spectral variability on timescales as short as ∼ 1000 s. In Ton S180 such fast spectral variability can be attributed
92 either to rapid changes in photon index or in the relative strengths of the soft hump and power-law. In addition to this evidence for divergent behavior on very short timescales, the soft hump and power-law show different trends on timescales of ∼ 1 week. The higher value of Fvar measured in the soft than in the hard band is probably due to the combined effects of a larger amplitude of variation of the soft hump over the baselines of the observation, and the changes in spectral slope (which could preferentially affect the soft band if the pivot point lies within the hard band). The correlation observed between the soft hump flux and the photon index in Ton S180 is expected in disk-corona models where an increase of the flux in the soft X-ray/UV component can cool the corona and steepen the power-law continuum (Haardt, Maraschi, & Ghisellini 1994; Pounds, Done, &
Osborne 1995). This is not the case for Ark 564, where we suggest (Chapter 2,
Turner et al. 2001a) the corona cooling might be saturated.
Iron Kα line emission is detected, with a broad, asymmetric profile. It is best parameterized by two Gaussian components; the narrow component peaks at EN =6.81 keV, which is consistent with emission from highly ionized gas, the broad component consistent with emission from either neutral or ionized gas. We caution about the reliability of this result because the S/N in this range is low and the errors on the fitted parameters are correspondingly large (Table 3.1).
Therefore it seems that the best parametrization for the Fe Kα line is a broad ionized line, which is consistent with previous BeppoSAX (Comastri et al. 1998)
93 and ASCA (Turner, George, & Nandra 1998) observations. The broad component
is generally thought to originate in the innermost regions of the accretion disk
around the central black hole (e.g., Fabian et al. 2000, and references therein),
and so it is due to highly ionized Fe, while the narrow component is thought
to be generated in the putative obscuring torus that extends on a parsec scale
(Ghisellini, Haardt, & Matt 1994; Krolik, Madau, & Zycki˙ 1994; Yaqoob et al.
2001), and so is believed to be due to neutral material. In this context, the result
of an ionized narrow Fe Kα component is therefore somewhat suprising. Note,
however, that observations of such a feature are not unprecedented: Sako et al.
(2000) and Ogle et al. (2000) detected narrow lines due to highly ionized iron in
Chandra observations of Mkn 3 and NGC 4151, respectively.
TheFeKα emission line has a very large equivalent width, EW ≈ 500 eV,
as previously observed in other ASCA and BeppoSAX observations of Ton S180,
Ark 564, and other NLS1s (Turner, George, & Nandra 1998; Comastri et al. 1998;
Turner et al. 2001a). These EWs are interesting compared to a sample taken across the Seyfert 1 population (Nandra et al. 1997b), which found an average
EW = 230 eV. One interpretation is in terms of an extreme Fe abundance in
NLS1s, as proposed by Turner, George, & Netzer (1999b) for Ark 564; this may, in turn, support the proposition that NLS1s are Seyfert galaxies in an early stage of evolution (Mathur 2000a). Alternatively, Matt, Fabian, & Ross (1996) show that the expected EWs for predominantly H- and He-like Fe can be a factor of a
94 few higher than those from neutral Fe. We do not detect significant variations of the Fe Kα line, on timescales of ∼ 1 day or of ∼ 1 week. We note that given our
S/N in the Kα energy range for the 14 time-selected spectra, the line flux would have had to varied by at least a factor 2 for us to have detected the variation at
> 99 % confidence level.
95 a a,b a 2 Model Γ2−10 EFe EW χ /dof (keV) (eV) (1) (2) (3) (4) (5) PLc 2.37 ± 0.01 ··· ··· 1518/1372 . +0.01 . +0.27 +120 PL + DISKLINE 2 43−0.02 6 40−0.00p 461−84 1353/1368 . ± . . +0.16 +123 PL + LAOR 2 43 0 02 6 55−0.15p 517−111 1352/1368 . ± . . +0.08 +30 PL+NG 239 0 01 6 75−0.04 182−26 1404/1370 . ± . . +0.12 +184 PL+BG 244 0 02 6 71−0.14 550−128 1352/1369 . +0.03 . +0.25 +225 PL + BG + NG 2 46−0.02 6 58−0.28 512−136 1328/1359 . +0.08 +33 6 81−0.12 90−31
Table 3.1: Spectral Fits in the 2–10 keV Band. Notes: (a) 90 % confidence level uncertainties. (b) Rest frame of Ton S180. (c) Over the whole hard energy range. The value of the photon index from the continuum fit is 2.44 ± 0.02 (§3.4).
96 Ton S180a Ark 564a Referencesb (1) (2) (3) (4) (5) z 0.06198 0.0247 1 2 −2 20 20 Galactic NH (cm )1.52×10 6.4×10 34 V (mag) 14.4 14.6 5 6 43 −1 L2−10 (10 ergs s ) 4.9 2.4 7(2.3) 8(3.3)
Rmax soft band 3.5 16 7(2.3) 8(3.3)
Rmax hard band 3.9 14 7(2.3) 8(3.3)
Rmax Soft Hump 2.33 ± 0.45 6.44 ± 3.30 7(2.5.3) 8(3.5.3)
Rmax 2–10 keV 1.65 ± 0.02 3.97 ± 0.06 7(2.5.3) 8(3.5.3)
Fvar (12 d, soft) (%) 19.12 ± 0.58 35.81 ± 0.81 7(2.3.1) 7
Fvar (12 d, hard) (%) 17.26 ± 0.65 35.73 ± 0.70 7(2.3.1) 7
Fvar (1 d, soft) (%) variable variable 7(2.3.1) 8(3.3.1) correlates with X-ray no no 7(2.3.1) 8(3.3.1, 3.7)
Fvar (1 d, hard) (%) variable, ICCF rmax (soft–hard) 0.748 0.942 7(2.3) 8(3.7) c ICCF τcent (95% limit, d) < 0.07 < 0.02 7(2.3) 8(3.7) Γ (Mean Spectrum) 2.44 ±0.02 2.538 ± 0.005 7(2.4) 8(3.4) Γmin–Γmax 2.38–2.62 2.45–2.72 7(2.5.2) 8(3.5.2) ∆Γ 0.24 0.27 7(2.5.2) 8(3.5.2) PL soft band contribution 72% 75% 7(2.4.1) 8(3.3.1) Soft Hump Fits: E (keV) 0.17 ± 0.17 0.57 ± 0.2 7(2.4.1) 8(3.4.1) +0.05 ± FWHM (keV) 1.01−0.12 0.36 0.01 7 8 n −2 −1 −2 +0.37 +0.12 (10 ph s cm )1.50−0.61 1.25−0.17 78 +23 +11 EW (eV) 94−38 110−15 78 Table 3.2: Continued at page 98. 97 Ton S180a Ark 564a Referencesb (1) (2) (3) (4) (5) α E d . +0.27 . +0.00 K diskline Fits: (keV) 6 40−0.00p 7 00−0.13 7(2.4.2) 8(3.4.2) +120 ± EW (eV) 461−84 351 85 7 8 i +22 ± (deg) 35−35p 26 278 α E e . +0.16 . +0.01p K laor Fits: (keV) 6 55−0.15p 6 99−0.13 7(2.4.2) 8(3.4.2) +123 ± EW (eV) 517−111 653 85 7 8 i +14 +11 (deg) 23−23p 17−17 78 Table 3.2: Continued. Comparison of the Properties of Ton S180 and Ark 564. Notes: (a) 90 % confidence level uncertainties. (b) The numbers in parenthesis refer to the sections where the relevant information is found in the given references. (c) Lag of hard relative to soft flux. (d) Best parametrization for Ton S180 is by E . +0.08 E . +0.25 two Gaussians with peak energy N =681−0.12 keV and B =658−0.28 keV. (e) Best-fit model for Ark 564 (Chapter 2, Turner et al. 2001a). References: 1: Wisotzki et al. (1995); 2: Huchra, Vogeley & Geller (1999); 3: Stark et al. (1992); 4: Dickey & Lockman (1990); 5: NED; 6: de Vaucouleurs et al. (1991); 7: Chapter 3 (Romano et al. 2002a); 8: Chapter 2 (Turner et al. 2001a). 98 Fig. 3.1.— Light curves for the ASCA data in cts s−1 and in 5760 s bins. The top panel is the SIS soft band (0.7–1.3 keV) light curve; the middle panel the GIS hard band (2–10 keV) and the bottom panel is the ratio of 0.7–1.3/2–10 keV. The background level in the source cell is about 4 % and 10 % of the SIS and GIS source count rates, respectively, and not plotted. The times are reported both in seconds from the start of exposure (top axis) and in JD-2450000 (bottom axis). 99 Fig. 3.2.— Same as Figure 3.1, for two “events” at JD ≈ 2451523. The time is seconds from the start of observation, the bin size is 256 s. In the first and second event, the SIS and GIS data show a variations up to a factor of 2 in ∆t = 1790 s; in the third event, the variation is of a factor of ∼ 1.3 in ∆t = 1536 s (§3.3). 100 Fig. 3.3.— Spectral and timing parameters obtained from fits to the individual time-resolved spectra. From the top, the light curves are the (model) continuum flux in the hard band, the photon index Γ, the soft hump flux, the Kα flux (laor), −11 −1 the fractional variability Fvar. The continuum fluxes are in units of 10 ergs s cm−2, the soft hump flux in units of 10−2 photons s−1 cm−2 and the Kα flux in units of 10−5 photons s−1 cm−2. 101 Fig. 3.4.— Top: Data/Model ratio where the model is a simple power law fit to the 1.8–4.71 keV data (observer’s frame). The 0.5–0.7 keV SIS data, though not used, are also plotted as empty circles. The crosses are the ASCA data from 1996 below 2 keV. The inset panel shows the ROSAT PSPC data compared to the continuum power law, illustrating the shape of the soft hump at lower energies (§3.4). Bottom: Data/Model ratio where the model is a power law plus soft hump (Gaussian; §3.4.1) plus 2-Gaussian model for the Fe Kα line (§3.4.2). 102 Fig. 3.5.— Combined SIS 0.7–1.3 keV light curve in cts s−1 and in 5760 s bins. The background level in the source cell is about 4 % of the source count rate, and not plotted or subtracted. The vertical dashed lines show our 14 time intervals within which spectra were extracted (§3.5). The “flares” shown in Figure 3.2 (§3.3) are within bin 9. 103 Fig. 3.6.— Ratio plots obtained by fitting the best-fit model for the first spectrum to the following 13 spectra. Energy ranges and instrument utilized are described in §3.5.2. Energies are in the rest frame of Ton S180. 104 Fig. 3.7.— Top panel: the ∆χ2 =2.3, 4.61, 9.21 contour levels for the soft hump flux (in units of photons s−1 cm−2) vs. photon index Γ. The full contours correspond to the hard state (§3.5.4), the dashed contours to the soft state, and crosses indicate the best-fit values. Bottom panel: the strength of the soft hump plotted against photon index Γ in our 14 time-selected spectra with high-state (open circles) and low-state (filled circles) points overlaid. 105 Fig. 3.8.— Panel a: the ∆χ2 =2.3, 4.61, 9.21 contour levels for Fe K-shell line intensity (in units of photons s−1 cm−2) vs. photon index Γ; the full contours correspond to the high state (§3.5.4), the dashed contours to the low state, and crosses indicate the best-fit values. Panel b: the Fe Kα regime compared to the continuum model; with overlay of high-state (open circles) and low states (filled circles). Panel c: same as a, with full contours corresponding to the hard state (§3.5.3), dashed contours to the soft state. Panel d: same as b, with hard-state (open circles) and soft-state (filled circles) overlaid. Panel e: same as a, with full contours corresponding to the high Kα flux bin 4 (§3.5.4), the dashed contours to the low Kα flux bin 13. Panel f: same as b, with bin 4 (open circles) and bin 13 (filled circles) overlaid. 106 Fig. 3.9.— The ratio of RMS/Mean spectrum for SIS1 and SIS0, using data of energy > 0.7 keV. The “spike-like” features at ∼ 1.7–2.5 keV are probably due to calibration uncertainties. 107 Chapter 4 FUSE Observations of Arakelian 564: Properties of the Warm UV–X-ray Absorber 4.1. Introduction In this Chapter we present the results from a 63 ks FUSE observation of Ark 564 obtained on 2001 June 29–30, focusing in particular on the O VI intrinsic absorption; we investigate the physical properties of the UV and X-ray absorbing gas using the constraints on column densities obtained during the multiwavelength observations of this AGN. In §4.2 we present the data. In §4.3 we describe our analysis methods. In §4.4 we test the hypothesis that the Warm UV-X-ray absorber are one and the same through photoionization calculations. In §4.5 we discuss some implications of our investigation. Our results are summarized in §4.6. In Chapter 6 we will analyze the intrinsic SED of Ark 564. 108 4.2. Observations and Data Reduction We observed Ark 564 with FUSE (Moos et al. 2000; Sahnow et al. 2000) for 63 ks starting on 2001 June 29 07:37:42 UT. The observations, consisting of 24 separate exposures, were performed in photon address (time-tag) mode through the 30 × 30 low-resolution (LWRS) aperture. During our observation, a high-voltage anomaly occurred, and Detector 1 did not collect useful data during the first 8 exposures. To best model the background for this observation, we first used the task ttag combine provided with the FUSE calibration pipeline, CalFUSE (version 2.0.5)1, to combine the last 16 exposures for Detector 1 (total exposure of 42 ks), and all 24 exposures for Detector 2. We then processed the combined exposures with the standard pipeline and extracted spectra from both detectors. The flux scale for the final spectra is accurate to ±10 %, while the wavelength scale is accurate to ±15 km s−1. As a result of the high-voltage anomaly and data screening, the effective on-source times were 41 ks in Detector 1A, 39 ks in Detector 1B, and 58 ks in Detector 2A, and 62 ks in Detector 2B. Consequently, the SiC1A and SiC1B spectra were discarded from further analysis, the final S/N being ∼< 1.5evenat0.6A˚ (100 pixels) resolution. We also discarded the LiF1B spectrum since it showed wavelength-dependent differences in flux of up to 1See http://fuse.pha.jhu.edu/analysis/calfuse.html. 109 30–50 % compared to the LiF1A probably due to the the “worm”, which cannot be corrected for by the pipeline (Oegerle, Murphy, & Kriss 2000). The full FUSE spectrum was obtained by combining the spectra extracted from the SiC2A, LiF2B, LiF1A, SiC2B, and LiF2A segments, yielding a wavelength coverage of 916-1175 A.˚ We then rebinned the full spectrum in a linear wavelength scale using 0.07 A˚ bins (10 pixels, here on our full-resolution spectrum, with an effective resolution of 20 km s−1), 0.2 A˚ bins (30 pixels, medium-resolution spectrum), and 0.6A˚ bins (100 pixels, low-resolution spectrum). Figure 4.1 shows the low-resolution spectrum after we cosmetically removed the strong airglow lines (mainly H I Lyman series, O I λ989, O I λ1027, and He I λ584 seen in second order at 1167A)˚ and C III λ977, which is scattered solar light in the SiC detector. The main spectral features are identified, the most prominent being the emission lines of the O VI λλ1032, 1038 resonance doublet. Strong absorption features due to Lyβ and O VI λλ1032, 1038 at velocities near the redshift of Ark 564 are also observed. We detect absorption from Galactic ISM molecular Hydrogen, mainly H2 Lyman series absorption (see Shull et al. 2000, Sembach et al. 2000, and Savage et al. 2000), and atomic Galactic ISM lines, including O VI λλ1032, 1038 (Mathur et al. 2002, in preparation). No intrinsic Lyman edge is detected. 110 4.3. Data analysis Our goal was to determine the column densities of the ionic species we observed in our spectrum, combine this information with the column densities available in the literature for Ark 564, and derive constraints on the physical parameters of the absorber (total density and ionization) through photoionization modeling. HST/FOS spectra of Ark 564 (Crenshaw et al. 1999) show the presence of strong intrinsic absorption lines of Lyα,NVλλ1238.8, 1242.8, Si IVλλ1393.8, 1402.8, and C IVλλ1548.2, 1550.8. Of these lines, which are resolved in STIS spectra into multiple components (Collier et al. 2001, hereafter Paper II, and Crenshaw et al. 2002, hereafter Paper IV), Lyα,NV,andCIV are completely saturated. Figure 4.2 shows the FUSE full-resolution (0.07 A,˚ 10 pixels) spectrum of Ark 564, in the Lyβ/O VI wavelength region. Close examination of the O VI troughs shows that the lines are heavily saturated, and their shape is mainly determined by partial covering effects (see §4.3.2 and Figure 4.3 below). Therefore, the absorption lines are not resolved into components at different velocities with respect to the systemic velocity, and we must treat each of the absorption troughs as a single absorption component, and we can only determine the velocity-averaged column densities of the observed species. We note, however, that O VIλ1032 is not completely black. Indeed, analysis of the spectra obtained with different pulse height restrictions and from 111 night-only data (we did not use the latter for this work, since the lower S/N did not allow a proper subtraction of Galactic molecular Hydrogen) shows that scattered light is marginal in this observation, and that there is no filling in of the absorption troughs. It is also clear that the uncertainty in the measurement of O VI absorption line parameters, hence O VI column density, is dominated by the uncertainty in the underlying emission-line profile. Additionally, there is contamination from absorption lines of Galactic molecular Hydrogen, with a column density log N(H2) ∼> 16 (K. R. Sembach 2002, private communication); this is not surprising, given the substantial amount of neutral atomic hydrogen 20 −2 (NH =6.4 × 10 cm , Dickey & Lockman 1990) along the line of sight toward Ark 564. Given these limitations, we proceeded as follows: we determined a power-law continuum underlying the Lyβ/O VI wavelength region using the low-resolution spectrum (S/N ∼< 15 in the continuum); we then modeled the Lyβ and O VI emission lines from the high- and medium-resolution spectra (S/N ∼< 10 in the emission lines for high-resolution), and used H2 templates to estimate the H2 contribution to the O VI absorption troughs. This part of the analysis was done using the IRAF2 task specfit (Kriss 1994) in the STSDAS package. Finally, we measured the absorption line parameters of the normalized line profiles using 2IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. 