A Thesis On Microquasars

Hard X-ray and Gamma Ray
Properties of Cosmic Sources

A Thesis submitted to the University of Mumbai for the
Ph. D. (Science) Degree

in Physics
Submitted by
Manojendu Choudhury
Under the Guidance of

Prof. A. R. Rao
Tata Institute of Fundamental Research

Mumbai 400 005
June 2004
ii
To My Parents
ii
Acknowledgements
First and foremost, I wholeheartedly acknowledge the institutional structure, geograph-

ical location, general ambiance and the colonnade of TIFR for providing the ideal site
where, perhaps, the most eventful years of my life, in both professional and personal
fronts, were unfolded. The academic environment of TIFR provided me a very thorough
introduction to the rigours of the research life, providing a gamut of diverse experiences,
transforming me from a “green rookie” to a level from where (hopefully) I can take off
to the next step to establish myself.
I will always remain in gratitude to my thesis supervisor Prof. A. R. Rao, for picking
me up from a situation of dire nothingness, setting me on a path, providing me a goal to
strive for, and for letting me believe in myself when most others had given up.
I am indebted to Prof. P. C. Agrawal, Prof. R. K. Manchanda, Prof. K. P. Singh,
Prof. J. S. Yadav, Prof. T. P. Singh and Prof. M. N. Wahia for providing me various
academic opportunities, discussions, and constant support during the various stages of
my doctoral tenure. I would like to mention Dr. B. Paul separately for all the above, plus
many a friendly chatting sessions. I thank the scientific staff of the department for a very
congenial atmosphere in the department, and I should mention Magnes and Shobha for
their general cheerful disposition and very helpful nature.
The greatest treasures that I will carry with me from my stay here are the friendships
that I forged over the years. Santosh, Gulab and Sachi provided a unique environment
of camaraderie and togetherness in the department. Discussions and conversations with
them consisted of a complete package of friendship, understanding and professional
acumen, covering all aspects of science, life and everything else that may follow! I re-
ally pride ourselves at creating an atmosphere devoid of petty individualism and jealous
competition, so common among contemporaries. I can defnitely not fail to mention our
immediate juniors in the office, Vikram and Sarita, who cheerfully sustained the very
friendly air in the office rooms and the tea-tables. In the department, I acknowledge
the companionship of Poonam, Sambaran, Harsha, Surajit, Uddipan, and last but def-
initely not the least, Rituparno. In the institute, Pratik-da always provided a sense of
togetherness unique to him (even now he is waiting for me to finish my typing, get the
i
ii Acknowledgements
printout, so that he may guide me to the book-binders). Funda-da (Pranab Sen) left a
unique impression in me, Surjeet was always a constant and consistent friend, Yogesh,
Yeshpal, Ashok (MP), Arvinder, Amitava, Krishnan, Rahul Jain, Neel made my later
years in the institute very enjoyable, while Dr. Paul (Dilip), Shubham, Tirtha, Roop, Ra-
jesh, Anwesh-da, Bhaswati-di, Arun, Rudrajyoti Palit, Bahniman-da, Soumen-da made
the early years unforgettable. I hope Shankar, Anand, Dipanker, Holla, Eknath, Suman,
Manna, Shamik and others of the footballing fraternity continue with the tradition of
kicking the ball regularly. I will always remember Girish Nathan, Kiran and Anjum for
their company in our first year (graduate school), and I will never forget Biswajit’s an-
tics, in and out of the cricket field. I may have, inadvertently forgotten, to name many
people who made my stay in the hostel a memorable experience. It will be a sacrilege if
I fail to mention my ‘Guruji’, Shri Namdeo Panchalji, who introduced Hindustani clas-
sical music to me in the last two and a half years, and in the process rendered me a most
creative and fulfilling avocation.
I sincerely appreciate the friendship of Dr. C. H. Ishwar-Chandra and the contin-
ued academic collaboration with him. I am indebted to Dr. Ashok K. Jain for his sup-
port, encouragements, academic discussions and research collaborations. I look forward
to the continued association along with the current ensuing collaboration with Vivek
Agrawal and V. Girish. I cherish the friendship and support of Dr. Sergio Mendoza,
and I hope to continue my association with him and his beloved country, Mexico, where
I was treated with great warmth and affection. I also acknowledge the hospitality of
Dr. Divakara Mayya in Instituto Nacional de Astrofisica, Optica y Electronica (INAOE),
Puebla, Mexico.
I gladly acknowledge the Kanwal Rekhi Scholarship of the TIFR Endowment Fund,
which provided partial support to this thesis. I acknowledge the Department of Astron-
omy and Astrophysics, TIFR, and the NATO Advanced Study Institute for providing me

financial and other logistical support enabling me to attend the Summer School at
Les Houches, France on, “Accretion discs, jets and high energy phenomena in astro-
physics”. I thank the IAU for providing partial support for my participation in the “IAU
Colloquium 194: Compact binaries in the Galaxy and beyond”, at La Paz, Mexico, as

well as providing complete support for the “IAU Asia Pacific Regional Meeting” at
Tokyo, Japan. I also acknowledge the Kyoto University (Dept. of Physics Yukawa Insti-
tute), ISAS and University of Tokyo (Dept. of Physics) for providing the complete sup-
port for my participation in the international conference on “Stellar-mass, Intermediate-
mass and Supermassive Black Holes” at Kyoto, Japan. I acknowledge Dr. G. C. De-
wangan (Gulab) and Prof. R. Griffiths for inviting me at the Carnegie Mellon University,
Pittsburgh for a short visit. I should mention that Dr. S. V. Vadawale (Santosh) played
iii
host to me in Boston during my participation in the international meeting on “X-ray Tim-

ing: RXTE and beyond”, organized by the Harvard University. I also acknowledge the
Instituto de Astronomia, Universidad Nacional Autonoma de Mexico (UNAM), as well
as INAOE, Puebla, Mexico, for inviting me to short visits at the respective institutes. I
also acknowledge the National Centre for Radio Astronomy, for providing support for
my participation at the “Summer school on radio interferometry and aperture synthesis”,
which introduced the rudiments of radio astronomical data analysis, with emphasis on
GMRT, which I hope will be of enormous benefit to me in near future.
No words can do justice to the support of my family, especially my wife, Rajul, who
has been the bed-rock of my support base through all times, good and bad; without her
understanding and endearing inspiration I wouldn’t have reached this day, when I can
see the completion of my thesis. Lastly, I would like to express all my love and wishes
to the little bundle of joy, our eleven days old son.
Contents
Acknowledgements i
Synopsis vii
1 Introduction 1
1.1 High energy physical processes and phenomena . . . . . . . . . . . . . 1
1.1.1 Accretion disc as a source of high energy emission . . . . . . . 4
1.2 Black hole sources: stellar mass and supermassive . . . . . . . . . . . . 5
1.3 Microquasars: general properties and behaviour . . . . . . . . . . . . . 8
1.3.1 Transient X-ray blackhole binaries . . . . . . . . . . . . . . . . 9
1.3.2 Persistent sources: canonical states of X-ray emission . . . . . 10
1.4 Accretion in X-ray binaries . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4.1 Unification of hydrodynamic solutions of accretion flow . . . . 17
1.4.2 Hard X-ray emission models . . . . . . . . . . . . . . . . . . . 19
1.4.3 Geometrical structure of the accretion system . . . . . . . . . . 20
1.5 Outflows from microquasars . . . . . . . . . . . . . . . . . . . . . . . 21
1.6 Aim of this thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2 X-ray detectors and techniques of instrumentation; Radio astronomy 26

2.1 X-ray detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Modern X-ray observatories . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.1 The Rossi X-ray Timing Explorer (RXTE) . . . . . . . . . . . . 31
2.2.2 Compton Gamma Ray Observatory (CGRO) . . . . . . . . . . . 34
2.2.3 Other notable X-ray missions . . . . . . . . . . . . . . . . . . 37
2.3 X-ray astronomical data analyses and techniques . . . . . . . . . . . . 38
2.3.1 RXTE data analysis . . . . . . . . . . . . . . . . . . . . . . . . 40
2.3.2 Timing analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 41
iv
CONTENTS v
2.3.3 Spectral analysis . . . . . . . . . . . . . . . . . . . . . . . . . 42

2.4 Radio astronomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.4.1 Green Bank Interferometer (GBI) observatory . . . . . . . . . . 46
2.4.2 Giant Metrewave Radio Telescope (GMRT) . . . . . . . . . . . 47
3 Cygnus X-3: spectral studies 49

3.1 Why Cygnus X-3? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2 General properties of Cygnus X-3 . . . . . . . . . . . . . . . . . . . . 50
3.3 General spectral features of Cygnus X-3 . . . . . . . . . . . . . . . . . 53
3.3.1 A historical perspective . . . . . . . . . . . . . . . . . . . . . . 53
3.3.2 X-ray wide band spectra from RXTE . . . . . . . . . . . . . . . 54
3.4 Correlation of radio & X-ray emission in Cygnus X-3: Spearman’s Par-
tial Rank Correlation test . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.4.1 Spearman’s Partial Rank Correlation test . . . . . . . . . . . . 60
3.5 X-ray spectral pivoting in the low (hard) state . . . . . . . . . . . . . . 63
3.6 X-ray spectral evolution driving the radio flares: high (soft) state . . . . 67
3.7 Complete X-ray spectral evolution . . . . . . . . . . . . . . . . . . . . 71
4 Cygnus X-3: temporal studies 73

4.1 Binary modulation and correction with a given ephemeris . . . . . . . . 73
4.2 Radio X-ray correlation of Cygnus X-3 . . . . . . . . . . . . . . . . . 79
4.3 Power Density Spectrum (PDS) . . . . . . . . . . . . . . . . . . . . . 81
4.3.1 Power Density Spectrum (PDS) of Cygnus X-3 . . . . . . . . . 84
4.4 Time lag between soft and hard X-rays . . . . . . . . . . . . . . . . . . 91
5 Disc-jet connection in microquasars: low (hard) states 94

5.1 Radio:X-ray correlation of the persistent sources: hard states . . . . . . 97
5.1.1 Cygnus X-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.1.2 GRS 1915+105 . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.1.3 Cygnus X-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.2 Uniform behaviour of X-ray spectral shape with radio emission . . . . . 106
5.2.1 The X-ray soft state and suppressed radio emission . . . . . . . 109
5.3 Universal correlation and its origin . . . . . . . . . . . . . . . . . . . . 111
5.4 X-ray spectral shape as the “driver” of the radio emission . . . . . . . . 114
5.5 Summary: the generalized picture of the accretion - ejection mechanism
in the ‘low’ - hard state of Galactic microquasars . . . . . . . . . . . . 115
vi CONTENTS
6 Two Component Accretion Flow model 117

6.1 Two Component Accretion Flow (TCAF) . . . . . . . . . . . . . . . . 118
6.2 Outflow of mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.3 The magnetized TCAF model . . . . . . . . . . . . . . . . . . . . . . 124
6.4 Phenomenological picture of accretion and ejection connection . . . . . 125
6.4.1 Cygnus X-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.4.2 Hybrid Comptonization . . . . . . . . . . . . . . . . . . . . . 129
7 Summary and conclusions 131

7.1 Summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . . 131
7.1.1 Cygnus X-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
7.1.2 Generalized picture of disc-jet connection in Galactic micro-
quasars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.2 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Synopsis
An X-ray is a quantum of electromagnetic radiation with an energy, to an order of magni-

tude approximation, some 1000 times greater than that of optical photons. Traditionally,
the soft X-ray band is defined as the energy range 0.5 – 12 keV (corresponding to wave-

length of 25 – 1 ), the hard X-ray extends to 50 keV and the energy range beyond
it till a few MeV is regarded as soft gamma rays, although this classification is not very
stringent. High energy astronomy pertains to the observation of the sky in this regime
of the electromagnetic spectrum. The study of cosmic sources at these high energies of
X-rays and gamma-rays began only in the early 1960’s, after the serendipitous discov-
ery of the low mass X-ray binary (LMXRB) Sco X-1, which houses a neutron star (the
compact object) and a low mass optical companion in the main sequence.
Simple extrapolation from the optical regime suggests that, assuming the physical
processes giving rise to these X-ray, gamma ray emissions are thermal, the temperature
of the radiating matter should be of the order

K for X-ray photons and greater
for gamma ray photons. The fundamental physical mechanisms which give rise to high
energy emissions from a thermalised distribution of matter are few, viz. thermal black
body radiation, bremsstrahlung, Compton scattering. Soon, however, it was discovered,
mainly from the supernova remnants, that non-thermal physical processes also play very
important part in these high energy emission. Such physical processes may also involve
bremsstrahlung and Compton scattering, in addition to synchrotron emission.
This interplay of thermal / non-thermal emission is best observed in accreting black
hole (both stellar mass and super massive) and neutron star systems. Various theoretical
paradigms exist today which attempt to comprehensively explain the accretion phenom-
ena in these systems. Shakura & Sunyaev laid the foundation of the first standard disc
models (now known as SS discs). However these discs were untenable due to instabil-
ity arising out of temperature crossing the hydrogen ionization point, and this led to the
classic disc instability paradigm which sought to explain various transitions and variabil-
ities in the accretion disc systems. Improved wide band X-ray observational capabilities
enabled the spectral energy distribution (SED) of these sources to reveal the ubiquitous
vii
viii Synopsis
presence of power law component extending well into hard X-ray and gamma ray band,
along with any soft component (black body, generally multicoloured). Thereafter var-
ious paradigms involving hybrid (i.e. both thermal and non-thermal) Comptonization,
advection dominated accretion flow (ADAF), Compton reflection, bulk motion Comp-
tonization, two component accretion flows, etc., with or without magnetic field, were
developed to explain the hard power law extension, the various states of black hole tran-
sitions and their variabilities. Models involving synchrotron emission, Compton self-
synchrotron, etc., are also used to explain the high energy emission in these sources.
Long term monitoring of the various high energy SEDs and variabilities of these sources
are needed to devise any comprehensive physical and geometrical picture of the pro-
cesses.
Quasars, which fall in the broader classification of radio loud Active Galactic Nu-
clei (AGN), were discovered in the radio band of electromagnetic radiation during the
era of the very early discovery of X-ray sources. These were subsequently identified to
be accreting supermassive black holes

of galactic scales with outflows
in the form of a jet observable in the radio band by virtue of the physical mechanism
of synchrotron emission. Therefore a paradigm of accretion being related to ejection
was gradually developed, although it was not observable in the radio quiet AGNs. The
discovery of Galactic X-ray binaries exhibiting (superluminal) radio jets, with both phys-
ical and temporal (variability) scale roughly at 6 orders of magnitude less than those of
quasars, led to the notion of ubiquitous presence of outflow in the form of collimated jets
in accreting black hole systems and low magnetic field (
G) neutron stars, lending
them the terminology of microquasars. The observation of microquasars over the AGN
is advantageous for, chiefly, two reasons. Firstly, these sources are located within the
Galaxy, the astronomical equivalence of our own backyard. And, secondly, the charac-
teristic dynamical time scales in the flow of matter are proportional to the black hole
mass and any variability time scale of hours to days of microquasars correspond to anal-
ogous phenomena with duration of hundreds to thousands of years in AGNs, assuming
that the same fundamental physical processes underlie the behaviour of these sources.
Therefore monitoring the microquasars for a few days may sample phenomena not pos-
sible to observe in quasars. These features led to the current upsurge in the observational
study of these sources. The aim of this thesis is to gather together a comprehensive pic-
ture of the high energy observational features of Galactic microquasars, with a particular
emphasis on the enigmatic source Cygnus X-3, in order to develop a phenomenological
understanding of the fundamental processes and geometrical structure of these systems.
Since accretion (inflow of matter, generally in the form of discs) and ejection (outflow of
matter in collimated jets) are closely related, a correlated study of high energy and radio
ix
emission is presented to provide a coherent picture of the systems.