112 the high-resolution spectrum, and determined the column densities with the apparent optical depth method, which we briefly describe below (§4.3.2). 4.3.1. Intrinsic O VI Emission Models In addition to the power-law continuum (that we kept fixed relative to the fit of the low-resolution spectrum) our models for the adopted “continuum” under the absorption troughs included a pair of broad O VI emission lines (FWHM = 4000–5000 km s−1), a pair of narrow O VI emission lines (FWHM = 1000–1100 km s−1), and a broad and narrow Lyβ emission lines. All lines are taken to have Gaussian profiles. The intensity of the O VI doublet lines was fixed to the optically thin value 2:1 for both broad and narrow lines, while their wavelengths were linked to the ratio of their laboratory values; the FWHM and wavelength of the broad Lyβ line were linked to those of the O VI lines, though a small shift in wavelength was permitted, consistent with increasing blueshift with respect to the systemic velocity as the ionization increases. To assess the possible range of absorption-line parameters, we considered five different models for the emission lines, shown in Figure 4.2, namely, low-lying (Low), medium (Med1 and Med2), and high-lying (High, that best follows the O VI λ1038 peak, though it clearly over-predicts the O VI λ1032 peak). In all four cases the Gaussian profiles for the emission lines were symmetrical. As observed 113 in many AGNs (e.g. Marziani et al. 1996), the emission lines in all our models are blueshifted with respect to the systemic redshift ze =0.02467 as derived from H I measurements (de Vaucouleurs et al. 1991) by 390–1240 km s−1. Finally, we considered a model in which the broad lines have the least blueshift (100 km s−1) with respect to the systemic redshift, and the narrow lines are at the systemic redshift; in this case, in order to model the profile, the narrow emission lines must be highly asymmetrical (skewness = 0.2). The motivation for considering the last model is the increasing evidence that the high-ionization emission lines in NLS1s are broader and present an excess of flux in the blue with respect to the low-ionization lines (Laor et al. 1997b; Peterson et al. 2000; Mathur 2000b; Leighly 2001, and references therein). With our choice of Low, Med1, Med2, and High emission line profiles the absorption is redshifted with respect to the emission lines. This is highly unusual, though not unprecedented (Mathur, Elvis, & Wilkes 1999; Goodrich 2000). When modeling the emission with the Skew profile, the absorption is in part blueshifted, in part redshifted with respect to the emission lines. Table 4.1 lists the O VI model emission-line parameters. Since no inflection points are clearly seen in the observed emission line profiles, our models may not have a direct physical interpretation. Also, while we believe the true shape of the emission line profile may be most realistically represented by the Med1 and Med2 profiles, we consider the most extreme profiles (Low, High and Skew) to bracket the constraints on the values of column density. 114 Our spectrum also shows narrow H2 and H I absorption-lines from the inter-stellar medium (ISM) in the Lyβ/O VI wavelength region; in particular, the top panel of Figure 4.3 shows six of such absorption lines that lie outside the absorption troughs, which we fit as Gaussians with specfit (FWHM = 30 km −1 s ). We then used H2 and H I absorption-line templates (S. McCandliss 2001, private communication; K. R. Sembach 2002, private communication) to predict the position of the lines in the absorption troughs and derive their intensity by scaling them to the two lines at ∼ 1065 A.˚ Incidentally, we note a 0.115 A˚ shift between the predicted and observed wavelengths of H2 absorption lines. The amount of this shift is constant along our spectrum, hence we interpreted it as a residual zero-point offset in the wavelength calibration of our spectrum; as reported in many observations3, these offsets can be as high as 0.25 Ainthe˚ LWRS aperture. We matched templates and spectra accordingly. Figure 4.3 shows the line profiles normalized with respect to the combined continuum, emission-line, and H2 profiles, for the five different O VI emission-line models. 4.3.2. Intrinsic O VI Absorption Measurements As mentioned above, the O VI lines are so heavily saturated that we must treat each of the absorption troughs as a single absorption component (as opposed to the many components observed in Si IV and Si IIIλ1206.5; Paper IV). This 3 http://fuse.pha.jhu.edu/analysis/calfuse wp1.html. 115 assumption is only strictly valid if the physical conditions are approximately constant along the line profile, i.e., as a function of radial velocity. We consired the possibility of partial covering of the lines and used the apparent optical depth method to determine the column densities. Following Hamann et al. (1997), we calculated the lower limit to the line-of-sight covering factor Cf from the residual intensities Ir in the troughs as a function of radial velocity, and the corresponding apparent optical depth τ, Ir Cf ≥ 1 − , (4.1) I0 I τ ≥ ln 0 , (4.2) Ir where I0 is the assumed continuum intensity across the absorption line (in the case of O VI, I0 is our combined power-law continuum and the emission line models corrected for the Galactic Hydrogen absorption as described in §4.3.1). Column densities are then obtained by integrating the apparent optical depth across the line profile using (e.g., Savage & Sembach 1991) m c N ≥ e τ(v) dv (4.3) ion πe2fλ where λ and f are the laboratory wavelength and oscillator strength of the transition, respectively. 116 Figure 4.4 shows the normalized line profiles, the covering factor, and the optical depth as a function of radial velocity relative to the systemic redshift ze =0.02467 for the Med1 emission line profile. Predictably, Cf ≈ 1 for most of the O VI λ1038 profile and is consistent with unity for O VI λ1032. The arrows show the position of the zero velocity with respect to the broad emission lines (BEL) and narrow emission lines (NEL). As noted in §4.3.1, while most of the absorption is blueshifted with respect to the systemic redshift, the absorption troughs are completely or at least partially (Skew model) redshifted with respect to the broad and narrow emission lines. The absorption troughs also show the presence of gas which is redshifted with respect to the systemic velocity. This may indicate that the absorbing gas is undergoing net radial infall, as is the case for NGC 5548 (Mathur, Elvis, & Wilkes 1999) and RX J0134-42 (Goodrich 2000). We used log fλ =2.137 for O VI λ1032 and log fλ =1.836 for O VI λ1038 (Morton 1991). Table 4.2 reports the values of O VI column densities 15 −2 for our five assumed emission line models; NOVI =[2.31, 2.65] × 10 cm and [5.28, 5.96] × 1015 cm−2 when measured from O VI λ1032 and O VI λ1038, respectively. The errors quoted in Table 4.2 are relative to the measurement of the integral of τ in velocity space only. We estimate that molecular Hydrogen lines contribute ∼ 10 % to the flux in the absorption troughs. We adopt (5.7 ± 0.07) × 1015 cm−2, obtained averaging the values from Med1 and Med2 emission 117 line models for the O VI λ1038 line, as a conservative lower limit on the O VI column density. 4.3.3. C III column density We also determined the column density of C III from the C III λ977 absorption trough with the apparent optical depth method described in §4.3.2 (log fλ =2.872, Morton 1991). The Lyγ/C III wavelength region does not necessarily require the same degree of complication in the emission line profile as the Lyβ/O VI wavelength region; however, though we performed fits and measurements separate from the Lyβ/O VI ones, for consistency we adopted the same, albeit with different flux normalization, emission model as Med2: one broad (FWHM = 4000 km s−1) and one narrow (FWHM = 1100 km s−1) Gaussian emission line for C III and one broad Lyγ Gaussian emission line; the FWHM and wavelength of the broad Lyγ line were linked to the ones of the broad C III line. We obtain a C III column density of (3.19 ± 0.05) × 1014 cm−2 (errors are relative to the measurement of the integral of τ in velocity space only, while a 10 % contribution is due to molecular Hydrogen absorption lines). 118 4.3.4. Velocity Centroids Table 4.3 reports the values of the radial velocity centroids relative to the systemic redshift of O VI,CIII,Lyβ,andLyγ absorption lines, along with the measured column densities. We also show the results of Paper IV to emphasize the good agreement between the centroid velocity shifts (relative to systemic redshift) obtained for the H I LymanseriesintheFUSE and HST spectra. In Paper IV, it is also noted that while saturation is probably responsible for the discrepancy in the values of the centroids in the different ions, the most saturated lines, i.e., the H I Lyman series, give us an estimate of the total coverage of the absorber. Paper IV reports a range [−420, +180] km s−1 for Lyα,andweobtain [−431, +174] km s−1 for Lyβ,[−412, +147] km s−1 for Lyγ,[−395, +177] km s−1 for C III,and[−412, +130] km s−1 for O VI λ1038. The less saturated O VI λ1032 yields [−374, +142] km s−1. Our FUSE spectrum was obtained a year after the last of the HST/STIS spectra were taken, and we confirm the finding of Paper IV that there are no changes in radial velocity coverage of the absorber. 4.4. Photoionization Modeling It is common practice to use photoionization codes to predict the fractional abundance of an element in a given ionization state, fion, given an input continuum, density n, total column density NH, and ionization parameter U of 119 the gas. The fractional abundance of an ion of an element X is related to its column density Nion and the abundance of its element NX by Nion = NH NX fion, which provides a prediction of Nion that can be tested against observations. 4 We used Cloudy (v94.00, Ferland 1996) to calculate fion for the ionic species for which we measured column densities from the FUSE spectrum, and for the species with published column densities, which are reported for easy reference in Table 4.3. We considered a range of values of NH for a range of input continua (described in detail in §4.4.1), a total Hydrogen density of 105 cm−3, and assumed solar abundances relative to Hydrogen. In the case of “table agn” (see §4.4.1), we specified a grid of NH and U values. For all other models, we normalized the SEDs with respect to the measured X-ray luminosity in the absorption-corrected 43 −1 rest-frame 2–10 keV energy range (L2–10 keV =2.4 × 10 ergs s ; Chapter 2), and specified the radius of the cloud, thus obtaining U. We note that the use of the observed SEDs assumes that the absorbing gas sees the same ionizing continuum as the observer does. 4.4.1. Input continua Figure 4.5 illustrates our choices of input continua for Cloudy. 4http://www.pa.uky.edu/∼gary/cloudy/. 120 1. The Cloudy “table agn” continuum, which is the Mathews & Ferland (1987) continuum modified with a sub-millimeter break at 10 µm, so that the −α spectral index is changed from −1to−5/2 (specific flux Fν ∝ ν ) for frequencies below the millimeter break. While “table agn” is unlikely to be a representative spectral energy distribution (SED) of Seyferts, we use this continuum for comparison with the literature. 2. A combination of the SED described in Laor et al. (1997a) and Zheng et al. (1997) for radio-quiet objects (LZ in Figure 4.5); we have extended the original SED in Laor et al. (1997a) to cover the whole 10−5–7.354 × 106 Ryd energy range as required by Cloudy. At the low energies, we defined a sub-millimiter break and at the high energies, a break at 100 keV (with a spectral index of −5/3), analogous to the ones in “table agn”. This “composite” continuum might be a typical AGN continuum. 3. Observed SEDs: we used data obtained during the multiwavelength monitoring campaign performed in 2000, that included simultaneous observations of Ark 564 from ASCA (Chapter 2), HST (Paper II), and from many ground-based observatories (Paper III). In addition, Infrared Astronomical Satellite5 (IRAS) measurements (Moshir et al. 1990) and our FUSE observations have been used. While the full extent of the data is used 5 IRAS was a joint US/UK/the Netherlands IR satellite, launched on 1983 January 25 and was operated for 10 months. 121 to create a quasi-simultaneous SED (Chapter 6), our adopted continuum for Cloudy only consists of selected points (also shown in Figure 4.5). All data have been corrected for redshift. Special care has been paid in correcting the data for reddening, given the indications (Paper IV) that strong intrinsic neutral absorption is present in Ark 564 in excess of the Galactic absorption. Therefore, we corrected the spectra for reddening in 2 different ways: (a) Using a standard Galactic extinction curve with E(B − V )=0.06 mag (Schlegel, Finkbeiner, & Davis 1998) (SED1 in Figure 4.5); (b) Using a standard Galactic extinction curve with E(B − V )=0.03 mag plus the intrinsic extinction curve Crenshaw et al. (2002) (Paper IV) derive for Ark 564 and E(B − V )=0.14 mag (SED2 in Figure 4.5). In the FUSE band we extrapolated the extinction correction linearly from the one relative to the HST band. We also note that to match our FUSE spectrum and HST spectrum in the overlapping region, we had to scale the FUSE fluxes by 0.75. This is not inconsistent with a combination of effects such as flux intercalibration uncertainties and, most importantly, source flux variability. In the X-ray, we used continuum points from the power-law fit (photon index Γ = α+1 = 2.538) and added a black body component of temperature 122 6 38 −1 T =1.8 × 10 K and luminosity Lbb =2.48 × 10 ergs s , as derived from fits to the mean ASCA spectrum. The square points in Figure 4.5 denote the adopted points for SED1, the circles for SED2, while the horizontal solid lines show the full ranges where data were available. The full details of the observed SED are presented in Chapter 6. 