Microquasars, generally black hole candidates (BHCs), mimic, at a much smaller
scale, the main astrophysical attributes of a quasar: general relativistic accretion identi-
fied by the X-rays and gamma rays from the surrounding accretion discs, and the special
relativistic outflows in the form of collimated jets with low opening angles (
!#" ) ob-
served by means of their synchrotron emission. Of the 200 Galactic X-ray binaries cata-
logued so far, about 20 are radio loud, half of which show evidences of radio jets, a few
of them superluminal (eg. GRS 1915+105, GRO J1655-40). These X-ray binary sources
have some common salient characteristics which may be enumerated as follows:-
Structural characteristics:-
$ They consist of one compact object (generally a black hole candidate) and one
normal star, generally from the main sequence.
$ The compact object accretes matter from the companion, via an accretion disc.
The donor may lose mass through Roche lobe overflow or via stellar wind. The
extent of the inner disc is a function of time (and probably accretion rate), the
explanation of the variability generally depends on the particular model adopted
to explain the X-ray characteristics.
$ The outflow of matter takes place via a collimated beam, visible in the radio, at
times infra-red or, arguably, even X-ray. The conical jet has a small opening angle
( %
! " ) directed perpendicular to the accretion plane. This system may show
precessional movement in some cases.
From the observational point of view, the study of the behavioural pattern of these
sources, in the various electromagnetic bands, may be classified into three different types
of analysis: 1) image analysis, which gives the (extended) spatial information about the
source, 2) temporal analysis, which gives the variability of the source with respect to
time, and 3) spectral analysis, which gives the pattern of the emission with respect to
the energy (or frequency / wavelength), and provides the best analytical tool for identi-
fying the physical processes giving rise to the emission. Some basic generic patterns of
the temporal and spectral characteristics (in the X-ray regime) of the microquasars are
highlighted below:-
Temporal characteristics:-
$ The X-ray light curve may show a variety of diverse variabilities, even for a single
source, depending on the particular states or transitions among them during the
period of observation.
x Synopsis
$ The power density spectra (PDS) shows, typically, a power law dependence with
a positive index in the region of 0.01 – 1 Hz, flat spectra for the next decade of
frequency range, followed by a power law decay (i.e. negative index) of power in
the 10 –100 Hz. The PDS of the neutron stars is generally shifted towards the
higher frequency region by an order of magnitude.
$ Various types of quasi periodic oscillations (QPOs) are observed in most of the
microquasars, prominently in the low-hard state of X-ray emission.
Spectral characteristics:-
$ The spectra constitutes of continuum component and line emissions (mostly Fe
K & ). The continuum, for a canonical black hole candidate, consists of two com-
ponents, a soft thermal (originating from a multi-coloured disc) and a hard non-
thermal (generally characterized by a power law). The Fe line is now considered
an essential feature of black hole X-ray spectroscopy.
$ The standard black hole candidates, viz. Cygnus X-1, exhibit two distinctly dif-
ferent kinds of behaviour, i) low-hard, with the soft X-ray flux low and spectral
behaviour comparatively harder, and ii) high-soft, with the soft X-ray flux higher
and the SED dominated by the softer X-ray. But the individual spectra of the
sources may differ dramatically from one another.
While these sources may or may not show radio flaring episodes which entail huge
blobs of matter being ejected (superluminally) from the system, recently it was realized
that non-thermal radio emission is a ubiquitous feature during the quieter phases. The
non-thermal emission forms a substantial fraction (5%–50%) of the energy budget. His-
torically, the radio and X-ray studies were done in a disjoint manner for these sources,
and the development of the models describing the accretion and ejection took place in-
dependently. In the last decade first efforts were made to create models to treat accretion
and ejection in a unified scenario, underlying the physical connection between the two.
Meanwhile, observational strategies were developed independently to monitor some of
these sources in the radio and X-ray at a regular basis to study the long-term behavioural
patterns in these systems, chiefly to observe the transient features of mostly transient
and a few persistent sources. The most methodical and consistently regular of these
strategies were the ones carried out in the radio at the Green Bank Interferometer (GBI)
operated by NRAO, and concurrently in the soft and hard X-rays by All Sky Monitor
aboard the Rossi X-ray Timing Experiment (RXTE - ASM, 2-12 keV) and the Burst and
Transient Sources Experiment aboard the Compton Gamma Ray Observatory (CGRO -
BATSE, 20-100 keV), respectively.
xi
Given the various diverse types of temporal variabilities and spectral characteristics
of the high energy emissions of the different microquasars, along with their different
types of radio emissions (from the outflow), there was no single consistent picture that
could provide a general scenario of the disc (accretion) – jet (ejection) connection in
these systems. Our investigation commenced with the next logical step of understanding
this connection at a broad scale across the diverse type of sources of this class exhibiting
their characteristic idiosyncratic behaviour, in order to provide a unified, consistent set of
observational features with the aim of developing a phenomenological model to unravel
the physical and geometrical structure of these X-ray binary systems. We achieved this
by carrying out a systematic correlation analysis among the radio, soft and hard X-rays,
for the sources Cygnus X-3, GRS 1915+105 and Cygnus X-1, using the available data
from the archives of GBI (2.2 & 8.3 GHz), RXTE-ASM (2-12 keV), CGRO-BATSE
(20-100 keV), during the long term steady hard states of these systems. These three
persistent X-ray as well as radio sources were the only ones monitored simultaneously
by these three observatories. The results of the correlation studies from these sources
was complemented by the observations reported for GX 339-4 (and also V404 Cyg),
scattered in the literature, to provide a qualitative self-consistent picture of the disc-jet
connection, using the Two Component Advection Flow (TCAF) model. In this thesis,
our emphasis lies in the detailed multi-band (X-ray and radio) study of the enigmatic
binary system Cygnus X-3, where we provide the complete evolution of the radio flaring
episodes of Cygnus X-3 driven by the X-ray spectral states in the system. In addition, we
report the temporal properties of the X-ray emission in this particular binary system and
provide a time scale of anti-correlation between the soft and hard X-rays in the system.
Of all the Galactic microquasars, Cygnus X-3 is one of the brightest in both radio
and X-ray bands, but one of the least understood of all binary systems. Located at a
distance of 9 kpc in one of the Galactic arms, it exhibits a binary period of 4.8 hours in
both X-ray as well as infra-red bands, while the radio emission doesn’t show the binary
modulation. The emission lines of He I and He II in the infra-red band with the absence
of any H lines suggest the presence of dense winds and the companion to be of the Wolf-
Rayet type, establishing the system to be a High Mass X-ray binary (HMXRB). The
nature of the compact object of the binary system is still not conclusively ascertained,
and a prime motive of undertaking the detailed X-ray spectral study was to glean the
observational features that may pertain to any particular class of compact objects, black
hole or neutron star.
To attain this goal we carried out a thorough and comprehensive analysis of the X-
ray emission of the source from the complete data set of the the RXTE archives available
publicly. This satellite observatory combined the dual advantage of the best X-ray tim-
xii Synopsis
ing capabilities (PCA) along with a very wide band X-ray spectral coverage (PCA and
HEXTE). Furthermore, RXTE has an extensive collection of data sets covering the mi-
croquasars at a (semi) regular basis. The analysis consisted of, downloading the raw data
of all the observations for this source, reducing and filtering the data in accordance with
the housekeeping files, extraction of lightcurve and the spectra keeping the diverse con-
ditions of the source as well as the observatory in consideration, creating the response
matrix for the obtained spectra, analyzing the lightcurves and the spectra using the desig-
nated software, and finally, interpreting the analyzed result in conjunction with, chiefly,
(contemporaneous) radio observations. The background spectra and lightcurves for PCA
were reduced from the model background event files provided by the PCA team, while
the two HEXTE detectors alternately point on and away from the source, measuring the
source and background flux with a duty cycle of 16 seconds. The PCA background has
a time resolution of 16 seconds, as a result the timing studies above '
!( keV can not
be done at a time scale less than this. All the X-ray spectra are obtained at the time
resolution of 16 seconds.
The long term X-ray spectral variation of this source shows two distinct states, high
and low, and correspondingly soft and hard (although the individual spectral model com-
ponents are different from a canonical high-soft and low-hard state of stellar mass black
hole candidates, characterized chiefly by Cygnus X-1). Correspondingly the radio emis-
sion can be broadly classified into two types, persistent and flaring. The persistent emis-
sion predominates the X-ray hard (as well as low) state. In the X-ray soft (as well as
high) state the persistent radio emission is quenched, interspaced by the flaring events,
both minor and major. The major radio flares are among the brightest in the Galaxy, oc-
curring only during the X-ray high (soft) state. The distinct features of the X-ray spectra
of this source are: 1) very high absorption in the soft X-ray regime, probably by the dust
and/or halo engulfing the binary system, whose most likely origin is in the wind from the
Wolf-Rayet companion, which obscures the thermal blackbody spectra, 2) above 5 keV,
the continuum is complex and a mixture of model components consisting of thermalised
Comptonization and a power law is needed to model the spectra in the low (hard) state,
while the high (soft) state, in most cases, shows a very strong multicoloured disc black
body emission and a Comptonization component, as a result the unusual hump in the
5-30 keV region present in the low (hard) state is absent in this high (soft) state, 3) the
hard X-ray tail, which may be characterized by power law, is present in both the low
(hard) and high (soft) states, 4) EXOSAT, ASCA & Chandra observations have revealed
three Fe lines (6.4 keV, 6.7 keV & 6.9 keV) along with two absorption edges (7.1 keV &
9.1 keV) present in the source, but the resolution of the RXTE-PCA is not good enough
to resolve these to desired accuracy.
xiii
Detailed analyses of the X-ray spectra show a definite pivoting at about 12-20 keV
in the hard state. In the soft state the picture is different with the thermal emission from
the accretion becoming the prominent feature of the SED, in most cases. There is a defi-
nite causal relationship between the X-ray spectral evolution and the radio flaring events,
which may be explained by classifying this state into three sub-states: 1) the radio qui-
escent phase, in which the thermal multicoloured disc black body and the Comptonizing
component are equally strong, 2) pre-radio flare, during which the Comptonizing com-
ponent becomes vanishingly small ( )*
!,+ ), suggesting the evacuation of the central
Compton cloud, resulting in the flare, which may occur at a time scale of less than a day,
3) post-radio flare, in which the succession of radio flares, both minor and major, are
brought to an end by a change in the X-ray spectrum, with the spectral shape harden-
ing and the thermal disc black body component vanishing, following which the source
makes transition into the low (hard) state. Unless this spectral hardening takes place,
the succession of flares, minor ones interspaced by a few major ones, continues with
the Comptonizing component remaining significantly less than the thermal component.
Thus, a picture of a complete evolution of the occurrence of radio flares in relation to the
X-ray spectra is obtained, for the first time for this source. These spectral features tilt the
balance of evidence in the favour of the compact object being a black hole, although a
definitive statement can only be made after ascertaining the mass function of the system.
To obtain the long term behavioural pattern of this source we carried out a systematic
correlation analysis among radio (GBI. 2.2 GHz), soft and hard X-ray (RXTE - ASM:
2-12 keV & CGRO - BATSE: 20-100 keV) emissions, during the period for which these
three observatories were simultaneously monitoring the source, as mentioned before.
We employ the Spearman’s Partial Rank correlation test for the correlation among the
three variables. In the high (soft) state, the long-term correlation results are not very
significant, due to the complex evolution of the radio as well as X-ray emission, as
explained earlier. In the low (hard) state, the soft X-ray is very strongly correlated to the
radio emission, while the hard X-ray is anti-correlated to both soft X-ray as well as radio
emission. This interesting correlation result is explained by the pivoting in the spectra,
at about 12-20 keV, being correlated to the radio emission.
The short term temporal properties of the X-ray emission of Cygnus X-3 is perhaps
the least studied aspect of this source, also among all other microquasars this source’s
X-ray temporal properties are least explored. The principal reason for this is the fact that
the binary modulation of 4.8 hours is too strong and smothers any other variability pat-
tern that may emerge out of any analysis in the study of the shorter time scale variability.
Hence, our first step was to correct for the binary modulation using the binary template
for a quadratic ephemeris which is amazingly stable for more than two decades of obser-
xiv Synopsis
vation. The long term RXTE-ASM monitoring data, containing all sorts of variabilities
including the period of major radio flares, folded by the template shows very meagre
residue, proving its applicability to correct for the binary modulation. The latest tem-
plate was obtained from the ROSAT, ASCA, BeppoSAX, RXTE and IXAE pointing and
monitoring, which was used to correct for the binary modulation in the flux variability
in the light curves obtained at all the different X-ray energy bands, after making ap-
propriate scaling adjustments, for all RXTE-PCA pointed observations analyzed in this
thesis. The binary correction using this template is very good at the rising and falling
phase, highlighting the small variations which were otherwise smothered by the binary
modulation of the ephemeris, whereas at the peak the smooth nature of the template is
not always successful to correct the generally random fluctuations in the emission. The
correlation among the radio, soft and hard X-ray emissions remains the same after the
correction for the binary modulation, and detailed tests show that the anti-correlation
time-scale between the soft and hard X-rays, due to the pivoting of the spectra, is less
than a day.
The power density spectra of this source has a feature distinct from its counterparts,
the shifting of the spectra towards low frequency. In this pattern of temporal variability it
resembles less like other Galactic microquasars and bit more like the massive AGNs with
a central massive black hole. One may reconcile the absence of power in the high fre-
quency regime to the reprocessing of the X-ray photons in the dust and/or halo engulfing
the system, reducing the amplitude of the fast X-ray variability.
The most interesting result of the X-ray timing properties reported of the source, by
us, is the anti-correlated time lag of 400–900 seconds between the soft and hard X-rays
in the low (hard) state of the system. Unless corrected for the binary modulation, this
delay is not observable. Also, this delay is observed only in the hard (low) states, the non
flaring soft (high) state doesn’t show any such delay. This anti-correlated delay between
the soft and hard X-ray flux, with the hard X-ray lagging, provides the dynamical time
scale of the pivoting of the spectra, in the hard state. It is noteworthy that from the
long term correlation among the monitoring data in the radio, soft and hard X-ray bands
we had predicted an anti-correlation time scale shorter than a day. This is the first such
observation for this source and will provide stringent constraints on the accretion models
for microquasars in general.
The generalization of the X-ray:radio correlation, during the low (hard) state, for
different types of microquasars, was done by repeating the correlation study of the long
term radio(GBI, 2.2 GHz), soft and hard X-ray (RXTE - ASM: 2-12 keV & CGRO -
BATSE: 20-100 keV) monitoring data of two more sources with apparently diverse be-
havioural pattern, viz. GRS 1915+105 & Cygnus X-1, and collating the existing results
xv
of another black hole candidate GX 339-4. It was successfully demonstrated that all
these three sources, plus Cygnus X-3, show a very similar behaviour during the low
(hard) state, i.e. pivoting of the X-ray spectra correlated to the radio emission, with the
radio emission being higher in the comparative softer state (within the bounds of the hard
state). The only difference lies in the pivoting energy of the individual sources. GRS
1915+105 has a pivot point between 20-30 keV, Cygnus X-1 between 50-100 keV and
GX 339-4 has it at -/. keV. We also established that the pivot point moved further
into the hard X-ray / low gamma ray regime as the intrinsic luminosity of the sources
in the soft X-ray band weakened. Furthermore, continuing the X-ray:radio association
beyond the X-ray state transition it is revealed that the radio emission is quenched, to
varying degrees, for all the sources (during the non-flaring periods). Therefore, for the
first time, a uniform behavioural pattern was found of the radio emission and correlated
X-ray spectral emission evolution, encompassing various microquasars with apparently
different characteristics. Most remarkably, all the sources (along with another black hole
candidate, V404 Cygni), were found to show a linear monotonic increase of radio emis-
sion with the soft X-ray flux, spanning a 5 orders of magnitude variation in the intrinsic
luminosities, with the radio emission suppressed for the intrinsically weaker X-ray emit-
ters (during the non-flaring state). Arguing that radiation in both the bands, radio as
well as X-ray, are unlikely to originate from a single mechanism (like synchrotron emis-
sion), we invoked the TCAF model to explain the accretion-ejection behaviour in these
systems.
According to the TCAF model, the Compton scattered X-rays in a black hole source
originate from a region close to the compact object, confined within the Centrifugal
Boundary Layer (CENBOL). The X-ray spectral shape in various ‘states’ of the source
essentially depends on the location of the CENBOL. At low accretion rates, the CEN-
BOL is far away from the compact object and the X-ray spectrum is dominated by a
thermal (-non thermal)-Compton spectrum, originating from the high temperature re-
gion within the CENBOL. In the transition state, the CENBOL comes closer to the
compact object and the CENBOL can sometimes give rise to radial shocks, causing in-
tense quasi-periodic oscillations. In the high state, the increased accretion rate produces
copious photons in the accretion disc which cool the Compton region, giving rise to very
intense disc blackbody emission along with bulk motion Comptonization (a power-law
in hard X-rays with a photon index of 2.5). The outflow rate is found to be a mono-
tonic function of the compression ratio of the gas at the shock region. In this scenario,
at low accretion rates, the CENBOL is far away from the compact object, a weak shock
can form with low compression ratio, giving low and steady outflow. If this outflow
gives rise to radio emission, one can expect a relation between the radio emission and
xvi Synopsis
the X-ray emission. In this state (off state to low-hard state), an increased accretion
rate increases the overall amount of energy available to the Comptonizing region and
hence increasing the X-ray emission. The CENBOL location would be pushed inward,
increasing the compression ratio (and hence increasing the radio emission) and also can
increase the temperature and optical depth of the Comptonizing region, thus giving rise
to a pivoting behavior at hard X-rays as seen in Cygnus X-1 and GX 339-4. At increased
accretion rate, the CENBOL can come closer to the compact region, giving the spectral
and radio properties as seen in GRS 1915+105 and Cygnus X-3. For a given accretion
rate the compression ratio, after reaching a critical value (with the shock region coming
correspondingly closer to the event horizon), causes the source to transit into the high-
soft state state, for which the radio emission is progressively suppressed, clearly seen in
Cygnus X-3, GRS 1915+105, Cygnus X-1 and GX 339-4.
During the flaring states, the TCAF model predicts ejection of the central Compton
cloud resulting in the radio flares, with the size of the ejected blob determining the flux
in the radio band. This model further predicts the absence of radio flares during periods
dominated by bulk motion Comptonization, when the infalling matter falls directly into
the event horizon due to its bulk motion, resulting in quenched radio emission. The
observational evolution of the X-ray spectra with the radio flaring of Cygnus X-3 may
be explained by transition between these two states, during the high (as well as soft)
state.
Thus, in this thesis we provide a comprehensive phenomenological modeling of the
class of Galactic X-ray binary system exhibiting jets, called microquasars, with very
special emphasis on the enigmatic source, Cygnus X-3. The future works should en-
tail a detailed quantitative modeling of the complete behavioural pattern of the source
Cygnus X-3. Then one has to reconcile the qualitative as well as quantitative features of
the accretion–ejection mechanism in these systems with other cosmic sources showing
accretion as well ejection at diverse physical scales, namely the AGNs at one and the
YSOs (Young Stellar Objects) on the other.
Chapter-wise organization of the thesis:-

$ Chapter 1 gives introduction to astronomical high energy phenomena, with em-
phasis on accretion discs as sources of high energy emission. It further discusses
the cosmic black hole systems: stellar mass and supermassive, and introduces the
concept of microquasars and their importance, laying down the aim of this thesis.
$ Chapter 2 deals with X-ray detectors and techniques of instrumentation as well
as radio astronomy observatories. The modern X-ray observatory RXTE and its
xvii
data analysis techniques is illustrated in detail, this being the principal observa-
tory whose data is analyzed in these thesis. Other high energy observatories, viz.
CGRO is also mentioned. A brief introduction to radio astronomical observato-
ries, namely GBI and GMRT are provided.
$ Chapter 3 starts with general introduction to Cygnus X-3. Thenceforth it pro-
vides the detailed spectral analysis and features of this source, providing an X-ray
spectral behavioural evolution during the radio flares.
$ Chapter 4 demonstrates the temporal properties of Cygnus X-3. Here the long
term radio : X-ray correlation tests and results are presented. Further, a recipe is
provided for the correction of X-ray binary modulation with the given ephemeris.
Thereafter the power density spectra and time lag between soft and hard X-ray
emissions is obtained.
$ Chapter 5 provides the generalized accretion - ejection mechanism of micro-
quasars by extending the long term radio : X-ray monitoring to other persistent
sources. Uniform behaviour of X-ray spectral shape with radio emission and uni-
versal correlation is obtained in both low (hard) as well as -non flaring- high(soft)
states.
$ Chapter 6 describes the TCAF model, elucidating a qualitative phenomenological
explanation of our multi-band data analyses results, providing a uniform picture
of the disc - jet connection in the microquasars.
$ Chapter 7 summarizes the results and the phenomenological models and outlines
future directions for further study of the accretion - ejection mechanism.
Chapter 1
Introduction
An X-ray is a quantum of electromagnetic radiation with an energy, to an order of magni-

tude approximation, some 1000 times greater than that of optical photons. Traditionally,
the soft X-ray band is defined as the energy range 0.5 – 12 keV (corresponding to wave-

length of 25 – 1 ), the hard X-ray extends to 50 keV and the energy range beyond
it till a few MeV is regarded as soft 0 -rays, although this classification is not very strin-
gent. High energy astronomy pertains to the observation of the sky in this regime of the
electromagnetic spectrum. The study of cosmic sources at these high energies of X-rays
and 0 -rays began only in the early 1960’s, after the serendipitous discovery of the low
mass X-ray binary (LMXRB) Sco X-1, which houses a neutron star (the compact ob-
ject) and a low mass optical companion in the main sequence. These early observations
were made by rocket flights, which provided only a few minutes of data. This field of
astronomy matured with the advent of use of balloon flights and the first generation X-
ray astronomy satellites during the 1970’s. Post 1980’s the enormous advancement and
progress in the computational and technological capabilities completely revolutionized
this branch of astronomy so much so that currently high energy astronomy is, perhaps,
the most ‘happening’ branch of astronomy and astrophysics.
1.1 High energy physical processes and phenomena