4.4.2. Physical conditions of the UV/X-ray absorber Following Arav et al. (2001) we constrain the characteristics of the absorber by plotting curves of constant Nion on the log U–log NH plane. In this plane, for each constant Nion curve, lower limits on the column densities, derived from apparent optical depth line fitting methods, exclude the area below it, while upper limits, derived from non-detections, exclude the area above it. Figure 4.6 shows the Nion constraints (see Table 4.3) for the “table AGN” input continuum, and solar abundances. Lower limits are shown as solid lines, upper limits as dashed lines, detections as dotted lines. The combination of constraints given by the column densities suggests that the absorber in Ark 564 is characterized by a narrow range in Nion and U, i.e., log U =[−1.74, −0.74] and log NH =[19.90, 21.89]. We note the consistency of all constraints without departure from solar abundances. Analogously, Figure 4.7 shows the Nion constraints for our SED1 input continuum and indicates 123 log U =[−1.99, −1.31] and log NH =[19.99, 21.09], Figure 4.8 (SED2 continuum) indicates log U =[−1.86, −1.02] and log NH =[19.95, 21.27], while Figure 4.9 (LZ continuum) indicates log U =[−1.97, −1.54] and log NH =[20.00, 20.79]. Thus, depending on the input continuum, there are small but significant differences in the derived properties of the absorber. 4.5. Discussion The UV absorber in Ark 564 is in a general state of outflow with respect to the systemic redshift (see Table 4.3). A very good agreement is found between the values of the velocity centroids we derive for the species observed in the FUSE spectrum and the ones derived for the HST/STIS spectrum (Paper IV); therefore, we adopt as the best estimate of the net radial velocity of the UV absorber the value obtained in Paper IV for Si III and Si IV, the least saturated lines: −1 Vout = −194 ± 5kms . The absorption troughs also show the presence of gas which is redshifted with respect to the systemic velocity. This can be explained in part as a saturation effect, as is shown in Figure 3 of Paper IV. Alternatively, a model with more than one kinematic component is required to explain the observed absorption troughs (i.e. Elvis 2000); in this scenario, in addition to the blueshifted absorption from an outflowing wind, we would be observing redshifted absorption from infalling material, such as an accretion flow. In addition to the 124 continuum source, the absorbing gas must cover a substantial portion of the BELR, since the absorption troughs are much deeper than the continuum level. Assuming the identity of the UV and X-ray absorbing gas, we have used the column densities of the observed species to constrain the physical conditions of the absorber. For the most realistic SED (SED2), we obtained log NH =[19.95, 21.27] and log U =[−1.86, −1.02]. These constraints can be used to determine the size of the absorber, its distance from the central continuum source, and the mass outflow rate. For the following order-of-magnitude arguments we adopt the mean 20 −2 values U =0.0363 and NH =4.07 × 10 cm . The size of the absorber is derived r . × 20 n−1 from the total column density, abs =407 10 H cm. For SED2 the number of ionizing photons is Q =6.68 × 1055 s−1, so the distance from the continuum R . × 22 n−1/2 R > source is abs =700 10 H cm. Using the lower limit on abs 95 pc found 4 −3 in Paper IV, this would indicate a total density nH > 5.70 × 10 cm . Assuming uniform density, and considering that rabs Rabs, the mass of the outflowing gas M . × 10 fn−1 M f is abs =211 10 H ,where is the covering factor, i.e., the fraction of the sky covered by the absorber as seen at the central source. The mass outflow ˙ 4 −1 rate is then Mabs = Mabs Vout /rabs =3.17 × 10 fM yr and the outflow M˙ V 2 / . × 44 f −1 carries out a rate of kinetic energy abs out 2=376 10 ergs s . To power 44 −1 Ark 564 at the observed luminosity (Lbol =10× L2–10 keV =2.4 × 10 ergs s )at ˙ −3 −2 an efficiency η =0.1, an accretion rate Macc =1.8 × 10 (L44/η)=4.3 × 10 M −1 44 −1 yr is required (L44 is the bolometric luminosity in units of 10 ergs s ). 125 The outflow carries out a kinetic luminosity about one order of magnitude smaller than the observed radiative luminosity of the source. However, the mass outflow rate is uncomfortably large unless the covering factor is very small. If ˙ ˙ −6 Mabs ∼< Macc, then it implies f ∼< 10 . Alternatively, our assumption rabs Rabs might not be valid. The absorber might be an extended, low density region. The assumption of a uniform density gas may not be strictly valid and the ionization parameter and density that we deduced should only be considered as “average” values. Recent Chandra observations have found extended warm gas in some AGN (Sako et al. 2000) with physical characteristics similar to that of a warm absorber, but seen in emission. So it is quite likely that the warm absorber in Ark 564 is also spatially extended along the line of sight. In Paper IV the UV absorber was modeled as a single zone with quasi-solar abundances (Carbon depletion being the main departure) and the best-fit values of log U = −1.48 and log NH =21.21 are consistent with our limits. Crenshaw et al. (2002, Paper IV) over-predicted Carbon and Oxygen column densities: 15 −2 17 −2 NC III =5.2 × 10 cm , NOVI =2.4 × 10 cm (cf. our measurements: 14 −2 15 −2 NC III =3.2 × 10 cm and NOVI =5.7 × 10 cm ). These predictions, however, are consistent with the upper-end values of our range of parameter space. We note that our modeling did not require Carbon to be depleted. 126 Finally, we can compare our solutions of log NH =21andlogU = −1.5, with the preliminary results of Matsumoto, Leighly, & Marshall (2001) based on analysis of a 50 ks Chandra observation of Ark 564. Their curve of growth analysis 17 −2 18 on the absorption lines indicates that NOVII =3.2 × 10 cm , NO VIII =1× 10 −2 17 −2 17 −2 cm , NNe IX =3.2 × 10 cm ,andNNe X =1× 10 cm , suggestive of log NH =21andlogξ =1.6–2. While there is agreement between the values of log NH and NOVII, the column densities they measure for O VII do not agree with the ones derived in Paper IV from the upper limits on the bound-free optical depths in the ASCA spectrum (Chapter 2). Given the high NO VIII we would expect an edge would be observable in the Chandra spectrum. 4.6. Summary We have presented a 63 ks FUSE observation of the NLS1 galaxy Ark 564. The observed spectrum is dominated by the O VI λλ1032, 1038 emission lines. As observed in many AGNs (e.g. Marziani et al. 1996), the emission lines in all our models are blueshifted (or at rest, as in the case of our Skew blue-asymmetric model) with respect to the systemic redshift by 100–1240 km s−1. Blue-asymmetric UV emission line profiles may be a characteristic of NLS1 galaxies. We concentrated on the analysis of the strong and heavily saturated absorption troughs due to Lyman series, O VI and C III λ977, which are 127 observed at velocities near the systemic redshift of Ark 564. In Chapter 6 we will analyze the intrinsic SED of Ark 564. Using the apparent optical depth method, we have determined that the column density of O VI is a few 1015 cm−2, 14 −2 and NC III =3.2 × 10 cm . We used these values in conjunction with the published column densities of species observed in the UV and X-ray spectra of this object to derive constraints on the physical parameters of the UV/X-ray absorbing gas through photoionization modeling. The combination of constraints, assuming the most realistic SED, indicates that the absorber is characterized by a narrow range of density and ionization parameter, log NH =[19.95, 21.27] and log U =[−1.86, −1.02]. There is excellent agreement between the kinematic properties of the UV absorber emerging from the combined analysis of the FUSE and HST/STIS spectra, i.e. distribution of gas in radial velocity (as derived from the extent of the absorption troughs) and net radial velocity (as derived from the velocity centroids). The UV/X-ray absorber in Ark 564 is in outflow with respect to the systemic redshift with a radial velocity of a few hundred km s−1, and it is likely spatially extended along the line of sight. The absorption troughs also show the presence of gas which is redshifted with respect to the systemic velocity. This may indicate that a component in the absorbing gas is undergoing net radial infall. This is highly unusual, though not unprecedented (Mathur, Elvis, & Wilkes 1999; Goodrich 2000). 128 Model Line λa Fluxa FWHM ∆V b (A)˚ (10−13 erg s−1 cm−2)(kms−1)(kms−1) (1) (2) (3) (4) (5) (6) Low BEL O VI λ1032 1053. 1.00 5000. −1240 BEL O VI λ1038 1058.82 0.50 5000. −1240 NEL O VI λ1032 1056. 0.90 1100. −390 NEL O VI λ1038 1061.83 0.45 1100. −390 Med1 BEL O VI λ1032 1055. 1.00 4000. −670 BEL O VI λ1038 1060.83 0.50 4000. −670 NEL O VI λ1032 1056. 1.00 1100. −390 NEL O VI λ1038 1061.83 0.50 1100. −390 Med2 BEL O VI λ1032 1056. 1.00 4000. −390 BEL O VI λ1038 1061.83 0.50 4000. −390 NEL O VI λ1032 1056.3 1.00 1100. −300 NEL O VI λ1038 1062.13 0.50 1100. −300 High BEL O VI λ1032 1055. 1.00 4000. −670 BEL O VI λ1038 1060.83 0.50 4000. −670 NEL O VI λ1032 1056. 1.25 1000. −390 NEL O VI λ1038 1061.83 0.63 1000. −390 Skew BEL O VI λ1032 1057. 1.25 2000. −100 BEL O VI λ1038 1062.84 0.63 2000. −100 NEL O VI λ1032 1057.37 0.60 1000. 0. NEL O VI λ1038 1063.21 0.30 1000. 0. Table 4.1: O VI Model Emission-Line Parameters. Note: (a) Observed values. (b) Velocities are relative to the systemic redshift ze =0.02467 (H I measurements, de Vaucouleurs et al. 1991). The shift toward longer wavelengths of 0.115 A˚ to match the Galactic molecular Hydrogen templates is not included. 129 Line Low Med1 Med2 High Skew (1) (2) (3) (4) (5) (6) O VI λ1032 2.31 ± 0.03 2.47 ± 0.05 2.54 ± 0.03 2.65 ± 0.03 2.80 ± 0.03 130 O VI λ1038 5.28 ± 0.05 5.60 ± 0.05 5.77 ± 0.05 5.96 ± 0.05 5.64 ± 0.05 Table 4.2: O VI Column Densities from Intrinsic Absorption in units of 1015 cm−2. The errors quoted are relative to the measurement of the integral of τ in velocity space only. We estimate that molecular Hydrogen lines contributes ∼ 10 % to the flux in the absorption troughs. Ion Wavelength/ Lower Limit Detection Upper Limit Velocity Ref. Energy (cm−2)(cm−2)(cm−2)(kms−1) (1) (2) (3) (4) (5) (6) (7) H I 973 A˚ ··· ··· ··· −108 1 C III 977.0 A3˚ .2 × 1014 ··· ··· −153 1 H I 1025 A˚ ··· ··· ··· −111 1 O VIa 1031.9,1037.6 A5˚ .7 × 1015 ··· ··· −79/ − 116 1 Si III 1206.5 A˚ ··· 2.6 × 1013 ··· −190 2 H I 1216 A1˚ .4 × 1015 ··· ··· −106 2 131 N V 1238.8,1242.8 A3˚ .1 × 1015 ··· ··· −152 2 Si II 1260.4 A˚ ··· ··· 7.4 × 1013 ··· 2 C II 1334.5 A˚ ··· ··· 5.4 × 1013 ··· 2 Si IV 1393.8,1402.8 A˚ ··· 1.6 × 1014 ··· −197 2 C IV 1548.2,1550.8 A2˚ .5 × 1015 ··· ··· −130 2 O VIIb 0.74 keV ··· ··· 2.2 × 1017 ··· 2 O VIIIb 0.87 keV ··· ··· 1.1 × 1016 ··· 2 Table 4.3: Column Densities from Intrinsic Absorption in Arakelian 564. Notes: (a) Average of values obtained from Med1 and Med2 emission line models. (b) Absorption edge. References: (1) This work. (2) Crenshaw et al. (2002). Fig. 4.1.— FUSE spectrum of Ark 564, binned to a resolution of 0.6 A˚ (100 pixels). In addition to the prominent emission lines from the O VI λλ1032, 1038 resonance doublet, we suggest identifications for the main emission and absorption features. The strong airglow lines (H I Lyman series, O I λ989, O I λ1027, and He I λ584 seen in second order at 1167A),˚ and the C III λ977 (which is scattered solar light in the SiC detector) have been removed. All other absorption features are due to Galactic or intergalactic absorption (indicated with “IS”). The absorption feature at ∼ 1080A˚ is partially due to a gap between detectors. 132 Fig. 4.2.— Full-resolution (0.07 A,˚ ∼ 10 pixels) spectrum of Ark 564, in the Lyβ/O VI wavelength region. The dashed vertical lines mark the rest-frame wavelengths of the O VI lines and the ticks mark Galactic absorption lines. Overalid are our five adopted models for the combined continuum and emission lines, low-lying (Low), medium (Med1 and Med2), and high-lying (High, that best follows the O VI λ1038 peak), and a model with skewed Gaussian emission lines (Skew). All absorption lines are saturated, but while Lyβ and O VI λ1038 are black, O VI λ1032 is not. The mean 1-σ errorbar on the spectrum is also shown for reference. 133 Fig. 4.3.— Top: the full-resolution spectrum of Ark 564, in the Lyβ/O VI wavelength region, with a model for the continuum, the emission lines (Med1) and molecular Hydrogen absorption lines overlaid as a solid thicker curve (log N(H2) ∼ 1016). The other panels show the normalized line profiles of the absorption system at za ≈ ze. The dashed vertical lines mark the rest-frame wavelengths of the O VI lines. 134 Fig. 4.4.— Top: normalized line profiles (Med1) of the absorption system as a function of radial velocity relative to a systemic redshift ze =0.02467 (H I). The arrows show the position of the zero velocity with respect to the broad emission lines (BEL) and narrow emission lines (NEL). Middle: covering factor as a function of radial velocity. Bottom: optical depth as a function of radial velocity. The solid lines refers to the O VI λ1032 profile, the dotted line to the the O VI λ1038 profile. 135 Fig. 4.5.— Comparison of the adopted SEDs for Ark 564 normalized to the absorption-corrected rest-frame flux at 2 keV. The square points denote adopted points for SED1, the circles for SED2, while the horizontal solid lines show the full ranges where data were available. 136 Fig. 4.6.— Photoionization curves at constant ionic column density on the plane of total hydrogen column density NH vs. ionization parameter U. The shape of the ionizing radiation is defined by the “table AGN” model in Cloudy, and the abundances are solar. Lower limits are shown as solid lines, upper limits as dashed lines, detections as dotted lines. For clarity, curves relative to ions of the same element have been drawn in the same color: C in red, O in light blue, Si in dark blue. The shaded region corresponds to the locus on the log U–log NH space where all conditions are met (see §4.4.2), that is, where log U =[−1.74, −0.74], log NH =[19.90, 21.89]. 137 Fig. 4.7.— Same as Fig. 4.6, but with a ionizing continuum with no reddening intrinsic to Ark 564 (SED1 in Figure 4.5). Solar abundances are assumed. Most constraints are met in log U =[−1.99, −1.31], log NH =[19.99, 21.09] (see §4.4.2). 138 Fig. 4.8.— Same as Fig. 4.6, but with a ionizing continuum with Galactic and intrinsic reddening (SED2 in Figure 4.5). Solar abundances are assumed. Most constraints are met in log U =[−1.86, −1.02], log NH =[19.95, 21.27] (see §4.4.2). The top axis indicates the distance of the absorbing gas from the continuum source. 139 Fig. 4.9.— Same as Fig. 4.6, but with a ionizing continuum described in Laor et al. (1997a) and Zheng et al. (1997) (ZL in Figure 4.5). Solar abundances are assumed. Most constraints are met in log U =[−1.97, −1.54] and log NH =[20.00, 20.79] (see §4.4.2). 