Simple extrapolation from the optical regime suggests that, considering a thermal origin
of the physical processes giving rise to these high energy X-ray & 0 -ray, the temperature
of the radiating matter should be of the order of
1 –
, K for X-ray photons and greater
for the 0 -ray emission. The fundamental physical mechanisms which give rise to high
energy emissions from a thermalised distribution of matter are few, viz. thermal black
1
2 Chapter 1. Introduction
body radiation, bremsstrahlung, Compton scattering. Soon, however, it was discovered,

mainly from the supernova remnants, that non-thermal physical processes also play very
important part in these high energy emission. Such physical processes may also involve
bremsstrahlung and Compton scattering, in addition to synchrotron emission.
Black body radiation. The spectrum of thermalised emission of a black body is given
by (see, for eg. Rybicki and Lightman 1979)
:
23 (<
546798 ; = 3?>@BA 8
: (1.1)
C
where 4ED temperature of the body, D frequency of the radiation and < , F ,
; D Plank constant, Boltzmann constant 8
and velocity of light, respectively. A
temperature of
K will render most of the radiation in the X-ray band.
Thermal bremsstrahlung. Free-free emission occurs with the interaction (accelera-

tion) of energetic electron with the near stationary nuclei, with a fractional depo-
sition of the energy into the electromagnetic radiation. For a hot thermal plasma
the spectrum is given by
2 :?JKJMLON
XZY
GIH R
P LQ 46 (1.2)
8 FS4UTWV 81[
J]L
where 4\D temperature,
J D electron density, HED atomic no. of
Y the con-
cerned nuclei, D space density of nuclei in laboratory frame and 4^_D
81[
the Gaunt factor which gives the correction due to the collision parameters after
integrating over velocity.
Synchrotron radiation. The process of radiation due to acceleration of charged par-

ticles, normally electrons, in a region containing magnetic field is known as syn-
chrotron radiation process for relativistic velocities of the electrons (in the non-
relativistic limit the process is known as cyclotron emission).
2 N
X
X bBe
Ga`Rbdc (1.3)
8 T V
V 8
where `D magnetic field and fgD spectral index
J when the spectrum is a power-
law of electron energy distribution, given by ih_jGkhmlon .
1.1. High energy physical processes and phenomena 3
Comptonization. The scattering of an energetic photon with a stationary electron is

known as Compton scattering, where fractional energy is transferred to the elec-
tron. The inverse of this process involves transfer of energy from energetic elec-
tronto a low energy photon (inverse Compton scattering), and it is a very important
physical mechanism generating high energy radiation in cosmic sources. The in-
teraction between ensembles of photons and electrons is known as ‘Comptoniza-
tion’. The evolution of the photon and electron energy distributions subject to
Comptonization is governed by the Kompaneets equation (Kompaneets 1957, also
see Rybicki and Lightman 1979)
prq p prq
N FS4
r
u x N qzyq : y
p#st 7 PR; : T u : p uwv p u
T{
(1.4)
st}| q LB~ A ; s
where D time measured in units of mean time between scat-
terrings. Eq. 1.4 may be expressed in terms of the Compton parameter in an
integrated form p#q p prq

u x N qzyq : y
p :
7 u p uv p u (1.5)
T{
L L P L ; : ~ A J LBo q
where u 7< FS4 , 7FS4 and is the occupation number
8r
which is defined as the number of particles or photons per state. If the energy
y 3
density of isotropic radiation in the 3 frequency interval to is , then the
8 8 8
number density of photons is < and the occupation number is given by
8r 8
3 Q
q ;
7 (1.6)
8 < Q
8
The value of the Compton parameter determines the type of Comptonization i.e.
saturated or unsaturated. General solutions of the Kompaneets equation need to be
obtained numerically, although it is possible to find analytic solutions, under some
approximations. There have been many such attempts in this direction (e.g. see
Sunyaev and Titarchuk 1980, 1985, Lightman and Zdziarski 1987, Titarchuk 1994,
Poutanen and Svensson 1996, etc.) and accordingly many different Comptoniza-
tion models, which are valid under various sets of approximations, are available
for modeling the X-ray spectra from cosmic sources.
X-ray line emission. Excitation of the lowest-shell electrons (K,L, etc.) into the un-
bound state results in emission of radiation in the X-ray band when the empty
levels are filled by transition form outer shells. However, in order to excite these
low shell electrons the excitation energy need to be very high as well, which is
only possible in environment with extremely high temperature or with extremely

high density high energy photon fields. Thus, the X-ray line spectroscopy provides
us a powerful method for plasma diagnostics under extreme conditions.
1.1.1 Accretion disc as a source of high energy emission

The most common physical phenomenon that gives rise to emission in the X-ray band is
accretion of matter onto compact objects. This process involves accumulation of diffuse
gas or matter onto the compact object under the influence of gravity, and is expected to
be responsible for the observed properties of a wide range of X-ray sources from X-ray
binaries to Active Galactic Nuclei (AGNs) (Frank et al. 1992).
A mass P being accreted to a body of mass and radius will lose potential
, which, if converted to electromagnetic radiation, will cause the system to
energy _
have a luminosity of
P
PR; :1! :
7 7 7k PR; (1.7)
(
where 7( ; : D Schwarzschild radius, and R7: ! gives the efficiency

of the accretion process. Thus the efficiency of energy conversion due to accretion sim-
ply depends upon the compactness of the accreting body. For a white dwarf star with
7 , ! 7.m
x m and m
m ¡7.m
1l x . Correspondingly
for 1 neutron star with R = 15 km, 7E
1 , and for black holes ranges from
0.06 (Schwarzschild blackhole) to 0.426 (maximally rotating Kerr blackhole). Thus,
black holes, particularly, maximally rotating ones, are the most powerful energy sources
in the Universe and accretion is the process by which the energy is released. Despite
this high efficiency of emission of electromagnetic energy, the balance between the out-
wardly directed radiation pressure (obtained from Thomson cross-section) and inwardly
directed gravitational pressure limits the luminosity to a limiting value, called Eddington
luminosity, which is given by
j¢ Q
( ~ P ; Q N
7 A n 7£
.W¤
(1.8)
T ¥
It should be noted that this is derived assuming spherically symmetric geometry. It

is possible to exceed the Eddington limit by adopting a different geometry, but not by a
large factor. Further, the limit applies to steady-state situations and in none steady-state
situations like supernova explosions the Eddington limit can be exceeded by a large
margin.
1
the efficiency for nuclear burning in neutron stars is only ¦§¡¨© ¨B¨?ª
1.2. Black hole sources: stellar mass and supermassive 5
Thus observation of accretion of matter onto compact objects provides a unique op-
portunity to investigate the most powerful energy sources in the Universe. Two such
classes of objects exist in the universe known to us:-
1. Active Galactic Nuclei. These are supermassive blackholes, with the mass of
compact object «
lr¬ , at the centre of galaxies. The blackhole accretes
mass from the interstellar medium around it, disrupting the stellar structures and
consuming the matter within the envelope of its horizon.
2. Galactic X-ray binaries. These binary system have stars with a compact object,
neutron stars or blackhole candidates, accreting matter from the companion star,
which is normally in the main sequence.
1.2 Black hole sources: stellar mass and supermassive

Since the efficiency of conversion of gravitational energy into electromagnetic radiation
is more for blackholes, as compared to neutron stars, they provide a unique opportunity
to understand the behaviour of matter under extreme physical conditions. Also, the phe-
nomena of accretion makes the black hole ‘visible’ to observers from a distance, which
otherwise doesn’t allow any particle, including light photons, to escape and thus render
the direct observations of these sources impossible. Therefore these sources offer the
opportunity to experimentally verify the general theory of relativity directly, and hence
put to test the underlying fundamental principles of our understanding of the physical
structure of the universe.
It is presently established that blackholes exist in two classes, supermassive and
stellar mass blackholes.2 As mentioned above, the accretion phenomenon is ubiqui-
tously observed in Active Galactic Nuclei (henceforth referred to as AGNs) as well as
the Galactic X-ray blackhole binary systems. The more interesting observational fea-
ture seen in these sources is that accreted matter is more often than not ejected out in
the form of a jet perpendicular to the accretion disc. In fact, the observations, in the
radio band, of outflowing conical jets from the core of the galaxies, often exhibiting su-
perluminal (with apparent velocity greater than that of light) expansion, were reported
before the phenomenon of accretion was inferred from the observational studies, for
the extra-Galactic sources. Quasars, a sub-class of AGNs, were discovered in the radio
band of electromagnetic radiation during the era of the very early discovery of X-ray
2
recently there are growing arguments regarding a third kind, the intermediate mass blackhole, but a
discussion on this topic is beyond the purview of this thesis
sources. These were subsequently identified to be accreting supermassive black holes

6I
of galactic scales with outflows in the form of a jet observable in the
radio band by virtue of the physical mechanism of synchrotron emission. Therefore
a paradigm of accretion being related to ejection was gradually developed, although it
was not observable in the radio quiet AGNs. The discovery of Galactic X-ray binaries
exhibiting (superluminal) radio jets, with both physical and temporal (variability) scale
roughly at 6 orders of magnitude less than those of quasars, led to the notion of ubiq-
uitous presence of outflow in the form of collimated jets in accreting blackhole systems
and low magnetic field ( E
G) neutron stars, lending them the terminology of mi-
croquasars. There are obvious advantages of observing microquasars (over quasars) in
order to understand their physical and geometrical structures:-
1. Microquasars are located within the Galaxy, the astronomical equivalence of our
own backyard. Their proximity enables the study of both the lobes of the out-
flowing jet more feasible and practical. The measurement of both the components
as well as evolution of the flux and size ratio of the two components are essen-
tial to attempt a detailed modeling of the jets. Further, the improved accuracy of
the measurement of the jet properties provide more accurate values of the basic
parameters of such astrophysical structures.
2. The characteristic dynamical time scales in the flow of matter are proportional to
the black hole mass and any variability time scale of hours to days of microquasars
correspond to analogous phenomena with duration of hundreds to thousands of
years in AGNs, assuming that the same fundamental physical processes underlie
the behaviour of these sources. Therefore monitoring the microquasars for a few
days may sample phenomena not possible to observe in quasars.
Over the last few years of intense observations one disadvantage has become glaring in
the case of the microquasars:-
1. The dynamical time scales of the fast variability, milli-second variation for micro-
Q
quasars and correspondingly
s for AGNs, do not provide enough photon count
for any meaningful statistical analysis, because the present day detectors do not
possess the capability to capture enough photons necessary during the small time
span of the very fast variability.
Despite this one deficiency, in the recent past there has been tremendous upsurge in the
observational study of the microquasars, and understandably so.
1.2. Black hole sources: stellar mass and supermassive 7
Fig. 1.1: The similarities between quasars and micro-quasars. Both systems contain similar ba-
sic ingredients viz. (1) a central black hole, (2) an accretion disk and (3) collimated
jets of relativistic particles. The difference between them lies in the mass scale. Micro-
quasars have stellar mass blackhole of the order of a few solar mass and the jets travel
to a distance of a few light years whereas in quasars the black hole mass is supermas-
sive of the order of ¯®±°j²³¯®´ solar mass and the jet can travel to a distance of a few
million light years. This figure is taken from Mirabel and Rodriguez (1998)
1.3 Microquasars: general properties and behaviour