140 Chapter 5 The Spectral Energy Distribution of the Seyfert Galaxy Ton S180 5.1. Introduction As we have seen in Chapter 1, the examination of the SED of a NLS1, and comparison with that obtained for BLS1s should offer insight into the relative accretion rates across the Seyfert population. To this end, we undertook a multi-wavelength campaign to obtain a broad-band spectral-energy-distribution of Ton S180. We selected Ton S180 for study as it is bright in the soft X-ray regime and has low line-of-sight and intrinsic extinction, allowing a view of the bare continuum form. Ton S180 (PHL 912, z=0.06198; Wisotzki et al. 1995) has a low Galactic N . +0.27 × 20 −2 column density along the line-of-sight, H =155−0.13 10 cm (Dickey & Lockman 1990). The uncertainty on the column density represents the maximum 141 scatter of values of column density within a 1 degree cone centered on Ton S180. −11 The observed flux (i.e. no correction for extinction) is F0.5−2keV ≈ 1.1 × 10 ergs s−1 cm−2 (Turner, George, & Nandra 1998). The source is at the extreme end of the Seyfert range of line widths with FWHM Hα and Hβ ∼ 900 km s−1,making it a good choice for isolating the fundamental parameter which determines the classification of a Seyfert galaxy. BeppoSAX (Comastri et al. 1998) and ASCA (Turner, George, & Nandra 1998) data from Ton S180 indicated a steep spectrum in the 2–10 keV band, with α =1.5; both datasets also showed an Fe Kα emission line peaked at a rest-energy ∼ 7 keV, indicating that the circumnuclear material is strongly ionized. ASCA data confirm the complexity of the soft X-ray spectrum first noted in ROSAT PSPC data (Fink et al. 1997). A Chandra Low Energy Transmission Grating (LETG) observation has recently revealed the soft hump component to be a smooth continuum or extremely broadened reprocessed component (Turner et al. 2001b). In this Chapter we use energy index α for quantification of spectral indices, defined such that the flux density F (E) ∝ E−α at energy E.Alogofthe observations performed in support of the campaign is presented in Table 5.1. All the data were reduced using standard techniques as outlined in Turner et al. (2002) and below. 142 5.2. ASCA The reduction of the ASCA data has been described in 2.2. The results presented here are primarily based on the analysis of the time-averaged spectra obtained during the simultaneous ASCA–Chandra observations. This means that while all of the Chandra data were used, only a subset of the ASCA data were utilized. Figure 5.1 shows the periods covered by the FUSE and Chandra observations, with respect to the entire 12-day observation by ASCA, allowing us to see where these new datasets lie compared to the recent flux history of the source. In this paper, we use only the mean ASCA spectrum from the period simultaneous with the Chandra observation, i.e. within JD 2451526.576–2451527.498. As is evident from Figure 5.2, Ton S180 exhibited significant changes in flux. A detailed analysis of spectral variability over the full 12-day observation is presented in Chapter 3. We find the continuum fit to the mean spectrum to yield α =1.44 ± 0.02. A strong excess of emission is observed below 2 keV, and this soft component varies in strength down to the minimum timescale determinable via spectral analysis, ∼ 1 day. The variations in hump strength are correlated with the photon index and the 2-10 keV flux, consistent with disk-corona models (Chapter 3). The softness ratio shows rapid variability on timescales < 1000 s, indicating either a breakdown of the correlation between soft hump and power-law 143 fluxes on such short timescales, or rapid variations in the photon index. We also find a broad Fe Kα line with narrow peak at a rest-energy 6.8 keV, indicating an origin in ionized material. Analysis of the spectrum acquired simultaneously with Chandra data reveals α =1.44 ± 0.07 (in agreement with the mean for the full dataset). The soft +71 component shows an equivalent width EW = 63−50 eV when parameterized using a Gaussian model. As seen in Chapter 3, we found no evidence for variability in the flux or equivalent width of the Fe Kα line we do not fit for the Fe Kα parameters here (and exclude the 5.0-7.5 keV data when fitting for continuum slopes). 5.3. Chandra The data reduction of the Chandra data was described in Turner et al. (2001b) and Turner et al. (2002). The entire Chandra baseline is utilized in this analysis, as the ASCA observation overlaps the Chandra observation completely, as shown in Figure 5.2. Figure 5.2 also shows the light curve obtained from the ±1st-order Chandra/LETG data. The portion of the ASCA SIS light curve (from Figure 5.2) is overlaid for direct comparison. As might be expected, there is good agreement between the light curves from the two instruments. 144 Turner et al. (2001b) present the first order LETG spectra of Ton S180, finding no strong spectral features and concluding that the excess soft X-ray emission discovered using ASCA (Turner, George, & Nandra 1998) must be primarily due to a previously-unknown continuum component or very broadened reprocessed component. Turner et al. (2001b) note the lack of strong absorption features in the X-ray spectrum of Ton S180, in contrast to results from many Seyfert 1 galaxies (e.g. the BLS1s NGC 5548, Kaastra et al. 2000, and NGC 3783 Kaspi et al. 2000a, and the NLS1 NGC 4051, Collinge et al. 2001). Analysis using the new quantum efficiency (QE) file reveals an improvement to the agreement between ASCA and LETG data in the 1-2 keV regime (c.f. results presented by Turner et al. 2001b). We also re-examined the shape of the soft excess. The extrapolation of the hard-band power-law to soft X-ray energies reconfirms the presence of excess soft emission as expected, with a sharp turnover of the data below ∼ 0.3keV(∼ 7 × 1016 Hz). Figure 5.3 shows the form of the soft component. The turnover is sharper than that expected due to absorption by edges in neutral or ionized gas. In any case, if there were such a deep absorption edge in the X-ray regime, strong spectral features would be expected in other parts of the spectrum, which are not observed. The soft hump in the data was modeled using the XSPEC diskbb model (Mitsuda et al. 1984; Makishima et al. 1986). However, although diskbb has some intrinsic spectral curvature, this was not able to account for the shape of the LETG data. A fit to data above 0.3 keV 145 yielded a best-fit temperature of ∼ 98 eV at the inner edge of the disk. Possible explanations for the apparent shape of the soft excess include continuing problems with the softest energy calibration, and the more intriguing possibility of a peak due to the presence of a blend of broadened emission lines as suggested previously for some Seyfert galaxies (Branduardi-Raymont et al. 2001; Turner et al. 2001b). Comastri et al. (1998) present a BeppoSAX spectrum of Ton S180 down to ∼ 0.1 keV, with no evidence for a spectral drop below 0.3 keV, supporting the possibility that this is due to residual calibration problems in the LETG. Thus we do not perform detailed fitting to this feature. 5.4. EUVE Archival EUVE data are available covering a period which overlaps the Chandra observation, and their data reduction is detailed in Turner et al. (2001b). 5.5. FUSE FUSE observed Ton S180 on 1999 December 12, 05:50:32–19:41:14 UT, for a total observing time of 15.2 ks, through the LWRS apertures. Turner et al. (2002) describe the details of the data reduction. Figure 5.1 shows that while the FUSE 146 observation was not performed simultaneous with Chandra it does cover a similar mean flux state to that covered by the Chandra observation. As seen in Figure 5.4, the FUV spectrum of Ton S180 shows a bright, blue continuum and prominent broad O VI emission. Fainter emission from S VI λλ934, 945, C III λ977, N III λ991, and He II λ1085 may also be present. The foreground Galactic and intergalactic absorption visible in this spectrum has already been discussed by Savage et al. (2000), Sembach et al. (2000), and Shull et al. (2000); one noteworthy feature is the deep absorption by Lyβ. In addition to these foreground features, absorption at three velocities near the redshift of Ton S180 is visible in the O VI λλ1032,1038 resonance doublet (Figure 5.4). To measure the strengths of these features and that of the broad O VI emission, we used the IRAF task specfit (Kriss 1994). We used a power law for the underlying local continuum, a broad Gaussian for each of the O VI emission lines with their fluxes fixed at a 2:1 ratio, and Gaussians for the narrow absorption lines. The broad O VI lines have a FWHM of 2600 ± 186 km s−1 and a total flux of (1.10 ± 0.05) × 10−12 ergs s−1 cm−2. They are blueshifted relative to the systemic redshift by 490 ± 69 km s−1. Unfortunately, the Lyβ lines corresponding to these velocities fall in the gaps between the LiF detector segments, and so we must use the lower S/N SiC2A data to measure their strengths. No Lyβ lines are detected. 147 For the O VI absorption lines the measured equivalent widths (Wλ)column densities and FWHM are summarized in Table 5.2. The doublets have optical depths consistent with a 2:1 ratio, implying full coverage of the underlying continuum and broad emission lines. Given the strength of the O VI absorption and the weakness of any corresponding neutral hydrogen, we conclude that this gas is in a fairly high state of ionization. 5.6. HST As HST was in safe-mode during 1999 December, the earliest HST/STIS observations of Ton S180 could be obtained was 2000 January 22 (UT). The 52 x 0 .2 slit was used, and the exposure times were 1260 s for the G140L grating and 720 s for the G230L grating. A detailed description of the data reduction is given in Turner et al. (2002). Figure 5.5 shows the STIS spectrum of Ton S180. The underlying continuum form is the primary objective of this study, and to this end, we first fit simple power-law models to the STIS data, absorbed by E(B-V)=0.0296 mag, the extinction due to Galactic material in the line-of-sight. The best-fitting power-law has slope α =0.66 ± 0.14. The extrapolation of this continuum slope provides a good fit to the FUSE continuum, after correction of the FUSE data for absorption (using reddening curves from Hutchings & Giasson 2001 and Clayton et al. 1996). 148 However, the source continuum level in the STIS data is lower than that observed in Dec 1999 by FUSE. We find the normalizations of the STIS and FUSE datasets show a flux discrepancy, in the sense that the STIS data find the source at 55% of the flux level observed by FUSE. The uncertainty on absolute flux is ∼ 10% for FUSE and ∼ 2% for STIS data. Thus we attribute the discrepancy to variability in Ton S180 over the ∼ 5 weeks separating those observations. Figure 5.1 shows that the FUSE observation covered a similar flux state to the Chandra observation, so we do not want to rescale the FUSE data as it samples the same flux state as the X-ray data. Given an expectation of lags between emission in the different wavelength regimes of an AGN, it is always difficult to assess how to construct the most meaningful and instructive SED. The flux discrepancy in data from the overlapping bandpass of the STIS and FUSE data remove ambiguities as to breaks in intrinsic spectral shape, thus for construction of the SED of Ton S180 we scaled-up the STIS and ground-based data by a factor 1.78 to compensate for the flux variability. The absorbed power-law continuum extrapolates from the STIS band to agree with the FUSE continuum form. No spectral turnover is evident in these data, indicating that if the BBB component is contributing to the FUSE data, then its peak lies above ∼ 912 A(15eV).Theexcessesabovethecontinuumfit˚ are due to known emission features which are detailed in Table 5.3 and some 149 weak absorption features are also evident. In addition to the distinct lines, there is evidence for emission from Fe II and Fe III between 1900 and 2300 A˚ similar to that observed in another NLS1, I ZW 1 (Vestergaard & Wilkes 2001). Examination of the detailed STIS data shows that all of the emission lines expected for a Seyfert 1 galaxy are present, as well as a number of absorption lines from our Galaxy. The UV emission lines are narrow compared to those in typical Seyfert 1 galaxies; for example, the full-width at half-maximum of the C IV λ1549 A˚ line is ∼ 2300 ± 80 km s−1. The peak of C IV is blueshifted by 510 ± 20 km s−1 with respect to the systemic velocity. The emission lines are somewhat asymmetric, as illustrated by the C IV line, with more emission in the blue wing than red wing. There is no evidence for intrinsic UV absorption lines, which occur in ∼60 % of normal and narrow-line Seyfert 1 galaxies (Crenshaw et al. 1999). Although we cannot rule out the possibility of weak absorption at this resolution, we estimate an upper limit on the equivalent width of any C IV absorption to be 0.3 A.˚ 5.7. Ground-based Data Str¨omgren uvby observations (Str¨omgren 1956) were made of Ton S180 with the Danish 1.5m telescope at ESO, La Silla, on the night of 2000 January 20-21, and IR photometry was obtained on 2000 January 23 (JD 2451567.54) with the 150 near-IR imager/spectrograph OSIRIS mounted on the CTIO 1.5 m telescope. The details of the data reduction can be found in Turner et al. (2002). Table 5.4 reports the monochromatic fluxes obtained converting the uvby magnitudes f mλ/−2.5F F using λ =10 λ,mλ = 0 where the calibrating fluxes λ,mλ are taken from Pritchard et al. (1998) The IR monochromatic fluxes are given in Table 5.5. 5.8. Examination of the SED First we compared the SED data to the power-law continua determined for the UV and X-ray regimes. Figure 5.6 shows the extrapolation of the best-fitting power-lawtotheHST/STIS data (α =0.66) greatly overpredicts the X-ray flux. Clearly the spectrum must break somewhere between the UV and soft X-ray regimes. Also shown is the hard X-ray continuum slope, α =1.44, extrapolated to lower energies. This continuum intercepts the UV data around a few thousand A˚ but again, the hard X-ray power-law must terminate somewhere between the UV and soft X-ray regimes, as it overpredicts the optical and infrared data. In order to examine the approximate energy distribution for Ton S180, we first corrected the data for the small amount of extinction due to the Galactic line-of-sight gas. In the STIS band the reddening correction was made following Cardelli, Clayton & Mathis (1989) and in the FUSE band using Hutchings & Giasson (2001) and Clayton et al. (1996), both assuming E(B-V)=0.0296 mag, 151 the Galactic extinction. The absorption correction in the X-ray regime was made following Morrison & McCammon (1983) and assuming a Galactic value 1.55 × 1020cm−2. Table 5.6 summarizes some useful data from the SED. A simple parameterization was made of the spectral shape using the hard X-ray power-law, α =1.44 breaking to α =2.5 at 1 keV and then breaking to α =0.66 at 0.1 keV. This parameterization is shown as a solid green line in Figure 5.7. The peak of the SED in this case is 80 eV. The dotted green lines denote the uncertainty in the intrinsic SED, due to some uncertainty in the line-of-sight absorption measurement. Parameterization of the soft X-ray regime as a steep power-law is clearly inadequate, and we also overlay an alternative model with continuum plus diskbb soft component. It is interesting to see that the best-fitting diskbb model, which has a temperature of 98 eV at the inner radius, does not predict any BBB component would appear in the UV band. Even application of the physically meaningful models such as diskbb leave some unmodeled structure in the soft component, i.e. a sharp spectral break below 0.3 keV. It is currently unclear whether this structure represents the intrinsic form of the soft X-ray emission or whether it represents a residual uncertainty in the ACIS/LETG calibration. The break is not well modeled using neutral or ionized gas. The most obvious possibility remaining is that this sharp feature is due to the presence of emission features. However, uncertainty in calibration prompts us to note this structure but not to model it in detail. 152 5.9. Discussion 5.9.1. Interband Indices Table 5.7 shows the indices between various wavebands for Ton S180, compared to some values previously determined for other Seyfert type galaxies. A commonly cited slope is αox, and the value αox =1.52 ± 0.02 derived for Ton S180 α . +0.05 is consistent with ox =146−0.07 found for a sample of optically-selected radio quiet AGNs (Zamorani et al. 1981). Ton S180 appears X-ray weak compared to the mean index of 1.14 ± 0.18 determined for the ROSAT International X-ray/Optical Survey (RIXOS, Puchnarewicz et al. 1996) however, Ton S180 does lie within the the broad range found for RIXOS sources, which include both BLS1s and NLS1s. The value αox−hard was defined by Grupe et al. (1998) as the index linking 5500 A˚ and 1 keV, and Ton S180 lies within the broad ranges found for soft X-ray and hard X-ray selected AGN from that study, based on ROSAT observations. There are two questions of interest here, one is whether NLS1s as a class have systematically different interband indices to BLS1s, and the other is whether Ton S180 is unusual compared to other NLS1s. Nagao et al. (2001) have broached the first question by comparing the quantity αox for NLS1s and BLS1s. Those authors find average values and 1σ deviations αox−NLS1 =1.31 ± 0.16 and 153 αox−BLS1 =1.36 ± 0.24, thus concluding there to be no significant differences between this quantity for the two extremes of the Seyfert 1 population, contrary to some previous results (e.g. Puchnarewicz et al. 1996). In summary, based upon the comparison of interband indices with other studies yields no evidence that Ton S180 has an unusual ratio of optical/UV to X-ray flux. While opinions in the literature differ on whether there is a systematic difference in αox for the extremes of the Seyfert 1 population, it seems clear that interband indices have large ranges and their use is best suited to comparison of large samples of sources. In this study, we proceed by more detailed examination of the shape of the SED, and comparison of our data with other detailed SEDs. 5.9.2. The Form of the SED A SED for Ton S180 was first presented by Comastri et al. (1998), who found the soft X-ray component to contain the bulk of the energy in this Seyfert galaxy. This campaign of observations provides a more complete SED for Ton S180 than previously available, based on a large amount of simultaneous data. Examination of the detailed energy distribution of Ton S180 reveals significant differences compared to some other well-studied AGN. Figure 5.8 shows the extinction-corrected SED of Ton S180, and some parameterizations of its form. Overlaid on the parameterizations of the SED of Ton S180 are the 154 SEDs of other AGN; NGC 5548 is shown as a magenta dash-dotted line (Kraemer et al. 1998) while the mean radio-loud and radio-quiet quasars (from Elvis et al. 1994) are shown as dotted black and dashed blue lines, respectively. The most immediate result is that the SEDs of the Seyfert galaxies appear to peak somewhere in the extreme UV/soft X-ray band, while the quasars peak in the UV regime. Furthermore, the SED of Ton S180 peaks at a higher energy than that of NGC 5548. Some caution is required in the comparison of SEDs constructed with different datasets and various assumptions. Some apparent difference in SEDs could be an artifact of the assumption of some continuum form for the quasars, versus a simple joining of the soft X-ray to UV data for the Seyferts. However, such assumptions are only necessary in the problematic regime between ∼ 900A˚ and ∼ 0.1 keV. We find the evidence for true underlying differences between Ton S180 and the comparison sources to be strengthened by the absence of a contribution from the BBB in the UV band of Ton S180. A standard optically thick, geometrically thin accretion disk (Shakura & Sunyaev 1973) predicts the temperature of the peak of the disk spectrum T (R) to be a function of the mass of the central black hole and the accretion rate: T R ≈ . × 5 M/˙ M˙ 1/4 M −1/4 R/R −3/4 ( ) 6 3 10 ( Edd) 8 ( S) K (Peterson et al. 2000), where 8 R is the radius, RS is the Schwarzschild radius, M8 is the mass in units of 10 M 155 and M˙ is the accretion rate in units of the Eddington accretion rate. All other things being equal, a difference of two orders of magnitude in mass should yield a disk spectrum whose peak energy is a factor of 3 lower for the higher-mass system than the lower-mass system; similarly, a factor 100 increase in accretion rate (relative to the Eddington rate) shifts the peak to a factor 3 higher energy. Thus the RQQ SED (Figure 5.8) represents sources radiating at substantially below the Eddington accretion rate, and having a high central mass; these yield a relatively cool disk spectrum. At the other extreme, NLS1s are thought to be accreting close to the Eddington limit, and have a low mass; the disk spectrum appears hot. The BLS1 NGC 5548 represents an intermediate system in terms both of the accretion rate and central mass, and this appears to have an intermediate SED. Telfer et al. (2002) find that for a sample of QSOs, the entire continuum from 10 eV to 2 keV can be represented by a single power-law; this is clearly not the case in Ton S180 where the X-ray spectrum steepens below 2 keV, and neither the soft or hard X-ray components extrapolate to meet the UV data in a satisfactory way1. 1While the Chandra data show a turnover at 0.3 keV, this turnover is sharper than that expected from observation of the peak of the disk spectrum and there is some discrepancy between these data and data from other instruments. For these reasons, henceforth we will assume this turnover does not indicate the peak temperature of the disk. 156 8 NGC 5548 has a central black hole mass estimated at ∼ 10 M (Kaspi et al. 2000a). If the peak of the SED for NGC 5548 is close to the Lyman limit (T ≈ 1.6 × 105K) then an accretion rate of 11% of Eddington would be estimated, assuming a standard thin disk picture. Few strong constraints exist on the central mass in Ton S180. From > 6 variability observed in the X-ray regime (Chapter 3) we found MBH ∼ 8 × 10 M for Ton S180; however, there we assumed a bolometric luminosity which is lower > 7 than that revealed by this SED, leading to a revised limit MBH ∼ 8 × 10 M . Mass estimates such as these can be misleading if the X-ray variability is due, for example, to flares on the surface of the accretion disk, as the timescale of variation may not be directly related to the scale-size of the disk system. Thus we examine an alternative estimate of mass based upon the luminosity at 5100 A˚ 44 −1 (νLν ∼ 3 × 10 erg s ). Using the relation derived from other NLS1s (Peterson 7 et al. 2000, their Figure 7) we estimate a central mass M ∼ 2 × 10 M (with a factor ∼ 2 uncertainty) and the broad line region to exist at a radius ∼ 100 light days (Peterson et al. 2000, their Figure 6). This is in keeping with the systematically large BLR radii suggested by Giannuzzo & Stirpe (1996) for NLS1s compared to BLS1s. As the level of starlight contamination of the 5100 Afluxis˚ difficult to assess, this mass should be considered as an upper limit on the true central mass. The difference in SED peak energies is thus expected, as Ton S180 157 has a lower mass than NGC 5548, and NLS1s are thought to have systematically higher accretion rates than BLS1s. Unfortunately, the peak of the spectrum in Ton S180 remains loosely constrained. The EUVE data favor the simple parameterization of the X–UV spectrum (Figure 5.7) indicating that the peak lies close to or below 100 eV. Assuming a standard disk spectrum, the disk temperature must be greater than ∼ 15 eV, the peak of any cooler component of significant flux would show up in the FUSE data. The SED data indicate a peak close to 100 eV. Assuming a peak in disk emission for Ton S180 at this energy, which corresponds to 4 × 105 K, 7 ˙ ˙ then for M ∼ 2 × 10 M , M ≈ 0.88 × MEdd. For black holes operating near the Eddington limit the accretion disk surface is predicted to be highly ionized. There is certainly strong evidence for an ionized disk in Ton S180, as the Fe Kα line appears to be produced in highly ionized material in BeppoSAX (Comastri et al. 1998) and ASCA data (Turner, George, & Nandra 1998). The results from Ton S180 appear to fit into the standard disk picture. However, we also note that Cheng, Gaskell & Koratkar (1991) conclude that the standard disk model is not applicable to the UV spectra of quasars. Their case rests on the lack of any relation between αuv and luminosity. However, since the peak of the disk spectrum is generally at rest wavelengths of 1000 A˚ or shorter (Zheng et al. 1997; Telfer et al. 2002), αuv is indicative of only the rising edge of 158 the disk spectrum. In the standard disk model, the spectral slope in this region is relatively insensitive to luminosity, so one does not necessarily expect a strong correlation between αuv and luminosity of the BBB. As a final note, the lowest frequency IR point lies above the adjacent IR points. This has been observed in many AGN and is due to thermal emission from dust grains heated to close to their evaporation temperature (1500 K for graphite) close to the central engine (Rieke 1978). Recently, strong near-IR emission from the Seyfert 1 galaxy NGC 7469 has been attributed to very hot dust grains (T>900 K) associated with the putative torus (Marco & Alloin 1998), this is also observed in the SED of NGC3783 (Alloin et al. 1995). 5.9.3. The Energy Budget of Ton S180 The multi-power-law parameterization of the SED for Ton S180 makes it possible to estimate the luminosity in various energy regimes, helping to constrain reprocessing mechanisms and isolate the primary energy source. Table 5.8 shows the observed and intrinsic luminosities in several energy-bands, defined in the 46 −1 rest-frame of the source. The implied bolometric luminosity is Lbol ≈ 10 erg s . More luminosity emerges in the 10–100 eV regime than the 100 eV to 10 keV regime. Assuming that we are seeing all emitted radiation in each wavelength regime then this indicates that the EUVE–soft X-ray band contains the primary 159 spectral component, in keeping with disk-corona models (e.g. Haardt & Maraschi 1991). The energy budget and the SED show that Ton S180 is relatively X-ray weak above ∼ 2 keV (interband indices are insensitive to this, as historically X-ray fluxes for comparison with optical fluxes have been taken at soft X-ray energies). One possible reason for this is Compton-cooling of the hard spectrum by the large flux of soft X-ray and UV photons, as discussed by many authors, including Pounds, Done, & Osborne (1995). 5.9.4. The State of the Circumnuclear Gas The weak O VI absorption features detected in the FUSE data, the absence of absorption from lower ionization species in the HST data, and the lack of detectable X-ray absorption in the Chandra spectrum together indicate the presence of a small column of circumnuclear material which appears to be in a high state of ionization compared to that observed in other well-studied sources such as NGC 5548 and NGC 3783. The outflow velocity of ∼ 500 km s−1 is not unusual. Many Seyfert galaxies have shown evidence for outflow in UV and optical data (e.g. Crenshaw et al. 1997). In the X-ray regime Chandra grating observations have revealed supporting evidence for outflowing gas, with velocities of order a few hundred km s−1 (e.g. Collinge et al. 2001; Kaastra et al. 2000; 160 Kaspi et al. 2000b, 2001). A picture of Ton S180 being shrouded by highly ionized gas is consistent with earlier BeppoSAX (Comastri et al. 1998) and ASCA (Turner, George, & Nandra 1998) observations of Fe Kα emission, as well as the new ASCA data which show that the narrow component of Fe Kα is consistent with emission from Fe XXV–Fe XXVI. The ratio of the O VI to H I absorbing columns in the UV regime is comparable to that of the high-ionization component detected in Mrk 509 (Kriss et al. 2000), which was tentatively identified with the X-ray warm absorber in that object. However, in Ton S180 the total equivalent column density of hydrogen associated with the UV absorber must be < 1017 cm−2; too low to produce detectable X-ray absorption. However, the absence of ionized circumnuclear gas does not appear to be a general property of NLS1s. Some NLS1s do appear to show signatures of a warm absorber in the X-ray regime (e.g., Lee et al. 2001; Collinge et al. 2001) as well as UV absorption systems (Crenshaw et al. 1999). To examine the relation between the ionizing spectrum in Ton S180 and the circumnuclear gas, we took two estimates of the SED and total ionizing flux. The conservative estimate links the softest X-ray point at ∼ 0.3 keV to the highest end of the UV data with a simple power-law. For this SED, the total luminosity from 0.01–10 keV is ∼ 2.0 × 1045 ergs s−1 and the corresponding luminosity in ionizing photons is Q ≈ 2.9 × 1055 photons s−1. Taking instead the (extreme) 161 SED that peaks in the EUV (green line in Figure 7), the 0.01–10 keV luminosity is ∼ 3.8 × 1045 erg s−1 and Q ≈ 4.5 × 1055 photons s−1. Assuming the typical density and ionization parameter in the optical broad-line region (BLR) clouds, the radii at which the BLR exists can be estimated. Wandel, Peterson & Malkan (1999) used these “photoionization radii” and the measured line widths to derive black hole masses, which were in general agreement with those determined via reverberation mapping. In order to explore the role of luminosity on the line widths we instead use the masses derived from reverberation mapping and the photoionization radii (r) to estimate the line widths. Following Wandel, Peterson & Malkan (1999), we assume 2 10 Q/(4πr c ne) × ne ∼ 10 for the line emitting gas. Based on our estimates of the ionizing luminosity of the central source in Ton S180, we derive representative radial distances of r ≈ 8.8 × 1016 cm and ∼ 1.1 × 1017 cm, for the conservative and extreme cases, respectively (∼ 40 light days, somewhat smaller than the radius estimated from Peterson et al. 2000). If the BLR clouds are virialized around the central black hole, the FWHM of the emission lines should be roughly equal to 7 −1 GM/r; for a black hole mass of 10 M ,FWHM∼ 1290 km s (conservative case) and 1160 km s−1 (extreme case), in reasonable agreement with the observed FWHM of Hβ. 162 In Table 5.9 we show our estimates for the total ionizing flux, Q, for two BLS1s: for NGC 4151, Q is ∼ 2–8 ×1053 photons s−1 (Kraemer et al. 2001); for NGC 5548, Q is ∼ 1 × 1054 photons s−1 (Kraemer et al. 1998). Assuming the 7 black hole masses quoted by Wandel, Peterson & Malkan (1999) (1.2 x 10 M for 7 NGC 4151, and 6.8 x 10 M for NGC 5548), the corresponding “typical” BLR cloud distances and velocities are r ∼ 1.2 × 1016 cm and FWHM ∼ 3700 km s−1 for NGC 4151, and r ∼ 1.6 × 1016 cm and FWHM ∼ 7460 km s−1 for NGC 5548; the FHWMs are in rough agreement with the observed values. Clearly, the narrowness of the emission lines in Ton S180 is partially due to its stronger ionizing flux. Furthermore, given these “typical” BLR conditions, it is likely that the clouds within a few light days of the central source are too highly ionized even to produce much C IV emission (for a discussion of range in conditions in which emission lines form, see Baldwin et al. 1995). Hence, the higher value of Q for Ton S180 requires that the BLR gas is either more highly ionized than in BLS1s or, if the emission-line ratios are the same, it must be denser (i.e., denser gas is now ionized enough to contribute significantly to the emission-line spectrum). Either way, conditions are not identical to those in typical BLS1s. 163 5.10. Summary Construction of the spectral energy distribution for the bright NLS1 galaxy Ton S180 shows that most of the energy is emitted in the 10–100 eV regime, indicating that the primary source of emission dominates that band. The UV and X-ray data together constrain the peak of any BBB component to lie between 15 and 100 eV. This, and the overall shape of the SED indicate that emission from the accretion disk peaks at significantly higher energies in this source than in BLS1s, as expected if NLS1s have smaller central black holes and higher accretion rates. High-resolution spectra from FUSE reveal UV absorption due to O VI. The absence of absorption features in the HST data and the lack of neutral hydrogen absorption in the FUSE spectrum indicate a high-ionization state for the absorbing gas, while the absence of soft X-ray absorption shows that the column density is quite low. The highly-ionized state of the circumnuclear gas is most likely linked to the high luminosity and steep slope of the ionizing continuum in Ton S180. Given our constraints on the SED in Ton S180, we find that typical BLR emission lines would form at a radius which is an order of magnitude further out than in typical BLS1s. The BLR is estimated to exist at a radius ∼ 1017 cm, or 40 light days. 164 Observatory Instrument UT Dates Notes References (keV) (eV) (1) (2) (3) (4) (5) ASCA 1999 Dec 03–15 continuousa 1,2 Chandra LETG 1999 Dec 14–15 continuous 3 RXTEb PCA 1999 Nov 12–Dec 15 once every 96 min 2 EUVE 1999 Nov 12–Dec 15 continuousa,c 2 FUSE 1999 Dec 12 15.2 ks; 30 x30 (LWRS) 4 165 HST STIS 2000 Jan 22 6 ks; 52 x0 .24 ESO 1.5m 2000 Jan 21 uvby 4 CTIO 1.5m OSIRIS 2000 Jan 23 JHKs 4 Table 5.1: Observing Log for Ton S180. Notes: (a) except for gaps due to Earth occultation and passage of the spacecraft through the SAA. (b) Those data are not used in the construction of the SED, but were taken for the complementary timing project (Edelson et al. 2002). (c) A subset of these data were used in construction of the SED. References: (1) Chapter 3 (Romano et al. 2002a). (2) Edelson et al. 2002; (3) Turner et al. 2001b. (4) Chapter 5 (Turner et al. 2002). a Feature # Wλ Nion ∆v FWHM (A)˚ (1014 cm−2)(kms−1)(kms−1) (1) (2) (3) (4) (5) (6) O VI λ 1031.93 1 0.20 ± 0.03 1.7 ± 0.22 −146 ± 15 75 ± 11 20.07 ± 0.001 0.6 ± 0.01 −6 ± 15 27 ± 6 30.21 ± 0.02 2.0 ± 0.22 +109 ± 15 44 ± 5 Table 5.2: UV Absorption Lines. Notes: (a) Velocity is relative to a systemic redshift of z =0.06198. 