Microquasars, generally black hole candidates (BHCs), mimic, at a much smaller scale,
the main astrophysical attributes of a quasar: general relativistic accretion identified
by the X-rays and gamma rays from the surrounding accretion discs, and the special
relativistic outflows in the form of collimated jets with low opening angles ( µ 15 ¶ )
observed by means of their synchrotron emission. Of the 200 Galactic X-ray binaries
catalogued so far, about 20 are radio loud, half of which show evidences of radio jets,
a few of them superluminal (eg. GRS 1915+105, GRO J1655-40). These X-ray binary
sources have some common salient characteristics which may be enumerated as follows:-
Structural characteristics:-
$ They consist of one compact object (generally a black hole candidate) and one
normal star, generally from the main sequence.
$ The compact object accretes matter from the companion, via an accretion disc.
The donor may lose mass through Roche lobe overflow or via stellar wind. The
extent of the inner disc is a function of time (and probably accretion rate), the
explanation of the variability generally depends on the particular model adopted
to explain the X-ray characteristics.
$ The outflow of matter takes place via a collimated beam, visible in the radio, at
times in the infra-red or, arguably, even in the X-ray. The conical jet has a small
opening angle ( µ 15¶ ) directed perpendicular to the accretion plane. This system
may show precessional movement in some cases.
From the observational point of view, the study of the behavioural pattern of these
sources, in the various electromagnetic bands, may be classified into three different types
of analysis: 1) image analysis, which gives the (extended) spatial information about the
source, 2) temporal analysis, which gives the variability of the source with respect to
time, and 3) spectral analysis, which gives the pattern of the emission with respect to the
energy (or frequency / wavelength), and provides the best analytical tool for identifying
the physical processes giving rise to the emission. The spectral analysis identifies the in-
dividual physical mechanism which gives rise the electromagnetic emission. A complete
understanding requires the combined understanding of the all the three analysis. Some
basic generic patterns of the temporal and spectral characteristics (in the X-ray regime)
of the microquasars are highlighted below:-
1.3. Microquasars: general properties and behaviour 9
X-ray temporal characteristics:-

$ The X-ray light curve may show a variety of diverse variabilities, even for a single
source, depending on the particular states or transitions among them during the
period of observation.
$ The power density spectra (PDS) shows, typically, a power law dependence with
a positive index in the region of 0.01 – 1 Hz, flat spectra for the next decade of
frequency range, followed by a power law decay (i.e. negative index) of power in
the 10 –100 Hz range. The PDS of the neutron stars is generally shifted towards
the higher frequency region by an order of magnitude.
$ Various types of quasi periodic oscillations (QPOs) are observed in most of the
microquasars, prominently in the low-hard state of X-ray emission.
X-ray spectral characteristics:-

$ The spectra constitutes of continuum component and line emissions (mostly Fe
K & ). The continuum, for a canonical black hole candidate, consists of two com-
ponents, a soft thermal (originating from a multi-coloured disc) and a hard non-
thermal (generally characterized by a power law). The Fe line is now considered
an essential feature of black hole X-ray spectroscopy.
$ The standard black hole candidates, viz. Cygnus X-1, exhibit two distinctly dif-
ferent kinds of behaviour, i) low-hard, with the soft X-ray flux low and spectrum
comparatively harder, and ii) high-soft, with the soft X-ray flux higher and the
spectrum dominated by the softer X-ray. But the individual spectra of the sources
may differ dramatically from one another.
1.3.1 Transient X-ray blackhole binaries

Most of the Galactic blackhole candidates are transient in nature, which implies that
the source is not continuously detectable by the X-ray astronomical detectors given the
current sensitivities of these instruments (with improving sensitivity it is increasingly
possible to probe into the “off” state of X-ray emission of these sources). The iden-
tification of the optical counterparts during the “off” state of the X-ray emission has
recognized many new blackhole candidates by virtue of the mass function of the binary
system which is given by ·
Q Q QÀ¿
»K¼d½¾
£7¸º ¹ 7 y : (1.9)
( »

Fig. 1.2: The long term monitoring of transient X-ray binary blackhole candidates by RXTE–
ASM. These sources are also the Galactic superluminal sources.
where & » D masses of the companion star and the compact object respectively,
¿
D inclination angle of the orbit of the binary system, D period of the binary system
¸
and D velocity amplitude obtained from the Doppler shifts of the spectral lines of the
¹
companion due to the orbital motion. These sources are normally Galactic sources that
brandish superluminal motion, in the form of outflowing jets. Normally these superlu-
minal expansions occur during the X-ray “on” state. The X-ray bright state may persist
from less than a day to more than a decade (see Figure 1.2).
1.3.2 Persistent sources: canonical states of X-ray emission

These sources, in obvious contrast to the transient ones, never go “off” in the X-ray
band. Always ‘visible’ (Figure 1.3), they present the opportunity to study the X-ray
emission behaviour in the various states in which they may exist, evolving from one to
another in a cyclic process. There are very few persistent established Galactic blackhole
candidates, whose mass function is known, viz. Cygnus X-1, GX 339-4. As evident
from Figure 1.3, they exhibit prolonged episodes of high and soft states, whose spectral
shapes are intrinsically different and hence are classified into different states (Tanaka
1.3. Microquasars: general properties and behaviour 11
Fig. 1.3: The long term monitoring of persistent X-ray binaries by RXTE–ASM. Cygnus X-1
(top panel) is an arche-typical blackhole candidate, and Cygnus X-3 (bottom panel).
and Lewin 1995), the “high-soft” and “low-hard” states. The suffix ‘soft’ and ‘hard’
are added to emphasize the X-ray spectral shape during the concerned states. With
detailed studies of different Galactic BHCs the presence of other states like intermediate
and very-high states have emerged. In general the spectra is described by two continuum
components, 1) soft thermal component believed to originate from a multi-coloured disc,
and 2) hard non-thermal component characterized by a power law. Figure 1.4 shows the
wide band X-ray/0 -ray spectra of Cygnus X-1 in three different canonical states which
clearly demonstrate the differences in the spectral properties of these X-ray states.
High-soft state. In this state the soft X-ray flux is higher and the spectral energy dis-
tribution (SED) is dominated by the softer X-ray flux below 10 keV, i.e. of the two
continuum components describing the spectra the multicoloured disc black body com-
ponent dominates the SED, which most likely originates in the inner regions of the ac-
cretion disc. The powerlaw component is, at times, regarded as the blackhole accretion
signature (Tanaka 2000). The temporal characteristic feature of this state is the strong
aperiodic variability over a wide range of time scales from milliseconds to minutes re-
flected in a typical power-law shape of the power density spectrum.
Soft Intermediate
Hard
Fig. 1.4: The various canonical states of Galactic blackhole candidates, characterized by the
arche-typical BHC Cygnus X-1. The figure obtained from Gierliński et al. (1999)
Low-hard state. In this state the soft X-ray flux is low and spectral behaviour compar-
atively harder, i.e. the powerlaw component dominates the SED. This powerlaw com-
ponent is commonly regarded as a signature of Comptonization process in the system.
The temporal behaviour is characterized by the presence of quasi-periodic oscillations
(QPO).
Other states. During the intermediate state the shape of the SED is intermediate be-
tween the high-soft and low-hard states described by the two continuum components of
the canonical states, while the lightcurve shows the presence of QPOs. The very-high
state has the SED with a shape similar to the intermediate state but with total luminosity
higher than that of the high-soft state. For completeness sake one may also mention the
“off” state, during which the transient sources are not detectable.
1.4 Accretion in X-ray binaries

Accretion, by definition, means accumulation of diffuse gas or matter onto some ob-
ject under the influence of gravity. The importance of this phenomenon as a source of
1.4. Accretion in X-ray binaries 13
Fig. 1.5: The two types of accretion process in X-ray binaries. Left panel: accretion via stellar
winds, more common in HMXBs. Right panel: accretion via Roche-lobe overflow,
predominant in the LMXBs.
highly energetic emission was realized with the discovery of X-ray binary systems and
subsequently applied to interpret the properties of cataclysmic variables and AGNs.
The Newtonian dynamics states that a test particle with initial velocity Á Â falling
onto a massive object due to gravitational attraction (conserving angular momentum) is
described by :
: y < ( :
Á : 7* Á Â
(1.10)
Ä : Ä
where <Ã7 Á D specific angular momentum, and are polar coordinates. Hence
[
for a non-zero < the particle can never reach 7k . Using Einstein’s general relativistic
treatment the dynamics of the same particle in the Schwarzschild metric is given by
: :
: y < ( ( Å
< : ; :
Á : Q ; :
7 i
Æ C

(1.11)

where < & Æ are constants of motion, Á 7/Ç ÇrÈ where È is proper time, while : t < X and
: :
iÆ
are constants É< and Á Â (in Newtonian case). Due to the term l ÊÌË X the

particle is allowed to fall to 7Í if <£E( Î
; . Hence, relativistically the effect of
gravity is ‘increased’ close enough to 79 overcoming the centrifugal barrier. Also,
there is a last stable orbit (Schwarzschild metric) with radius 7I. , beyond which the
circular orbits cease to exist and the particle spirals rapidly towards the singularity (see
Longair 1994, for a detailed treatment).
Accretion flow is essentially of two types,
1) Bondi type flow, which has (quasi)
spherical geometry with angular momentum ) , and 2) disc type flow, where the
WÏ

inflow is flattened into a disc with high angular momentum, - , where D the
WÏ WÏ

angular momentum at minimum stable orbit . . Even for the first case it can be shown
that for a spherically symmetric infalling cloud with axisymmetric rotation the infalling
gas passes through a shock at the equator dissipating the kinetic energy perpendicular
to the equatorial plane, resulting (in most cases) in the formation of a disc (Hartmann
1998). This implies that even if the accreting matter is falling at an arbitrary angle with
respect to the general rotation axis, the velocity components parallel to the rotation axis
gets canceled and the matter eventually settles down in the plane perpendicular to the
rotation axis forming an accretion disk. In X-ray binaries, the accretion of material
happens in two modes (Figure 1.5):-
$ Wind Accretion. More common in the High Mass X-ray Binary systems (HMXB),
the companion loses mass in the form: of stellar wind. The material of the wind
t
within the capture radius, 7 X Ð , where » D mass of X-ray emitting
Ò Ñ
compact object, and Ó³D velocity of wind relative to the » , is trapped by the
gravitational potential and accreted in.
$ Roche-lobe accretion. Predominant in Low Mass X-ray Binary system (LMXB),
the companion expands to fill the Roche-lobe and matter flows through the inner
Lagrangian point to enter the gravitational pit of the compact object. The conser-
vation of angular momentum ensures the formation of an accretion disc (Frank et
al. 1992). In this case the efficiency of the disc formation is definitely higher.
The necessary conditions for the accretion to occur are two-fold, firstly, energy needs
to be dissipated as the matter falls into the gravitational well, and secondly, the angular
momentum needs to be transferred outwards. Without fulfilling these two conditions,
the matter will remain in a circular ring (keplerian stable orbit) around the compact
object. The theoretical difficulty lies in accounting for a physical mechanism of angular
momentum transfer as, the only obvious agent available, the ordinary fluid viscosity
is far too weak to be a significant factor. Shakura and Sunyaev (1973) took the first
step in developing a phenomenological understanding of the process by introducing an
anomalous internal stress (vertically averaged along the disc axis), Ô given by
Ô 7a& (1.12)
¸
where D vertically averaged pressure, &D normalized proportionality constant, such
¸
that U)k&)
. This idea perpetuated the concept of anomalous viscous stress, defined
by the parameter & , as the agent necessary to transfer the momentum and energy. 3 A
Sunyaev later commented that he was not absolutely convinced about the introduction of the Õ param-
3
eter. Little did they realize the forthcoming impact of this Õ

commonly used modification of the Shakura-Sunyaev : (SS) prescription: is to assume a

kinematic viscosity of the form &
; B
× Ø & ; × , where ; × D the ver-
8³Ö Ö ÙÛÚ Ö ¸ ±Ü
tically averaged sound speed ( is vertically averaged density, Ø 7 Ø iD vertical
Ü
half-thickness of disc, and D angular velocity of test particle circular orbits (Kep-
ÙÝÚ
lerian angular velocity). When inserted into the standard viscous form of the stress, this
gives
Ç Ç
Ô 7 Ù 7a& Ù (1.13)
Ü8 Ç# ¸ Ç#
ÙÑÚ
Here 7 iD actual angular velocity in the flow, which may differ from if the
Ù Ù Ù Ú
flow is not completely geometrically thin. If , then within factors of order of
Ù Ù6Ú
unity, i.e. the disc is geometrically thin equation 1.13 gives the same stress as equation
1.12, with the only difference that the stress here depends explicitly on the shear of the
flow, just like an ordinary viscous flow (for detailed explanation see Longair 1994, Frank
et al. 1992, Blaes 2003). The conditions for the validity of the thin disc are that 1) at any
radius , the vertical height Ø\Þ , 2) rotation velocity Ó1ß_à the sound speed ;!× within
the disc (i.e. internal pressure gradients should not inflate the disc). In such : cases the
accretion flow is extremely turbulent with the Reynolds number á9 £
.
Carrying on with the recipe of Shakura and Sunyaev (1973) for the thin disc, the
luminosity and the emission spectra of the disc are given by (Longair 1994)
Â N
h P Á ä
7â Êã s ( 7 (1.14)
T ( ä
> ÊæåÀçÌè > >
2 Q urx Q íì?î1ï u C
l u Q u
G â (1.15)
8 8 Ê
å ë
é ê
where äÑD inner most radius of the disc, and D the boundaries of the mul-
u jðòñ ôódõ
ticoloured disc considered
> and 7< FS4^ . The integral of equation 1.15 is definite,
2 Q 8r
hence ÝG between the frequencies corresponding to and radii. At
jðòñ
8 8
frequencies less than that corresponding to the temperature of the disk at öó÷õ , the spec-
2 : ôódõ
trum has a shape of the Rayleigh-Jeans spectrum i.e. öG whereas at frequencies
8 8
greater than that corresponding to the temperature at , the spectrum 3>@BA has a shape of
2 öðøñ =
the high frequency tail of the blackbody spectrum i.e. jG l .
8
The optically thick SS discs (or & discs) became the hallmark of the accretion disc
theory because of the rather robust spectral predictions. The predicted emission from
these disc should consist of a sum of blackbody spectra, peaking at a maximum tempera-
ture of k ëù keV for near Eddington accretion rates or at a . keV for an accretion rate
of ú
+ Eddington, onto a typical blackhole ( ú
, ). These discs, however, were
soon discovered to be unstable, at low mass accretion rate, due to instability induced
by the transition of neutral hydrogen to ionized gas (gas pressure dominated regime)
causing opacity variation in the solution (see Figure 1.6). At the temperature when the
hydrogen ionizes, the (Wien’s) photons are absorbed and the energy is not allowed to
escape, heating up the disc which causes a runaway heating until most of the hydrogen
is ionized. Thus, introduction of instability due to the temperature crossing hydrogen
ionization at any point in the disc causes the whole disc to become unstable. This led to
the classic disc instability paradigm, which was successful in explaining the dwarf-novae
type of outbursts in white-dwarf binaries. In the case of neutron stars and blackhole bi-
naries, the X-ray irradiation contributes to the hydrogen ionization and hence controls
the disc evolution. At higher mass accretion rates (the radiation pressure dominated
regime, i.e. total pressure G4 x ) instability is introduced as a small increase in temper-
ature causes large increase in heating rate, and hence to further increase in temperature.
This runaway heating stabilizes only when the time scale for the radiation to diffuse out
is longer than the accretion time scale, i.e. the photons are advected into the blackhole
(slim disc – optically thick accretion: Abramowicz et al. 1988). The spectra of these
slim discs differ from the SS discs as the energy generated in the innermost radii of the
disc is preferentially advected (Watarai and Mineshige 2001), but the disc structure at
such high mass accretion rate is uncertain.
Alternative solutions to the SS disc models predict a truncated disc with an optically
thin, X-ray hot accretion flow in the inner regions. The first such a solution was given
by Shapiro-Lightman-Eardley model (SLE: Shapiro et al. 1976) of accretion flow (see
Figure 1.6), where the electrons cool by radiation while the protons cool by Coulomb
collisions, and hence the flow is intrinsically a two temperature plasma. 4 In the SLE
flow the electrons radiate most of the gravitational energy through Comptonization of
photons from the outer disc. The other popular solution of the optically thin accretion
flow also assumes that the gravitational energy is given mainly to the protons, while the
electrons are heated via Coulomb interaction, forming, again, essentially a two temper-
ature plasma; but in this case the protons carry most of the energy inside the blackhole
horizon, resulting in an advection dominated accretion flow (ADAF: Narayan and Yi
1995). The ADAF solution is supposed to be more stable than the SLE solution. A
chief drawback in these models is that the proton temperature approaches the virial tem-
perature, hence the pressure support becomes important and the flow no longer remains
geometrically thin. Another model with the truncated disc geometry is the two com-
ponent accretion flow model (TCAF) (Chakrabarti 1996a), which assumes a thermal
ûoüþýaû , with maximal temperature possibly as high as û X

¨

4
K
b bÿ
V
ck
y thi ion
all ct
p tic adve
o
Radiation pressure
log m
.
instability
AF
AD
E
SL
Hydrogen ionization
instability
log τ (vertical optical depth through the disk)

Fig. 1.6: The multiple solutions to the accretion flow equations, plotted as a function of (mass

accretion rate) vs. (optical depth) of the fluid flow in the disc. The right hand solution
is the SS disc modified by advective cooling in the highest mass accretion rates and by
atomic opacities at lowest mass accretion rates. The left hand solution is the optically
thin flows, the ADAF and SLE solutions. (The figure obtained from Done (2002)
multicoloured outer disc and a hot inner sub-Keplerian flow separated by a centrifugally
bounded layer (CENBOL), with the presence of bulk-motion Comptonization in the in-
ner Comptonizing cloud (Chakrabarti and Titarchuk 1995, 1996). This TCAF model
will be discussed in more detail in chapter 6.
1.4.1 Unification of hydrodynamic solutions of accretion flow

Of the four physically distinct sequences of accretion discs thermal equilibria that exist
t
locally (Figure 1.7), two correspond to &w)& Ê , and other two correspond to &Å-
t t
& Ê , where & Ê D implies critical ‘viscosity’, which depends strongly on the radius
and weakly on the mass of the central accreting object. Chen et al. (1996) state a value
t t
of & Ê
while Bjornsson et al. (1996) find & Ê -£
after treating the microphysics
and inner boundary conditions more accurately. The two low viscosity flow sequences
of models are further differentiated by the optical opacity (Ô ) of the flow, optically thin

or thick, while the two high viscosity sequences by whether advection is negligible or

dominant (quantified by the ratio of the advected energy,
ó
Ò , to that of radiated energy

Ê ). These four sequences, which are effectively a highlight of the local classification
ó

Fig. 1.7: The four types of accretion flows labelled with their cooling mechanisms. The so-
lutions are plotted as a function of accretion rate ( ) against the vertically averaged

surface mass density ( ). (The figure is taken from Bjornsson et al. (1996))
at a fixed radius of the accretion disc, are as follows:-

t
Type (1): &) & Ê ÔI-*
[ [ ó
Ò Ê

ó
varies: This “S” shaped sequence has three
branches: 1) lower D gas pressure dominated, radiatively cooled, optically thick
classical SS disc with modified (multi-coloured disc) blackbody radiation, where
the bend (i.e. instability) is due to strong opacity dependence on temperature
(the characteristic of dwarf novae), for higher accretion rate the opacity is deter-
mined by electronscattering and the branch is both thermally and viscously stable;
2) middle D optically thick, radiation pressure dominated (and cooled), but ther-
mally and viscously unstable classical SS disc (physically unviable); 3) upper D
optically thick, radiation pressure supported, advection cooled, thermally & vis-
cously stable, slim accretion disc branch.
t
Þ
: This “ ” shaped sequence has three
Type (2): &Å-& Ê Ô varies,
[ ó
Ò Ê

ó
branches: 1) right D identical (with higher viscosity) to the lower branch of type
1; 2) upper D thermally and viscously unstable discs with × Ê
¸ ó ¸ ó
Í
and
ÔU
, the solution is obtained phenomenologically for the latter parameter value;

3) left D optically thin and very hot SLE solutions with protons having much
higher energies than electrons, where radiative processes include bremsstrahlung,

synchrotron & Comptonization.
t
Type (3): & )& Ê ÔE)
[ [ ó
Ò

Ê
ó
varies: This - shaped sequence has two
branches: lower D SLE models; upper D gas pressure dominated, very hot
ADAF discs (with very low radiative frequency), where the spectra are dominated
by Comptonization of bremsstrahlung and synchrotron soft photons and pairs are
suppressed by strong advective cooling.
t
Type (4): &9- & Ê Ô varies,
[ ó
Ò Ê
ó
-Í
: This straight line sequence has two
branches: 1) lower D ADAF solution, similar to upper branch of type 3; 2) up-
per D copious electron– positronpair production in hot, effectively optically thin

plasma cloudL where
: two conditions are met i) temperature of plasma is sufficiently
high, FS4 P ;
and ii) optical depth to photon-photon interactions is suffi-

ciently large. The radiation field may include bremsstrahlung, Comptonization of
internally generated bremsstrahlung photons, as well as from a hybrid population
of electron– positron pair population (Coppi 1992).
1.4.2 Hard X-ray emission models

Wide-band X-ray spectroscopy reveals the presence of hard X-rays, in all spectral states,
as a generic feature of accretion onto a blackhole and this emission cannot be explained
by the standard SS disc models. The only physical phenomenon that can account for
emission in this energy band is magnetic reconnection, while the physical mechanism
most easily and commonly used to explain the wide band spectra is Comptonization
(successive Compton scatterings of soft photons). A very brief outline of the attempts to
explain the hard X-ray emission models is as follows:-
$ MHD dynamo – disc viscosity. For an SS disc, to produce the hard X-rays a large
amount of gravitational energy has to be dissipated in optically thin medium, and
an obvious candidate (perhaps the only one) is the magnetic flares above the disc,
generated by the Balbus-Hawley MHD dynamo responsible for the disc viscosity
(Balbus and Hawley 1991). Buoyancy causes the magnetic field loops to rise to the
surface of the disc and hence they may reconnect in regions of fairly low particle
density.
$ Comptonization from hybrid thermal/non-thermal pair plasma. This model in-
volves the seed soft (X-ray) photons getting Comptonized by a cloud of hot elec-
tron or electron– positron pair plasma (Coppi 1992). The thermal population of
L A
plasma assumes a given temperature, 4 and a Thomson optical depth, Ô . The
deviation of the electronpopulation in the corona from the Maxwellian may be ex-
plained by a hybrid thermal /non-thermal population of (pair) plasmas (Poutanen
and Coppi 1998). Selected electron from a thermal distribution are accelerated to
relativistic energies (Zdziarski 2000), probably in the reconnections (for a review
see Poutanen 1998).
$ Compton reflection. This implies photo-electric absorption and Compton down-
scattering of hard radiation from the disc which may be cold (White et al. 1988,
Lightman and White 1988, George and Fabian 1991, Magdziarz and Zdziarski
1995) or ionized (Done et al. 1992) or may switch between the two states (Zycki et
al. 1998). These reprocessed photons may again act as seed for the Comptonizing
corona. The energy balance of the cold and hot phases determine the temperature
and the shape of the emerging spectrum (Haardt and Maraschi 1991, 1993, Stern
et al. 1995, Poutanen and Svensson 1996).
$ Bulk motion Comptonization. Accretion flow (passing through a shock) forms a
quasi-spherical inflow, the free-falling electrons near the horizon has large bulk
motion Comptonizes the seed photon from the optically thick, cold, geometri-
cally thin, accretion disk, giving rise to the power law spectrum (Chakrabarti and
Titarchuk 1995, 1996). This model will be discussed in more detail in conjunction
with the TCAF model in chapter 6.
1.4.3 Geometrical structure of the accretion system