166 Feature Flux FWHM (A)˚ (10−13 erg cm−2 s−1)(kms−1) (1) (2) (3) Lyαλ1216 30.8 ± 6.20 1961 ± 122 N V λ 1240 7.11 ± 1.40 1700 ± 848 Si II λ 1260 0.49 ± 0.19 ··· O I λ 1302 1.86 ± 0.45 921 ± 230 C II λ 1335 0.87 ± 0.30 1123 ± 280 S IV /OIV ] λ 1400 4.47 ± 0.94 ··· N IV] λ 1486 < 0.1 ··· C IV λ 1550 8.31 ± 0.61 2300 ± 80 He II λ 1640 1.95 ± 0.78 ··· O III ] λ 1663 0.60 ± 0.24 ··· N III] λ 1750 0.45 ± 0.22 ··· Si III] λ 1892 1.25 ± 0.25 ··· C III ] λ 1909 2.16 ± 0.43 ··· [ Ne IV] λ 2324 0.24 ± 0.10 ··· Mg II λ 2800 1.21 ± 0.24 1071 ± 428 Table 5.3: STIS Emission Lines. Notes: (a) Observed fluxes with no absorption correction. 167 a b Filter λ Mag. Fλ A(10˚ −19 erg cm−2 s−1 A˚−1) (1) (2) (3) (4) u 3500 14.94 ± 0.03 39.75 ± 1.02 v 4110 15.03 ± 0.02 11.94 ± 0.33 b 4670 14.74 ± 0.02 7.71 ± 0.13 y 5470 14.58 ± 0.02 5.26 ± 0.09 Table 5.4: uvby Photometry. Notes: (a) Str¨omgren magnitude (Str¨omgren 1956). (b) No rescaling applied for source variability. a Filter λ Mag. Fλ A(10˚ −19 erg cm−2 s−1 A˚−1) (1) (2) (3) (4) J 12150 13.22 ± 0.04 3.45 ± 0.10 H 16540 12.60 ± 0.03 1.40 ± 0.03 Ks 21570 11.67 ± 0.03 1.14 ± 0.03 Table 5.5: J, H and Ks Photometry. Notes: (a) No rescaling applied for source variability. 168 a a Rest Wavelength/Energy νLν (Observed) νLν (Intrinsic) (keV) (1044 erg s−1)(1044 erg s−1) (1) (2) (3) 2 µm 3.146 3.175 1 µm 3.842 3.953 7000 A˚ 4.283 4.544 5500 A˚ 4.485 4.823 3000 A˚ 5.272 6.124 2500 A˚ 5.356 6.371 1000 A˚ 6.053 8.892 0.25 keV 1.521 3.180 1 keV 0.655 0.686 2 keV 0.307 0.313 10 keV 0.156 0.156 Table 5.6: Data from the SED. −1 −1 Notes: (a) H0 =75kms Mpc ,q0 =0.5. 169 Index Definition Observed Intrinsic BLS1 (1) (2) (3) (4) (5) (6) a α αuv -2.096 log(F /F )0.53 ± 0.14 0.66 ± 0.14 0.85±0.06 3000 A˚−1000 A˚ 1000 A˚ 3000 A˚ b α -0.489 log(F0.25 keV/F )1.24 ± 0.03 1.12 ± 0.03 0.73 5500 A˚−0.25 keV 5500 A˚ c c α αox−hard -0.378 log(F1keV/F )1.37 ± 0.02 1.38 ± 0.02 1.13 5500 A˚−1keV 5500 A˚ 170 d α1µm−2keV αix -0.312 log(F2keV/F1µ)1.35 ± 0.02 1.35 ± 0.02 1.14-2.16 +0.05 e f α αox -0.384 log(F2keV/F )1.50 ± 0.02 1.52 ± 0.02 1.46− . 1.21 ± 0.02 2500 A˚−2keV 2500 A˚ 0 07 g αx -1.431 log(F2keV/F10 keV)1.44 ± 0.07 1.44 ± 0.07 0.91 Table 5.7: Spectral Indices Notes: (a) Index in the ∼ 2200-1200 A˚ band (Cheng, Gaskell & Koratkar 1991). (b) Turner et al. 1999a. (c) Grupe et al. 1998. (d) Lawrence et al. 1997. (e) Zamorani et al. 1981. (f) Puchnarewicz et al. 1996. (g) Nandra et al. 1997b. a a Energy Lobserved LIntrinsic (keV) (1044 erg s−1)(1044 erg s−1) (1) (2) (3) 5×10−4–0.01 13.94 15.69 0.01–0.1 2.86 28.80 0.1–1 2.03 8.92 1–10 0.65 0.66 5×10−4–10 19.49 54.07 0.1–10 2.69 9.57 Table 5.8: Luminosities. Object QMrFWHM 53 −1 7 −1 (10 s )(10M )(cm)kms (1) (2) (3) (4) (5) NGC 4151 2–8 1.2 1.6 3700 NGC 5548 10 6.8 1.2 7460 Table 5.9: Characteristic data for 2 BLS1s. 171 Fig. 5.1.— The 12-day ASCA light curve from Chapter 3. The periods covered by FUSE and Chandra observations are noted. EUVE data were extracted to be simultaneous with the Chandra observation. 172 Fig. 5.2.— The light curve of the combined ±1st order Chandra/LETG data (0.2- 10.0 keV) using 256 s bins. The corresponding portion of the ASCA SIS light curve is overlaid and denoted by the open symbols. 173 Fig. 5.3.— LETG (magenta) and ASCA (red and black are SIS, blue and green are GIS) data/model ratio compared to the α =1.44 power-law. The soft excess shows a shape which may be partially due to residual calibration uncertainties below 0.3 keV. 174 Fig. 5.4.— Top: FUSE spectrum of Ton S180 binned to a resolution of 0.6 A˚ (100 pixels). In addition to the noticeable broad O VI emission line, suggested identifications for other weak far-UV emission lines are marked. The associated O VI absorption lines are indicated. All other absorption lines are foreground Galactic or intergalactic features. Bottom: the section of the FUSE spectrum surrounding the peak of the broad O VI emission line is shown. The spectrum is binned to a resolution of 0.12 A˚ (20 pixels) to show the continuum and emission lines more clearly. The three associated O VI absorption systems are marked, as well as the corresponding locations expected for Lyβ absorption. The Fe II feature is foreground Galactic absorption. 175 Fig. 5.5.— The HST STIS (1150–3150 A)˚ data, the solid line shows a power-law of energy index α =0.66 convolved with extinction by E(B-V)=0.0296, (i.e. neither the data or model has absorption correction). The strong lines are labelled, lines are tabulated in detail in Table 5.3. 176 Fig. 5.6.— The multi-waveband data for Ton S180. In this plot, the data have not been corrected for absorption. The blue dashed line shows the (absorbed) power- law α =1.44 from the 2-10 keV regime, extrapolated to lower energies. The green solid line shows the power-law α =0.66 (convolved with line-of-sight absorption) from the fit to the STIS and FUSE data, extrapolated to higher energies. The open point between log ν =16–17 Hz represents the EUVE data, the circles represent the ground-based data. 177 Fig. 5.7.— The Spectral Energy Distribution of Ton S180. The data have been corrected for Galactic line-of-sight extinction. The circles represent the ground- based data. The simple model parameterization of the SED is shown as a solid (green) line. The dotted (green) line straddling this shows the uncertainty in the SED due to uncertainty in the galactic line-of-sight absorption. The dashed (BB) line is the blackbody and the dotted (PL) line the power-law model components of the SED, the dashed line (PL+BB) is their sum. 178 Fig. 5.8.— The Spectral Energy Distribution of Ton S180, represented by the solid green line from the simple parameterization. The dashed red line shows the sum of the blackbody and power-law model components (as in Figure 5.7). For comparison, the mean SED for radio-loud and radio-quiet quasars are also shown as dotted black and dashed blue lines, respectively (Elvis et al. 1994). The SED of NGC 5548 (a Sy1.5) is shown as a magenta dash-dot line (Kraemer et al. 1998). 179 Chapter 6 Quasi-simultaneous Spectral Energy Distribution of the Narrow-Line Seyfert 1 Galaxy Arakelian 564 6.1. Introduction In this Chapter we present a preliminary spectral energy distribution of the NLS1 galaxy, Ark 564, constructed with quasi-simultaneous data obtained during 2000. We compare this SED with that of Ton S180 (Chapter 5) and with those obtained for BLS1s to infer how the relative accretion rates vary among the Seyfert 1 population. In §6.2 we describe the observations and data reduction, summarize the main results of the monitoring campaign, and describe our methods for reddening correction. In §6.3 we present the SED of Ark 564. In §6.4 we discuss some 180 implications of our investigation, and in §6.5 we indicate possible next steps in our study of these objects. 6.2. Observations 6.2.1. Data Reduction Table 6.1 summarizes the log of the observations of Ark 564 used in this study. Column 1 and 2 report the observatory and instrument which obtained the data; Column 3 shows the date in which the data were obtained; Column 4 lists the ranges of energy and wavelength we used for our work; Column 5 lists the total exposure times and the slit size, along with other relevant notes for a particular observation; finally, Column 6 reports the references to the papers where the data were first published. Most of the data have already been published, therefore here we will only briefly summarize their reduction and analysis, referring the reader to the original papers for further details, and to §6.2.2 for a summary of the main results. The ASCA data (Chapter 2, Paper I) were obtained starting from Julian Date 2451697.024 for a total exposure time of ∼ 2.98 Ms, and were reduced using standard techniques as described in Nandra et al. (1997a) with the methods and screening criteria utilized by the Tartarus database (Turner et al. 1999c). 181 Data screening yielded an effective exposure time of ∼ 1.11 Ms for the SIS detectors and ∼ 1.29 Ms for the GIS detectors. The degradation of the low energy response of the SIS detectors was corrected for with the method of Yaqoob et al. (2000), i.e. parameterized the efficiency loss with a time-dependent absorption 20 −2 21 −2 (NH(SIS0) = 7.5 × 10 cm , NH(SIS1) = 1.05 × 10 cm , Chapter 2). FUSE observed Ark 564 for 63 ks starting from 2001 June 29 07:37:42 UT (Chapter 4, Paper V). The observations consisted of 24 exposures performed in photon address (time-tag) mode through the 30 × 30 low-resolution (LWRS) aperture, and the data reduction is described in Chapter 4. The HST data (Paper II and IV) were obtained with the Space Telescope Imaging Spectrograph (STIS) in 46 visits between 2000 May 9–July 8 (the first five visits were separated by intervals of 5 days, the remaining by 1 day). The spectra were obtained through the 52 × 0 .5 slit and the low-resolution G140L and G230L gratings which yield a spectral resolution of ∼ 1.2 Aand3.2˚ Ain˚ the 1150–1730 A˚ and 1570–3150 A˚ ranges, respectively. The data were reduced using the IDL software developed at NASA’s Goddard Space Flight Center for the STIS Instrument Definition Team (Lindler 1998). The spectra have been corrected for small wavelength intercalibration uncertainties following Korista et al. (1995). The uncertainty in the relative wavelength calibration is on the order of 0.6 Aand1.7˚ A˚ for the G140L and G230L gratings, respectively. A separate 182 mean spectrum was created for the G140L and G230L grating separately, given the different resolutions. In the optical we combined two spectra. The first one, which covers the 3170–4160 A˚ wavelength range, was obtained in 1980 at Lick Observatory (D. E. Osterbrock 2002, private communication). The second one is the mean of the spectra taken between 1998 Nov to 2001 Jan at the Tel Aviv University Wise observatory (Paper III). The host galaxy starlight contribution has been estimated by measuring its flux through PSF fitting to field stars in V-band images of the galaxy taken at Wise Observatory (corresponding to 40 % of the −15 −1 −2 −1 total light at 5200 A,˚ i.e. Fgal =2.4 × 10 ergs s cm A˚ , Paper III), and subtracted from the mean spectrum. We also retrieved archival IRAS flux measurements at 12, 25, 60, and 100 µm (Moshir et al. 1990) through the NASA/IPAC Extragalactic Database (NED). Figure 6.1 shows the quasi-simultaneous SED of Ark 564 before correction for intervening absorption is applied. We note that while the HST and Wise spectra are simultaneous (as well as simultaneous with the ASCA spectrum), the FUSE and Lick spectra were obtained one year later, and 20 years earlier, respectively. The FUV-optical rest-frame spectrum of Ark 564 covering the 1000-7790 A˚ wavelength region is presented in Figure 6.2 (labeled as (a)). 183 6.2.2. Summary of results from the multi-waveband observations The continuum fit to the mean ASCA spectrum (with a power-law model 20 −2 modified by Galactic absorption, NH =6.4 × 10 cm ,Dickey&Lockman 1990) yields a slope Γ = 2.538 ± 0.005 (Chapter 2). The strong excess of emission observed below 2 keV was parameterized as a Gaussian of peak energy E . ± . +11 =057 0 02 keV and mean equivalent width (EW)= 110−15 eV. The soft hump component is also found to be variable in flux down to timescales of about a day and in shape down to timescales of days (Chapter 2). Parameterization of the soft excess as a black-body yields a temperature T =1.8 × 106 K and luminosity 38 −1 Lbb =2.48 × 10 ergs s (Chapter 4). A strong, ionized (E ≈ 7keV)FeKα line is detected, which shows variations in flux and EW down to timescales of a week (Chapter 2). The FUSE spectrum is dominated by the strong emission in the O VI λλ1032, 1038 resonance doublet, and heavily saturated absorption due to Hydrogen Lyman-Werner bands, O VI is observed at velocities near the systemic redshift of Ark 564 (Chapter 4). The available data suggest that the UV and X-ray absorbers are physically related, possibly identical, and spatially extended along the line of sight, and characterized by total column density and ionization parameter log NH ≈ 21 and log U ≈−1.5. The absorbing gas is in a state of 184 outflow with respect to the nucleus and carries out a kinetic luminosity about one order of magnitude smaller than the observed radiative luminosity of the source (Chapter 4). The UV continumm flux (1365 A,˚ Paper II) is well correlated with both the hard (2-10 keV) and soft (0.75-2 keV) X-ray fluxes and with the soft hump flux. No significant lags are detected between the X-ray and UV bands (Chapter 2). There is evidence for UV/optical continuum time delays. The variations of the continuum at 3000 A˚ (Paper II) and at 4900 A˚ (Paper III) lag behind those at 1365 A˚ by about 1 day and 1.8 days, respectively. The UV/optical delays were interpreted as evidence for a stratified continuum reprocessing region, possibly an accretion disk. The variations of the Lyα emission line, which lag the variations of the continuum at 1365 A˚ by 3 days, were used, in conjunction with the line width, to 6 determine the virial mass of the central black hole, 8 × 10 M (Paper II). This estimate is however uncertain due to the low amplitude of the Lyα emission line variations (1%), which may indicate that the bulk of the emission is at larger radii; however, the estimate in Paper II agrees with the one obtained by Pounds et al. (2001) based on a fluctuation power spectrum analysis of X-ray variability. While the black hole mass and 5100 A˚ luminosity of Ark 564 are consistent with the hypothesis that NLS1s have lower black hole masses and higher accretion rates 185 than BLS1s of comparable luminosity, the low level variability observed in the emission lines is different from most Seyfert 1 galaxies, which characteristically display variations of 10 % on similar timescales. 