There are various proposed geometrical structures defining the accretion flow into the
compact object (blackhole event horizon) and the Comptonizing corona which is the
origin of the high energy tail in the spectra:-
$ A hot (magnetic) slab-corona sandwiching the cold accretion disc. In extreme
cases of complete dissipation of energy the spectra resemble the Seyfert galaxies
(Haardt and Maraschi 1991, 1993). The predicted steep spectra is not in accor-
dance with the observations (Poutanen 1998, Done 2001).
$ Patchy static magnetic corona above the disc. For the cold disc the predicted spec-
trum is harder, but the reflection albedo expected is generally too small for neutral
disc (Gierlinski et al. 1997). Also, the predicted anisotropic break is not really
observed. For the ionized disc, the formalism of accretion disc is complicated
with repercussions on the Compton reflected spectra (Ross et al. 1999, Done and
Nayakshin 2001b,a).
1.5. Outflows from microquasars 21
$ Cloudlets. The cold disc within the hot corona is disrupted to form cold dense
optically thick clouds, that can reprocess hard X/0 -ray radiation to produce seed
photons (feedback) for Comptonization (Kuncic et al. 1997).
$ Truncated disc with inner X-ray hot flow / Sombrero (Poutanen 1998). This is
favoured by advective flow model (with the optically thin Bondi like flow replac-
ing the inner part of the accretion disc), as well as the bulk motion Comptonization
or the TCAF paradigm, which can explain the state transitions (Figure 1.8).
Fig. 1.8: The geometry for the sombrero

shaped truncated disc with inner
hot X-ray flow. (The figure is
taken from Done 2002)
1.5 Outflows from microquasars

Relativistic outflows, or ‘jets’, represent one of the most obvious, important but poorly
explained phenomena associated with accreting relativistic sources in general and mi-
croquasars in particular. Historically, the key observational aspects were studied in the
radio regime of the electromagnetic spectrum. Their earliest associations were observed
to be with the quasars, the radio-loud sub class of the AGNs, and it was subsequently
recognized that jets are a common consequence of the accretion process. However, the
study of the outflow (in conjunction with the inflow – accretion) in the Galactic micro-
y
quasars didn’t commence until the discovery of superluminal motion in GRS 1915 105,
in the last decade (Mirabel and Rodriguez 1994). Till date a wide variety of radio emis-
sion behavioural characteristics, patterns and evolutions have been observed, and only
recently any systematic understanding of the accretion – ejection connection has gradu-
ally started to develop. In Figure 1.9 the radio outflow in three microquasars systems is
depicted.
Flare events. Short term injection of energy and matter in the form of an expand-
ing plasma cloud collimated into a jet results in a radio flaring event, which are char-
acterized by an optically thin spectrum and are normally associated with the X-ray
transients (in their outburst/bright phase) and persistent sources like Cygnus X-3 and
Fig. 1.9: Radio images of relativistic jets from Galactic microquasars, a) Superluminal motion

in GRS 1915 105 observed by MERLIN (Fender et al. 1999b), b) arcsec- scale jet
from Cygnus X-3 observed by VLA (Martı́ et al. 2001), c) two VLBI images of XTE J
1550-564 shortly after and four years later following a major flare (Hannikainen et al.
2001). This figure is taken from Fender (2003)
1.5. Outflows from microquasars 23
y
GRS 1915 105 which exhibit such flaring episodes in their X-ray high (as well as soft)
states. The synchrotron bubble model is characterized by the rise phase corresponding to
a decreasing optical depth at frequencies which were initially synchrotron self-absorbed
and this is manifested in an inverted radio spectrum during this rise phase, with possi-
ble Doppler effects on the profile. A possible alternative scenario is that the rise phase
represents a finite period of particle injection/acceleration resulting in an outflow; the
characteristics of this phenomena would be an optically thin spectrum and a duration at
least coupled to events in the observer’s frame, i.e. emission from the accretion disc.
Currently, observations are not in a position to rule out either of these possibilities (see
Fender 2003, and references therein). The time delays in the propagation of a shock may
(misleadingly) mimic the ’synchrotron bubble’ effect.
Steady jets. In addition to the flaring events, persistent sources show a steady radio
emission during the low-hard states, even for sources which do not exhibit the flaring
events. Figure 1.10 shows two such observations using VLBI. These steady jets are
characterized by ‘flat’ spectrum ( &9 ) which may extend through and beyond the
radio band, and may possess polarization at a level of 1-3%. The variability in the
radio emission in this state is supposed to be associated with the X-ray emission and this
feature will be studied in great detail in this thesis, for Cygnus X-3 in particular (chapter
3) and other microquasars in general (chapter 5) to understand the disc-jet connection.
The attempts to form a theoretical understanding of the origin of the collimated
outflows in cosmic sources have been aplenty and varied. Earlier, hydrodynamic and
magnetohydrodynamic formulations were made considering the jets as a separate entity
(Fukue 1982, Chakrabarti 1986). Thereafter efforts were made to correlate the disc struc-
ture with the outflow (eg. Konigl 1989, Chakrabarti and Bhaskaran 1992). Chakrabarti
(1999) investigates the outflow from shock compressed fluid around compact objects.
Numerical simulations, perhaps, provide one of the most realistic approximation to the
actual physical situation. The chief thrust in this field has been in understanding matter
deflection from the equatorial plane towards the axis (eg. Eggum et al. 1985, Molteni et
al. 1994), which currently involves the magnetohydrodynamical formalism (see, for eg.
Kato et al. 2004a,b). Blandford and Begelman (1999) proposed a generalized ADAF
model in which only a small fraction of the gas supplied actually falls on to the black
hole, as the gas is supplied conservatively to a black hole at rates well below the Ed-
dington rate it may not be able to radiate effectively and the net energy flux, including
the energy transported by the viscous torque, is likely to be close to zero at all radii,
resulting in a gas that accretes with positive energy so that it may escape. This paradigm
was known as the advection dominated inflow-outflow solutions (ADIOS).
Fig. 1.10: Steady radio jets in the low-hard state of X-ray emission for Cygnus X-1 (left panel:

Stirling et al. 2001) and GRS 1915 105 (right panel: Dhawan et al. 2000), as ob-
served by the VLBI.
1.6 Aim of this thesis
In the cosmic sources which are intrinsically bright in the high energy band of the elec-
tromagnetic spectrum, the primary phenomenon that gives rise to this X/0 -ray emission
is accretion. In the previous sections it has been highlighted that diverse types of possi-
ble mechanisms exist for the accretion to take place in these class of sources. Given the
multitudinous observational features, as well as the various possible theoretical alterna-
tives, a clear picture of the physical and geometrical structure of the accretion process
is yet to emerge. Further, in addition to this phenomenon, the process of Comptoniza-
tion plays an equally important role in generating the emission in this high energy (hard
X-ray and soft 0 -ray) regime.
According to the CGRO–BATSE catalogue of sources monitored by the occultation
method, the Galactic blackhole (X-ray) binary system Cygnus X-1 is the brightest in the
sky in this regime of X/0 -ray emission (Harmon et al. 2004). This source, as mentioned
before, falls in the class of Galactic microquasars systems, which as a class display this
accretion phenomenon most evidently among the Galactic sources. Till recently, in most
of the attempts to understand these physical processes and phenomena in this class of
1.6. Aim of this thesis 25
sources, the inflow used to be treated in an isolated manner from the outflow, which is
ubiquitously present, normally in the form of collimated jets. Understanding the basic
mechanism of the jet formation and particle acceleration requires a detailed study of the
energetics (Fender 2001a) and a long term multi-wavelength monitoring of these systems
spanning the extremes of the electromagnetic band. Thus far, the observed features of
the outflow and its association with the inflow, across different sources, has failed to
produce any universally consistent picture of the nature of the disc-jet association in
these sources. It has to be realized that accretion and ejection need to be considered
in a unified manner to develop the primary paradigm of understanding of the physical
processes in this systems, i.e. the high energy (X/0 -ray) emission needs to be studied
along with the radio emission from these sources (microquasars).
Considering the diverse theoretical and observational features of the accretion-ejection
phenomena, the pressing need of the hour, from an observational approach, is to inves-
tigate and understand this phenomena across the diverse type of sources, in order to
provide a unified, consistent set of observational features with the aim of developing a
phenomenological model to unravel the physical and geometrical structure of these X-
ray binary systems. This approach requires a long term study of the associated radio as
well high energy (X/0 -ray) emission of these sources, and this defines the primary goal
of this thesis.
Chapter 2
X-ray detectors and techniques of

instrumentation; Radio astronomy
Although astronomy is one of the ancient forms of science, the field of high energy
astronomy is newly developed, reaching its maturity only in the last decade. The pri-
mary, and perhaps the only, reason for it is the interaction of X-rays with the material
of the atmosphere, as a result of which the observations need to be conducted above
the atmosphere. Figure 2.1 shows the transparency of the Earth’s atmosphere for pho-
tons as a function of the energy (frequency) of the electromagnetic waves. As depicted
in the figure, electromagnetic waves in the high energy regime, from far ultra-violet to
high energy 0 -rays cannot penetrate beyond an altitude of 30 km from the surface of
the earth. Balloons, which can attain this height, were extensively used in the early his-
torical phase of this branch of astronomy. These observations are limited, in the low
energy, to 20 keV only. Soft X-ray observation (0.5 – 20 keV) was made possible by
the advent of the satellite technology. Satellites have also made long duration observa-
tions possible, otherwise the total duration of observations done using rocket or balloon
flights didn’t normally last for more than a few minutes or hours, respectively. The role
of the balloon flights is now limited to the development and/or testing of new instru-
ments. Currently the prominent operational satellites observatories for observing in the
X-ray range are CHANDRA, XMM-NEWTON, RXTE, INTEGRAL. The CHANDRA and
XMM-NEWTON X-ray observatories employ the method of X-ray focusing using the
grazing angle X-ray optics. Using the X-ray focusing method these latest generation ob-
servatories have reached comparable sensitivity and data quality in the other established
branches of astronomy such as optical astronomy and radio astronomy. In this chapter
the basics of X-ray detectors are discussed, with some detailed information regarding
two observatories, RXTE and CGRO, whose data have been used in this thesis. For the
26
2.1. X-ray detectors 27
Fig. 2.1: Transmission properties of the Earth’s atmosphere at different wavelengths. The solid
line shows the altitude at which half of the radiation from space get attenuated.
sake of completeness we also mention the other notable X-ray observatories of recent
past and the present. Thereafter we provide a brief note on the data analysis techniques
for these satellite observatories. The last section describes radio astronomy with GBI,
whose regular monitoring of the Galactic X-ray binaries has played an extremely im-
portant part in the research of the disc-jet connection in the Galactic microquasars, and
whose data has been used extensively in this thesis.
2.1 X-ray detectors