6.2.3. Reddening Correction Given the indications (Paper IV) that strong intrinsic neutral absorption is present in Ark 564 in excess of the Galactic absorption, special care has been paid in correcting the data for reddening. We adopted two methods: 1. Using a standard Galactic extinction curve (Cardelli, Clayton & Mathis 1989) with E(B − V )=0.06 mag along the line of sight through our Galaxy (Schlegel, Finkbeiner, & Davis 1998). From here on the SED obtained with this reddening correction will be called SED1. 2. Using a standard Galactic extinction curve with E(B − V )=0.03 mag plus the intrinsic extinction curve derived by Crenshaw et al. (2002, Paper IV) derive for Ark 564 and E(B − V )=0.14 mag. In the FUSE band we extrapolated the extinction correction linearly from the HST band, as suggested by Hutchings & Giasson (2001) and Sasseen et al. (2002). From here on the SED obtained with this reddening correction will be called SED2. 186 The effects of reddening corrections in the optical–UV bands are presented in Figure 6.2, where the observed spectrum in the 1000-7790 A˚ wavelength region (labeled as “(a)”) is compared to the spectra obtained correcting for absorption according to the two methods described above (labeled as “(b)” and “(c)”, respectively). 6.3. The SED of Ark 564 Figure 6.3 shows the SEDs in the optical–X-ray range. A power-law fit of the continuum in the optical-FUV region1 yields spectral indices α = −0.82±0.012 for −α SED1 and α = −1.58 ± 0.01 for SED2 (specific flux Fν ∝ ν ). Extrapolation of these power-laws into the X-ray regime either greatly underpredicts or overpredicts the X-ray flux. Similarly, the hard X-ray continuum slope (α =Γ− 1=1.538, §6.2.2, thick solid line) extrapolated to lower energies overpredicts the optical-FUV flux. Clearly, both the optical-FUV and the X-ray power-laws must break at some energy between the FUV and soft X-ray. Unfortunately, the position of this break is quite uncertain, due to the fact that, depending on whether internal 1The fit was made to 9 bands: λ = 1005–1007, 1029.5–1030.5, 1101–1107, 1114–1118, 1155– 1180, 1350–1380, 1460–1500, 1620–1660, and 7040–7050 A.˚ 2In both cases the uncertainties are purely statistical. The continuum fit in Paper II was performed only on the HST data and produced a spectral index α = −0.88 ± 0.01, comparable to our value, since the uncertainties are underestimated in both cases. 187 extinction is considered or not, the optical-FUV part of the SED assume quite different shapes. Figure 6.3 shows the ranges of energy of the break, and relative position of the SED peak. If no intrinsic extinction correction is applied, then the SED peaks in the soft X-ray, as previously observed by Comastri et al. (2001). If intrinsic extinction is assumed to be present, then the SED peaks at ∼ 50 eV. Table 6.2 summarizes some relevant data from the SEDs, derived using the simple parameterization described above. Column (1) is the rest wavelength/energy, Column (2) the observed value of νLν, Column (3) and (4) are the reddening corrected values for SED1 and SED2, respectively. 6.4. Discussion A non-simultaneous optical, UV and X-ray SED was presented by Comastri et al. (2001) who found that it peaks in the soft-X-ray band. Here we present an SED which is obtained from simultaneous data covering almost 4 decades in energy3. Simultaneity is particularly important for NLS1s, since, as a class, they can be extremely variable in time, although Ark 564 has shown only weak variability in the optical-UV bands (Paper II-III). 3The Lick spectrum, which was obtained in 1980 is merely used to fill in a small gap in the optical data, and the 2001 FUSE spectrum shows a continuum slope consistent with the one obtained from the HST spectrum. This latter fact which suggests that, although the flux level changed between 2000 and 2001, the overall shape of the optical-FUV SED did not change. 188 For an optically thick, geometrically thin accretion disk (Shakura & Sunyaev 1973), the radial dependence of the temperature can be expressed as a function of the mass of the central black hole and the accretion rate as, T R ∼ . × 5 m 1/4 M −1/4 R/R −3/4 R ( ) 6 3 10 (˙) 8 ( S) K (Peterson et al. 2000), where 8 is the radius, RS is the Schwarzschild radius, M8 is the mass in units of 10 M ˙ ˙ andm ˙ = M/MEdd is the accretion rate in units of the Eddington accretion rate. Assuming that the disk radiates locally as a black-body, and that most of the emission originates in the innermost regions of the disk, the observed peaks of the SEDs correspond to T (SED1) ≈ 3.2 × 106 KandT (SED2) ≈ 1.9 × 105 K. For M8 ≈ 0.1, as derived from Pounds et al. (2001), these temperatures indicate accretion ratesm ˙ (SED1) = 1700 andm ˙ (SED2) = 0.02. According to Czerny & Elvis (1987), the black-body approximation leads to much higher effective temperatures in the disk, hence an overestimation of the accretion rate by a factor of ∼ 5. Even with this correction,m ˙ (SED1) seems uncomfortably large. Because of this, as well as the fact that Crenshaw et al. (2002) find evidence of reddening in excess of Galactic, we consider SED1 no further. We note that the estimate of the accretion rate for SED2, is strongly affected by the uncertainty on the mass of the central black hole. An alternative mass estimate derived from the M–σ relationship (e.g., Ferrarese et al. 2001), would probably lead to more interesting limits on the accretion rate. The largest uncertainty on the accretion rate, however, is due to the loose constraint on the 189 peak of the SED, hence on the temperature (m ˙ ∝ T 4). Although no conclusive support for the idea that NLS1s are low-mass objects accreting at high accretion rates can be derived from these arguments, a comparison of the SED2 shape with the SEDs of other AGNs can be used to infer how the relative accretion rates might vary among the Seyfert 1 population. Figure 6.4 compares the SED of Ark 564 with the mean SED for radio-quiet quasars (Elvis et al. 1994) and LZ (Laor et al. 1997a; Zheng et al. 1997), the Seyfert 1.5 galaxy NGC 5548 (Kraemer et al. 1998), and the NLS1s Ton S180 (Turner et al. 2002). There are significant differences in the intrinsic shape of the SED across the AGN population (see Turner et al. 2002), the most evident being the energy of the peak and the presence (or lack of) of the big blue bump (BBB), the signature of the emission from the accretion disk. Among the NLS1s, RE J1034+396 peaks at ≈ 250 eV, and accretes atm ˙ =0.3–0.7 (Puchnarewicz et al. 2001), Ton S180 peaks at ∼< 100 eV and accretes atm ˙ =0.88 (Turner et al. 2002), and SED2 peaks at ∼ 50 eV. In none of these NLS1s an indication of the presence of an optical/UV BBB, and a strong soft X-ray excess is seen instead. In this light, Ark 564 is also consistent with the idea that the accretion disk is so hot in NLS1s that the BBB is shifted in the EUV–soft X-rays. Table 6.3 lists the observed and intrinsic (SED2) luminosities in different 45 energy bands. We estimate that the bolometric luminosity is Lbol ∼ 10 ergs 190 s−1. More energy is emitted in the 100 eV–1 keV band than in any other decade, constituting roughly half of the emitted energy in the opt–X-ray ranges. This implies that the primary spectral component peaks in the soft X-ray band, as generally consistent with disk-corona models (Haardt & Maraschi 1991). Finally, we note that Ark 564 is relatively FIR bright compared to the sample of radio-quiet quasars and the LZ sample. Crenshaw et al. (2002) have suggested that the associated warm UV absorber is lukewarm and dusty. The FIR emission observed in this object could then be thermal emission from the dust grains embedded in the absorber as they are heated by the strong continuum. Given the IR brightness in this object and in many NLS1s (Moran, Halpern, & Helfand 1996), it is not unlikely that a contribution might be coming from the host galaxy, in the form of a nuclear starburst (Mathur 2000a). 6.5. Future Work The next step in this project is to address the energy budget of Ark 564, by examining the emission lines observed in this object. We will use the code Cloudy4 (v94.00, Ferland 1996) to infer the physical properties of the line-emitting gas through photoionization modeling. In particular, it will prove of interest to address the issue of the weakness of [O III] in NLS1s compared to BLS1s. The 4http://www.pa.uky.edu/∼gary/cloudy/. 191 ionization potentials O+,O++,andO+++ are 5.117 eV, 54.934 eV, and 77.413 eV, respectively. Our calculations should be able to determine if the different ionizing fields in NLS1s and BLS1s are responsible for the different strength of O III,and in particular, if a viable explanation is that, given the SED of NLS1s, most of the Oxygen may be in the stage of O IV. We will also compare the equivalent width of O VI in NLS1s and BLS1s. Another issue of general interest is the strength of Fe II, which is considered to be one of the defining characteristic of NLS1s. The strength of Fe II in NLS1s compared to BLS1s has led Mathur (2000a) to consider that NLS1s may present super-solar abundances. If this were the case, it would agree well with our findings of Fe Kα emission line having a large equivalent width both in Ton S180 and in Ark 564. 192 Observatory Instrument UT Dates Wavelengtha Notes References (1) (2) (3) (4) (5) (6) ASCA 2000 Jun 1–Jul 6 0.75–9.76 keV 2.98 Ms, continuousb 1 FUSE 2001 Jun 29-30 1000–1175 A63ks;30˚ x30 (LWRS) 2 HST STIS/G140L 2000 May 9–Jul 8 1175–1711 A˚ 554304 s; 52 x0 .53,4 HST STIS/G230L 2000 May 9–Jul 8 1711–3143 A˚ 24216 s; 52 x0 .53,4 193 Lick 1980 3170–4160 A5˚ Wise FOSC 1998 Nov–2001 Jan 4160–7790 A6˚ IRAS 12, 25, 60, 100 µm7 Table 6.1: Observing Log for Arakelian 564. Notes: (a) Observed-frame wavelength/energy bands utilized. (b) Except for gaps due to Earth occultation and passage of the spacecraft through the SAA. References: 1: Turner et al. (2001a); 2: Romano et al. (2002b); 3: Collier et al. (2001); 4: Crenshaw et al. (2002); 5: ); 6: Shemmer et al. (2001); 7: Moshir et al. (1990). a a a Rest Wavelength νLν (Observed) νLν (SED1) νLν (SED2) /Energy (×1043 ergs s−1)(×1043 ergs s−1)(×1043 ergs s−1) (1) (2) (3) (4) 1 µmb 2.045 2.082 2.165 7000 A˚ 1.851 1.956 2.660 5500 A˚ 1.73 1.875 3.057 3000 A˚ 1.461 1.685 4.338 2500 A˚ 1.388 1.632 4.819 1000 A˚ 1.075 1.390 8.178 0.25 keVb 0.011 7.114 7.114 1 keV 2.569 4.592 4.592 2 keV 2.026 2.371 2.371 10 keV 0.895 0.957 0.957 Table 6.2: Data from the SEDs −1 −1 Notes: (a) H0 =75kms Mpc ,q0 =0.5. (b) Extrapolated value. 194 a Energy Lobserved LSED2 (keV) (×1043 ergs s−1)(×1043 ergs s−1) (1) (2) (3) 0.001–0.01 ··· 0.221 × 10−5 0.001–0.05 ··· 0.276 × 10−3 0.01–0.05 ··· 0.274 × 10−3 0.05–0.1 ··· 9.069 0.01–0.1 ··· 9.343 0.1–1 2.201 16.314 1–10 4.086 4.485 0.05–10 6.288 29.867 0.1–10 6.288 20.799 0.001-10 ··· 30.143 Table 6.3: Luminosities 195 Fig. 6.1.— Quasi-simultaneous SED of Ark 564. The data have not been corrected for reddening. 196 Fig. 6.2.— (a): FUV-Optical rest-frame spectrum of Ark 564, obtained combining the FUSE spectrum, the HST G140L and G230L mean spectra, the Lick spectrum, and the Wise mean spectrum. (b): Full spectrum after correction for reddening (see §6.2.3) using a standard Galactic extinction curve with E(B −V )=0.06 mag. (c): Full spectrum after correction for reddening using a standard Galactic extinction curve with E(B − V )=0.03 mag plus the intrinsic extinction curve Crenshaw et al. (2002) derive for Ark 564 and E(B − V )=0.14 mag. All data have been corrected for redshift (z =0.02467; de Vaucouleurs et al. 1991). The spectra (b) and (c) have been offset by 1 × 10−14 ergs s−1 cm−2 A˚−1 for clarity. 197 Fig. 6.3.— Reddening corrected, rest-frame SEDs of Ark 564 (see §6.2.3). The optical-FUV data are as in Figure 6.2. The long-dashed and short-dashed lines show the power-law fits of the data reddening corrected with the two methods (spectral indices α = −0.82 ± 0.01 and α = −1.58 ± 0.01, respectively). The latter is extrapolated to higher energies. The green line is the X-ray data. The solid line is the power-law fit to hard X-ray data (§6.2.2) extrapolated to lower energies. The dash-dot line is the black-body model for the soft X-ray excess. 198 Fig. 6.4.— Comparison of the SED of Ark 564 with the mean SED for radio-quiet quasars (RQ; Elvis et al. 1994) and LZ (Laor et al. 1997a; Zheng et al. 1997), the Seyfert 1.5 galaxy NGC 5548 (Kraemer et al. 1998), and the NLS1s Ton S180 (Turner et al. 2002). All SEDs have been normalized to the absorption-corrected rest-frame flux of Ark 564 at 2 keV. 199 Chapter 7 Conclusion 7.1. Summary This dissertation study has investigated the nature of Narrow-Line Seyfert 1 galaxies, a subclass of AGNs whose extreme continuum and emission line properties make them ideal for testing the models of AGNs. The hypothesis we are testing is that NLS1s have relatively lower black hole masses and higher accretion rates than normal broad-line Seyfert 1 galaxies (BLS1s) of the same luminosity. This study has concentrated on Ark 564 and Ton S180, two NLS1s that have been the object of extensive multiwavelength spectroscopic campaigns. Chapter 2 and 3 presented an analysis of the X-ray variability properties of Ark 564 and Ton S180, based on 35- and 12-day continuous observations with ASCA, respectively. Chapter 4 presented the FUSE spectrum of Ark 564, which is dominated by the strong emission in the O VI λλ1032, 1038 resonance doublet, 200 and an analysis of the properties of the UV–X-ray warm absorber. Chapter 5 and 6 presented an investigation of the spectral energy distributions of Ton S180 and Ark 564. Our main results can be summarized as follows. 1. X-ray time-selected spectroscopy (a) The mean spectrum of Ark 564 and Ton S180 is characterized by a very steep power-law continuum (Γ = 2.54 and 2.44, respectively), a strong soft excess (soft hump) at energies < 2keV,andanFeKα line that has a large equivalent width and originates in ionized material. (b) The amplitudes and timescales of the rapid variations may be a result of stochastic noise in the number of reprocessing clumps, which depend on the evolution of magnetic loops. The observed fast variability of the soft hump rules out an origin of the soft emission in large-scale components (e.g. circumnuclear starburst). (c) In Ton S180, we observe a photon index–soft hump correlation that is consistent with the soft hump being produced by up-scattering of the accretion disk radiation within a flaring disk corona, where an increase of the flux in the soft X-ray/UV component can cool the corona and steepen the power-law continuum. 201 (d) The photon index–soft hump correlation is not as tight in Ark 564; however, the good correlation between the UV continuum flux with both the hard X-ray flux and the soft hump flux is consistent with reprocessing models. 2. UV–X-ray warm absorber (a) There is excellent agreement between the kinematic properties of the UV absorber emerging from the combined analysis of the FUSE and HST/STIS spectra, i.e. distribution of gas in radial velocity (as derived from the extent of the absorption troughs) and net radial velocity (as derived from the velocity centroids). (b) The available UV, FUV and X-ray data on absorption lines suggest that the UV and X-ray absorbers in Ark 564 are physically related, and possibly identical. (c) The combination of constraints on column densities indicates that the absorber is characterized by a narrow range in total column density NH and U, centered at log NH ≈ 21 and log U ≈−1.5. (d) The UV–X-ray warm absorber may be spatially extended along the line of sight. 3. Spectral energy distributions (a) Ton S180. 202 i. Most of the energy is emitted in the 10–100 eV regime, indicating that the primary source of emission dominates that band. ii. The UV and X-ray data together constrain the peak of any BBB component to lie between 15 and 100 eV. iii. The emission from the accretion disk peaks at significantly higher energies in this source than in BLS1s, as expected if NLS1s have smaller central black holes and higher accretion rates. (b) Ark 564. i. The peak of the SED is less well constrained. ii. However, in our simple parameterization, most of the energy is emitted in the 0.1–1. keV regime. (c) These results are consistent with the predictions for an AGN with high accretion rate (M˙ ≈ 0.1–0.9) onto a relatively low-mass black hole 6−7 (M ≈ 10 M ). 7.2. Future Work As noted in Chapter 6, the logical next step in the Ark 564 SED project is to address the energy budget of Ark 564, by examining its emission lines. We will investigate the physical properties of the line-emitting gas through photoionization modeling. In particular, it will prove of interest to address the 203 issue of the weakness of O III in NLS1s compared to BLS1s. The ionization potentials O II,OIII,andOIV are 5.117 eV, 54.934 eV, and 77.413 eV. Our calculations should be able to determine if the different ionizing field in NLS1s and BLS1s is responsible for the different strength of O III, and in particular, if a viable explanation is that, given the SED of NLS1s, most of the Oxygen may be in the stage of O IV. We will also compare the equivalent width of O VI in NLS1s and BLS1s. AmoregeneralissueisthestrengthofFeII, which is considered to be a defining characteristic of NLS1s. Although there are rather large differences in the amount of Iron emission observed in the spectra of NLS1s, as our two objects highlight (Ton S180 having rather strong Iron lines, but Ark 564 having unexceptional Fe II strengths), the strength of Iron in NLS1s compared to BLS1s has led Mathur (2000a) to speculate that NLS1s may have super-solar metal abundances. If this were the case, it may agree well with our findings of Fe Kα emission line having a large equivalent width both in Ton S180 and in Ark 564. We will also investigate the properties of the warm absorber in another nearby, bright NLS1 galaxy NGC 4051, which has a comparatively accurate determination of the mass of the central black hole from emission-line reverberation 6 mapping (M ≈ 1.4 × 10 M ; Peterson et al. 2000), and which has been the object of a very extensive AGN Watch monitoring program with RXTE and ground 204 based observatories. Simultaneous Chandra and HST observations (Collinge et al. 2001) showed that in the X-ray spectrum two distinct absorption systems are resolved, a high-velocity blueshifted system at −2340 ± 130 km s−1 and a low-velocity blueshifted system at −600 ± 130 km s−1. In the UV the strong absorption from C IV and N V is resolved in as many as nine different intrinsic absorption systems with velocities between −650 and 30 km s−1. Though the low-velocity X-ray absorption is consistent in velocity with many of the UV absorption systems, the high-velocity X-ray absorption seems to have no UV counterpart. Our work on the O VI absorption lines with our FUSE data (which have already been collected, but are not reduced at the time of writing) will further constrain the characteristics of the warm absorber in this object. The FUSE data will also allow us to use the FUV emission-line diagnostics to cover the gap between the well-known optical-UV and EUVE–X-ray properties. Given the 20 low Galactic column density along the line-of-sight to NGC 4051 (NH =1.3 × 10 cm−2; Elvis, Lockman, & Wilkes 1989), NGC 4051 is also an excellent subject for determination of the optical-to-X-ray SED, and it will be interesting to compare its SED with that of other NLS1s and BLS1s. 205 BIBLIOGRAPHY Alloin, D., et. al. 1995, A&A, 293, 293 Arav, N., et al. 2001, ApJ, 561, 118 Baldwin, J., Ferland, G., Korista, K., Verner, D. 1995, ApJ, 455, L119 Ballantyne, D. R., Iwasawa, K., & Fabian, A. C. 2001, MNRAS, 323, 506 Boller, Th., Brandt, W. N., & Fabian, A.C. 1997, MNRAS, 289, 293 Boller, Th., Brandt, W. N., & Fink, H. 1996, A&A, 305, 53 Boroson, T. A. & Green, R. F. 1992, ApJS, 80, 109 Brandt, W. N., Boller, T., Fabian, A. C., & Ruszkowski, M. 1999, MNRAS, 303, L53 Brandt, W. N., Fabian, A. C., Nandra, K., Reynolds, C. S., & Brinkmann, W. 1994, MNRAS, 271, 958 Brandt, W. N., Mathur, S., & Elvis, M. 1997, MNRAS, 285, L25 206 Branduardi-Raymont, G., Sako, M., Kahn, S. M., Brinkman, A. C., Kaastra, J. S., & Page, M. J. 2001, A&A, 365, L140 Brotherton, M. S., Green, R. F., Kriss, G. A., Oegerle, W., Kaiser, M. E., Zheng, W., & Hutchings, J. B. 2002, ApJ, 565, 800 Burke, B. E., Mountain, R. W., Harrison, D. C., Bautz, M. W., Doty, J. P., Ricker, G. R., & Daniels, P. J. 1991, IEEE Trans. ED-38, 1069 Cardelli, J. A., Clayton, G. C., Mathis, J. S. 1989, ApJ, 345, 245 Cheng, F. H., Gaskell, C. M., Koratkar, A. P. 1991, ApJ, 370, 487 Clayton, G. C., et al. 1996, ApJ, 460, 313 Collier, S. J., et al. 2001, ApJ, 561, 146 (Paper II) Collinge, M. J., Brandt, W. N., Kaspi, S., Crenshaw, D. M., Elvis, M., Kraemer, S. B., Reynolds, C. S., Sambruna, R., & Wills, B. 2001, ApJ, 557, 2 Comastri, A., et al. 1998, A&A, 333, 31 Comastri, A., et al. 2001, A&A, 365, 400 Crenshaw, D. M., 1997, in ASP Conf Ser 113, Emission Lines in Active Galaxies: new Methods and Techniques, ed. B.M. Peterson, F.-Z. Cheng & A.S.Wilson, IAU Collq. 159 (San Francisco: ASP), 240 207 Crenshaw, D. M., Kraemer, S. B., Boggess, A., Maran, S. P., Mushotzky, R. F., & Wu, C. 1999, ApJ, 516, 750 Crenshaw, D. M., et al. 2002, ApJ, 566, 187 (Paper IV) Czerny, B. & Elvis, M. 1987, ApJ, 321, 305 de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H. G., Buta, R. J., Paturel, G., & Fouque, P. 1991, S&T, 82, 621 Dickey, J. M., & Lockman, F. M. 1990, ARA&A, 28, 215 Edelson, R., Turner, T. J., Pounds, K., Vaughan, S., Markowitz, A., Marshall, H., Dobbie, P., Warwick, H. 2002, ApJ, 568, 610 Elvis, M. 2000, ApJ, 545, 63 Elvis, M., Lockman, F. J., & Wilkes, B. J. 1989, AJ, 97, 777 Elvis, M., et al. 1994, ApJS, 95, 1 Fabian, A. C., Iwasawa, K., Reynolds, C. S., & Young, A. J. 2000, PASP, 112, 1145 Fabian, A. C., Rees, M. J., Stella, L., & White, N. E. 1989, MNRAS, 238, 729 Ferland, G. J. 1996, Hazy, a brief introduction to Cloudy 94.00, Univ. Kentuky Dept. Phys. Astron. Int. Rep. 208 Ferrarese, L., Pogge, R. W., Peterson, B. M., Merritt, D., Wandel, A., & Joseph, C. L. 2001, ApJ, 555, L79 Fink, H. H., Walter, R., Schartel, N., & Engels, D. 1997, A&A, 317, 25 Forster, K. & Halpern, J. 1996,ApJ, 468, 565 Frank, J., King, A., & Raine, D. 1992, Accretion Power in astrophysics (2d ed.; Cambridge, Cambridge University Press) Gaskell, C. M., & Peterson, B. M. 1987, ApJS, 65, 1 Gaskell, C. M., & Sparke, L. S. 1986, ApJ, 305, 175 George, I. M., Turner, T. J., Netzer, H., Nandra, K., Mushotzky, R. F., & Yaqoob, T. 1998, ApJS, 114, 73 Ghisellini, G., Haardt, F., & Matt, G. 1994, MNRAS, 267, 743 Giannuzzo, M. E., Stirpe, G. M. 1996, A&A, 314, 419 Goodrich, R. W. 1989, ApJ, 342, 234 Goodrich, R. W. 2000, New Astronomy Review, 44, 519 Grupe, D., Beuermann, K., Thomas, H.-C., Mannheim, K., & Fink, H. H. 1998, A&A, 330, 25 Guainazzi, M., Piro, L., Capalbi, M., Parmar, A. N., Yamaguchi, M., & Matuoka, M. 1998, A&A, 339, 327 209 Guilbert, P. W., Fabian, A. C., & McCray, R. 1983, ApJ, 266, 466 Haardt, F., & Maraschi, L., 1991, ApJ, 380, 51 Haardt, F., Maraschi, L., & Ghisellini, G. 1994, ApJ, 432, L95 Hamann, F., Barlow, T. A., Junkkarinen, V., & Burbidge, E. M. 1997, ApJ, 478, 80 Huchra, J. P., Vogeley, M. S., & Geller, M. J. 1999, ApJS, 121, 287 Hutchings, J. B. & Giasson, J., 2001, PASP, 113, 1205 Kaspi, S., Brandt, W. N., Netzer, H., Sambruna, R., Chartas, G., Garmire, G. P., & Nousek, J. A. 2000a, ApJ, 535, L17 Kaspi, S., Smith, P. S., Netzer, H., Maoz, D., Jannuzi, B. T., & Giveon, U. 2000b, ApJ, 533, 631 Kaspi, S. et al. 2001, ApJ, 554, 216 Kaastra, J. S., Mewe, R., Liedahl, D. A., Komossa, S., & Brinkman, A. C. 2000, A&A, 354, L83 Korista, K., Ferland, G., & Baldwin, J. 1997, ApJ, 487, 555 Korista, K., et al. 1995, ApJS, 97, 285 Kraemer, S. B., Crenshaw, D. M., Filippenko, A. V., & Peterson, B. M. 1998, ApJ, 499, 719 210 Kraemer, S. B., et al. 2001, ApJ, 551, 671 Kriss, G. 1994, in ASP Conf. Ser. 61, Astronomical Data Analysis Software and Systems III, ed. D. R. Crabtree, R. J. Hanisch, & J. Barnes (San Francisco: ASP), 3, 437 Kriss, G. A., et al. 2000, ApJ, 538, L17 Krolik, J. H., Madau, P., & Zycki,˙ P. T. 1994, ApJ, 420, L57 Kuraszkiewicz, J., Wilkes, B. J., Brandt, W. N., & Vestergaard, M. 2000, ApJ, 542, 631 Laor, A. 1991, ApJ, 376, 90 Laor, A., Fiore, F., Elvis, M., Wilkes, B. J., & McDowell, J. C. 1997a, ApJ, 477, 93 Laor, A., Jannuzi, B. T., Green, R. F., & Boroson, T. A. 1997b, ApJ, 489, 656 Lawrence, A., Elvis, M., Wilkes, B. J., McHardy, I., Brandt, N. 1997, MNRAS, 285, 879 Lee, J. C., Ogle, P. M., Canizares, C. R., Marshall, H. L., Schulz, N. S., Morales, R., Fabian, A. C., Iwasawa, K. 2001, ApJ, 554, 13 Leighly, K. M. 1999a, ApJS, 125, 297 Leighly, K. M. 1999b, ApJS, 125, 317 211 Leighly, K. M. 2001, in ASP Conf. Ser. 224, Probing the Physics of Active Galactic Nuclei, ed. B. M. Peterson, R. S. Polidan, & R. W. Pogge (San Francisco: ASP), 293 Lindler, D. 1998, CALSTIS Reference Guide (CALSTIS Version 5.1) Makishima, K., Maejima, Y., Mitsuda, K., Bradt, H. V., Remillard, R. A., Tuohy, I. R., Hoshi, R., & Nakagawa, M. 1986, ApJ, 308, 635 Malkan, M. A. & Sargent, W. L. W. 1982, ApJ, 254, 22 Maraschi, L., Haardt, F. 1997, in IAU Colloq. 163, Accretion Phenomena and related Outflows, ed. D. Wickramasinghe, L. Ferrario, & G. Bicknell, Publ. Astron. Soc. of Australia Press Marco, O., & Alloin, D. 1998, A&A, 336, 823 Marziani, P., Sulentic, J. W., Dultzin-Hacyan, D., Calvani, M., & Moles, M. 1996, ApJS, 104, 37 Mathews, W. G. & Ferland, G. J. 1987, ApJ, 323, 456 Mathur, S. 1994a, ApJ, 431, L75 Mathur, S. 2000a, MNRAS, 314, L17 Mathur, S. 2000b, New Astronomy Review, 44, 469 Mathur, S., Elvis, M., & Wilkes, B. 1995, ApJ, 452, 230 212 Mathur, S., Elvis, M., & Wilkes, B. 1999, ApJ, 519, 605 Mathur, S., Wilkes, B., Elvis, M., & Fiore, F. 1994b, ApJ, 434, 493 Matsumoto, C., Leighly, K. M., & Marshall, H. L. 2001, X-ray Emission from Accretion onto Black Holes, E46 Matt, G., Fabian, A. C., & Ross, R. R. 1993, MNRAS, 264, 839 Matt, G., Fabian, A. C., & Ross, R. R. 1996, MNRAS, 278, 1111 Mitsuda, K., et al. 1984, PASJ, 36, 741 Monier, E. M., Mathur, S., Wilkes, B., & Elvis, M. 2001, ApJ, 559, 675 Moos, H. W., et al. 2000, ApJ, 538, L1 Moran, E. C., Halpern, J. P., & Helfand, D. J. 1996, ApJS, 106, 341 Morrison, R., & McCammon, D. 1983, ApJ, 270, 119 Morton, D. C. 1991, ApJS, 77, 119 Moshir, M. & et al. 1990, IRAS Faint Source Catalogue, version 2.0 (1990), 0 Nagao, T., Murayama, T., Taniguchi, Y. 2001, ApJ, 546, 744 Nandra, K., George, I. M., Mushotzky, R. F., Turner, T. J., & Yaqoob, T. 1997a, ApJ, 476, 70 213 Nandra, K., George, I. M., Mushotzky, R. F., Turner, T. J., & Yaqoob, T. 1997b, ApJ, 477, 602 Nandra, K., Le, T., George, I. M., Edelson, R. A., Mushotzky, R. F., Peterson, B. M., & Turner, T. J. 2000, ApJ, 544, 734 Oegerle, B., Murphy, E., & Kriss, G. A. 2000, FUSE Data Handbook V1.1, 42 Ogle, P. M., Marshall, H. L., Lee, J. C., & Canizares, C. R. 2000, ApJ, 545, L81 Ohashi, T., et al. 1996, PASJ, 48, 157 Osterbrock, D. E., & Pogge, R. W. 1985, ApJ, 297, 166 Peterson, B. M. 1993, PASP, 105, 207 Peterson, B. M., Wanders, I., Horne, K., Collier, S., Alexander, T., Kaspi, S., & Maoz, D. 1998, PASP, 110, 660 Peterson, B. M. et al. 2000, ApJ, 542, 161 Pounds, K. A., Done, C., & Osborne, J. P. 1995, MNRAS, 277, L5 Pounds, K. A., Edelson, R., Markowitz, A., & Vaughan, S. 2001, ApJ, 550, L15 Pritchard, J. D., Tobin, W., Clark, M., & Guinan, E. F. 1998, MNRAS, 297, 278 Puchnarewicz, E. M., Mason, K. O., Romero-Colmenero, E., Carrera, F. J., Hasinger, G., McMahon, R., & Mittaz, J. P. D., Page, M. J., Carballo, R. 1996, MNRAS, 281, 1243 214 Puchnarewicz, E. M., Mason, K. O., Siemiginowska, A., Fruscione, A., Comastri, A., Fiore, F., & Cagnoni, I. 2001, ApJ, 550, 644 Reynolds, C. S. 1997, MNRAS, 286, 513 Rieke, G. H. 1978, ApJ, 226, 550 Rodriguez-Ardila, A. Binette, L., Pastoriza, M. G., Donzelli, C. J. 2000, ApJ, 538, 581 Romano, P., Turner, T. J., Mathur, S., & George, I. M. 2002a, ApJ, 564, 162 (Chapter 3) Romano, P., Mathur, S., Pogge, R. W., & Peterson, B. M., & Kuraszkiewicz, J. 2002, ApJ, in press (Chapter 4; Paper V) Ross, R., Fabian, A. C., & Mineshige, S. 1992, MNRAS, 258, 189 Sahnow, D. J., et al. 2000, ApJ, 538, L7 Sako, M., Kahn, S. M., Paerels, F., & Liedahl, D. A. 2000, ApJ, 543, L115 Sasseen, T. P., Hurwitz, M., Dixon, W. V., & Airieau, S. 2002, ApJ, 566, 267 Savage, B. D. & Sembach, K. R. 1991, ApJ, 379, 245 Savage, B. D., et al. 2000, ApJ, 538, L27 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 215 Sembach, K. R., et al. 2000, ApJ, 538, L31 Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337 Shemmer, O., et al. 2001, ApJ, 561, 162 (Paper III) Shields, G. A. 1978, in Pittsburgh Conf. on BL Lac Objects, ed. A. M. Wolf, (University of Pittsburgh), 257 Shull, J. M., et al. 2000, ApJ, 538, L73 Stark, A. A., Gammie, C. F., Wilson, R. W., Bally, J., Linke, R. A., Heiles, C., & Hurwitz, M. 1992, ApJS, 79, 77 Str¨omgren, B. 1956, Vistas in Astron. 2, 1336 Telfer, R.C., Zheng, W., Kriss, G.A., & Davidsen, A. F. 2002, ApJ, in press Turner, T. J., George, I. M., Grupe, D., Nandra, K., Remillard, R. A., Leighly, K. M., Marshall, H. L., Kraemer, S. B., & Crenshaw, D. M. 1999a, ApJ, 510, 178 Turner, T. J., George, I. M., & Nandra, K. 1998, ApJ, 508, 648 Turner, T. J., George, I. M., & Netzer, H. 1999b, ApJ, 526, 52 Turner, T. J., Romano, P., George, I. M., Edelson, R., Collier, S. J., Mathur, S., & Peterson, B. M. 2001a, ApJ, 561, 131 (Chapter 2; Paper I) 216 Turner, T. J., et al. 1999c, in Proceedings of the 19th Texas Symposium on Relativistic Astrophysics and Cosmology, ed. J. Paul, T. Montmerle, & E. Aubourg (Saclay: CEA), E441 Turner, T. J., et al. 2001b, ApJ, 548, L13 Turner, T. J., et al. 2002, ApJ, 568, 120 (Chapter 5) Urry, C. M., & Padovani, P. 1995, PASP, 107, 803 Vestergaard, M., & Wilkes, B. J. 2001, in Spectrscopic Challenges of Photoionized Plasmas. ASP Conference Series, Vol TBD, 2001, ed. G. Ferland & D. W. Savin Vaughan, S., Pounds, K. A., Reeves, J., Warwick, R., & Edelson, R. 1999, MNRAS, 308, L34 Vaughan, S., Reeves, J., Warwick, R., & Edelson, R. 1999, MNRAS, 309, 113 Walter, R. & Fink, H. H. 1993, A&A, 274, 105 Wandel, A., Peterson, B.M., & Malkan, M. 1999, ApJ, 526, 579 White, R. J., & Peterson, B. M. 1994, PASP, 106, 879 Wisotzki, A., Dreizler, S., Engels, D., Fink, H. H., & Heber, U. 1995, A&A, 297, L55 Yaqoob, T., et al. 2000, ASCA GOF Calibration Memo, ASCA-CAL-00-06-01, v1.0 217 Yaqoob, T., George, I. M., Nandra, K., Turner, T. J., Serlemitsos, P. J., & Mushotzky, R. F. 2001, ApJ, 546, 759 Zamorani, G., et al. 1981, ApJ, 245, 357 Zheng, W., Kriss, G. A., Telfer, R. C., Grimes, J. P., & Davidsen, A. F. 1997, ApJ, 475, 469 218