X-ray detectors are essentially photon counting devices, with the detection of one photon
being registered as an event. These events are tagged by the time of event occurring and
also the energy channel to which it belongs to, the latter being converted into energy by
the response matrix of the concerned detector. The interaction between the high energy
photons, that constitute the high energy band of electromagnetic radiation, and matter
28 Chapter 2. X-ray detectors and techniques of instrumentation; Radio astronomy
is capable of depositing large amount of energy in the matter, which may be collected
and measured electronically. The matter (of the detector) that interacts with the photons
may be in any of possible states, solid, liquid or gas. Basically, there are three physical
mechanisms by which the high energy photons interact with the matter in the detector
and transfer its energy (Longair 1994):-
$ Photo-electric effect. The X-ray photon is absorbed by the atom and an elec-
tron is ejected from the inner shell, with an energy equal to the difference of the
initial photon energy and the binding energy of the electron. This mechanism
dominates for photons with energy µ 100 keV. The cross section of interaction

is proportional to H , where H is the atomic number of the material, and hence
material with high H are normally chosen for such detectors.
$ Compton scattering. The photon is scattered by an outer shell electron by an
Ä Ä
angle , depositing some energy on the scattered electron. determines the energy
transferred to the electron, and may take any value from zero to a large fraction of

the incident photon energy. This mechanism dominates for initial photon energy
400 keV. The cross-section of interaction is given by the relativistic Klein-
Nishima formula, and depends linearly on the atomic number H of the matter,
Ä
apart from a function of the incident photon energy and scattering angle .
$ Pair production. In an interaction with the coulomb field of a nucleus, the pho-

ton gets absorbed and an electron- e pair is generated. The photon need to have
energy more then twice the electron rest-mass energy i.e. - 1.02 MeV, and any ex-
cess photon energy goes into the kinetic energies of the newly generated electron-

e pair. The cross-section is very low upto several MeV of incident photon energy.
An X-ray detector converts the energy of the incident X-ray photons into a measur-
able electrical signal. The resolution and accuracy of the various measurements, viz. the
position, energy and arrival time of the incoming photon, depends on the characteris-
tics of the detector. An ideal X-ray detector should combine the precise measurement
of position, energy and time of the X-ray interaction along with a large collecting area,
good detection efficiency, ability to handle large count rate, low internal background,
stable performance over long time etc. Clearly it is impossible to have an ideal X-ray
detector and the different types of X-ray detectors available have different strengths and
weakness. Depending on the needs of the observations required of a detector, the one
with the most desirable characteristics are chosen, with a compromising trade-off with
the other features that are non-essential for the particular type of observation. A detailed
discussion of various X-ray detectors is given in Knoll (2000). The most common class
of detectors are as follows:-
2.1. X-ray detectors 29
Proportional counters. Proportional counters have been the ‘workhorse’ of X-ray as-
tronomy since its early inception and are still in active use. These consist of a windowed
gas cell, subdivided into a number of low and high electric field regions by some ar-
rangement of electrodes. Incident X-ray photon interacts with the gas by photo-electric
effect ejecting an electron. This energetic electron further ionizes the gas generating
more electron-ion pairs until it dissipates all its energy. The number of electron-ion pair
thus generated is proportional to the energy of the incident X-ray photon. These elec-
trons are accelerated when they enter the regions of high electric field and generate more
electron-ion pairs, which are all attracted towards the electrodes of opposite charges and
generate an electric pulse. The electric field is controlled in such a way that the final
number of electron-ion pairs is still proportional to the energy of the incident photon.
Thus, measuring the height of the electric pulse gives the energy of the X-ray photon.
Proportional counters are sensitive at energies less then 50 keV.
Scintillation detectors. These work by converting the energy of the incident X-ray
photon into visible light. The X-ray photon first interacts by the photo-electric effect
generating an energetic electron. The interaction of this electron with the crystalline
structure of the detector then generates visible photons as a result of a complex sequence
of excitations and de-excitations. The visible photons are then converted into an electric
signal by means of a photo-multiplier tube (PMT). The alkali halides NaI and CsI, ac-
tivated by a small amount of either thallium or sodium impurity, are the scintillators of
choice so far in X-ray astronomy. This is because these can be made into large area crys-
tals, have good X-ray stopping power, are efficient light producers and are transparent
to their own light. Some other materials such as plastics or the higher-Z bismuth ger-
manate (BGO) are also used as X-ray scintillators. Scintillators can be used for detecting
photons with energies up to a few MeV. They also offer very good time resolution. How-
ever again their major weaknesses are poor energy resolution and the lack of imaging
capabilities.
Semi-conductors (solid state detectors). The principle of operation of semi-conductor

detectors is analogous to that of the proportional counters. The intrinsic crystalline struc-
ture takes the role of the detecting volume (of gas in the proportional counters) where
the incident X-ray photons interact by photo-electric effect. However, in the solid state
detectors, electron-hole pairs are created, instead of the electron-ion pair in the propor-
tional counters. Average energy to create one electron-hole pair in a semi-conductor
is smaller by a factor of 10 compared to the average energy to create one electron-ion
pair in the proportional counter gas. Thus for a given energy of incident X-ray photon,
the efficiency of creating electron-hole pairs in a semi-conductor detector is about ten

times that of the primary electron-ion pairs in proportional counter. Therefore the semi-
conductor detectors offer better energy resolution. Typical semi-conductors are based
on Silicon (Si) or Germanium (Ge). However, on the downside, Si or Ge based detector
have to be operated at cryogenic temperatures to overcome the thermal noise, and hence
it is not currently feasible to devise large area detectors based on these materials. In the
past few years new class of semi-conductors such as Cadmium-Zinc-Telluride (CZT) or
Cadmium-Telluride (CdTe) have been discovered which may be operated at near room
temperature. At the same time, advances in micro-electronics have presented the op-
portunity to use large number of such small detectors in an array formation, providing
position sensitivity and hence imaging capability. It is believed that such array of small
CZT/CdTe detectors will become the most widespread detectors in the field of X-ray
astronomy in future.
X-ray CCDs. The charge coupled device (CCD), a special type of semi-conductor
detector, is intrinsically pixillated and hence has the inherent advantage of imaging ca-
pabilities without the need of creating an array. CCDs are mainly Si based devices which
offer very good energy resolution. As with any general semi-conductor, X-ray photon
interacts in the active volume of CCD, generating electron-hole pairs. However, unlike
the general solid state detectors, the electron are not immediately absorbed by the elec-
trodes. Rather, they are stored in the particular pixel at which the interaction has taken
place, and are read out later. This storage capability of CCD has revolutionized the field
of optical astronomy where the photons can be accumulated over large time interval
yielding very high sensitivity. However, in the X-ray regime, the CCD has to be read out
after every X-ray photon interaction. The readout from a CCD is time consuming and
hence even though CCDs offer very good imaging and spectral resolution, their temporal
resolution is limited. The process of manufacturing CCDs with large area is a difficult
engineering feat. Another practical difficulty, from astronomical point of view, of the
present day CCDs is that, because of low H of Si, they are only sensitive to X-rays with
energy ) 10 keV. Hence, CCDs are almost always used as focal plane detector for an
X-ray telescope, which themselves are limited to energies ) 10 keV.
2.2. Modern X-ray observatories 31
Fig. 2.2: Schemtic representation of the The Rossi X-ray Timing Ex-
plorer (RXTE)satellite observatory, with its three detector systems (see
http://heasarc.gsfc.nasa.gov/docs/xte/xtegof.html). (The figure is obtained from
http://heasarc.gsfc.nasa.gov/Images/xte/xte spacecraft.gif)
2.2 Modern X-ray observatories

2.2.1 The Rossi X-ray Timing Explorer (RXTE)
The Rossi X-ray Timing Explorer (RXTE) is a satellite X-ray astronomical observatory,
launched on December 30, 1995 from NASA’s Kennedy Space Center. The mission is
managed and controlled by NASA’s Goddard Space Flight Center (GSFC) in Greenbelt,
Maryland. The most attractive feature of RXTE is the unprecedented time resolution
in combination with wide-band X-ray spectral capability with moderate spectral resolu-
tion, to explore the variability of X-ray sources. Time scales from microseconds (lowest
achieved for X-ray astronomy, so far) to months are covered in an instantaneous spectral
range from 2 to 250 keV (ideally). Originally designed for a required lifetime of two
years with a goal of five, RXTE has passed that goal and is still performing well. The
satellite has a low-earth circular orbit at an altitude of 580 km, corresponding to an or-

bital period of about ù minutes, with an inclination of 23 ¶ . The three instruments on
RXTE are the Proportional Counter Array (PCA), co-pointed with detectors on the High
Energy X-Ray Timing Experiment (HEXTE), and the All-Sky Monitor (ASM). The first
two are for pointed observations (PCA & HEXTE), while the last (ASM) monitors the
X-ray sky, scanning 70% of the sky in each satellite orbit. In this thesis, data from each
of these instruments have been used extensively, and hence features of all of these three
are presented here. Schematics of the instruments in the orbiting satellite is shown in
Figure 2.2, and the salient features of the various design parameters and performance
are tabulated in table 2.2.1. Major scientific results from RXTE are: the discovery of
kHz QPO and the detection of X-ray afterglow from the gamma-ray bursts. RXTE also
carried out extensive observations of black hole candidates and bursting pulsars which
provides stringent test of accretion theories. A brief introduction to each of the instru-
ments follows:-
Proportional Counter Array (PCA). Developed at GSFC, it consists of : a system of

five large proportional counter units
: (PCU), each with an area of 1300 cm yielding a
total collecting area of 6500 cm . The large area along with the active anti-coincidence
features provides very high sensitivity, and hence the extraordinarily high time resolu-
tion. Each unit is filled with Xenon gas and achieve low background through efficient
veto schemes including side and rear chambers and a propane top layer. A mechanical
hexagonal collimator is carried on each proportional counter which provides a field of
view of 1 ¶ (FWHM). Because of large effective area, a few seconds exposure detects
weak sources (flux
mCrab). The PCA is effective over the energy range of 2 60
keV with 18% energy resolution at 6 keV. The detailed description on the PCA design
as well as the in orbit performances are given in Jahoda et al. (1996).
The High Energy X-ray Timing Experiment (HEXTE). Developed by Center for
Astrophysics & Space Sciences (CASS), University of California, San Diego (UCSD),
it consists of two independent clusters (called Cluster A and B), containing four NaI/CsI
phoswich scintillation detectors (each containing a collimator, phoswich detector, pho-
tomultiplier tube, gain control detector and associated electronics), passively collimated
to a 1¶ field of view co-aligned
: with the PCA field of view. Each detector has a net
open area of about 225 cm and covers the energy range of 15 250 keV with an av-
erage energy resolution of 15.4% (FWHM) at 60 keV. Each cluster can “rock” (beam
switch) along mutually orthogonal directions to provide background measurements 1 ¶ .5
or 3 ¶ away from the source every 16 to 128 s. The field of view of each: detector is lim-
ited by a Pb collimator. Automatic gain control is provided by using a x Am radioactive
source mounted in each detector’s field of view. The detailed description on the HEXTE
design as well as the orbit performances are given in Rothschild et al. (1998). Events
detected by HEXTE will be processed on board by its own data system before inser-
tion into the telemetry stream at an average data rate of 5 kbit/s. Data products include
Table 2.1: Design parameters and performances of various RXTE instruments (from RXTE web
site)
PCA Energy range 2 60 keV
Energy resolution 18% at 6 keV
Time resolution
1 È
Spatial resolution collimator with 1 ¶ FOV (FWHM)
Detectors 5 proportional
: counters
Collecting area 6500 cm
Layers 1 Propane veto; 3 Xenon, each split
into two; 1 Xenon veto layer
Sensitivity 0.1 mCrab
Background 2 mCrab
HEXTE Energy range 15 250 keV
Energy resolution 15% at 60 keV
Time sampling 8 microsecond
Field of view 1 ¶ FWHM
Detectors 2 clusters of: 4 NaI/CsI scintillation counters
Collecting area 2 800 cm
Sensitivity 1 Crab = 360 counts s l per HEXTE cluster
Background 50 counts s l per HEXTE cluster
ASM Energy Band 2 10 keV
Energy resolution 20% in 2 10 keV (3 energy channels)
Scan time 90 minutes, 80% of the sky per orbit
Angular resolution 0.2 ¶
Spatial resolution 3’ 15’
Detectors 3 Scanning
: Shadow Cameras
Collecting area 180 cm: (without masks)
Net area 90 cm (3 detectors)
Sensitivity 20 mCrab in 90 minutes and 10 mCrab a day
event mode, binned spectra and light curves, and a burst-triggered event buffer. The de-
tailed description on the HEXTE design as well as the in orbit performances are given
in Rothschild et al. (1998).
All Sky Monitor (ASM). Developed by Center for Space Research (CSR) at Mas-
sachusetts Institute of Technology (MIT), it is perhaps the most ingenious of all the three
instruments aboard RXTE. It consists of three Scanning
: Shadow
: Cameras (SSC) on one
rotating boom with a total effective area of 90 cm (180 cm without masks). Each SSC
is a one-dimensional “Dicke camera” consisting of one dimensional mask and a one di-
mensional position sensitive proportional counter. The field of view of a single SSC is
6¶1 90¶ FWHM and the angular resolution in the narrow direction is 0. ¶ 2. A motorized
drive rotates the SSCs from field to field in 6 ¶ steps. At each resting position, a 100 s
exposure of the X-ray sky is made and hence a complete rotation of the SSCs takes place
in 90 minutes. During each rotation, about 80% of the sky is surveyed to a depth of
20 mCrab. The limiting sensitivity of the SSCs becomes 10 mCrab in one day. The
detailed description on the ASM design as well as the in orbit performances are given in
Levine et al. (1996).
2.2.2 Compton Gamma Ray Observatory (CGRO)

Historically the second of NASA’s Great Observatories was christened Compton Gamma
Ray Observatory (CGRO)(the first one being the Hubble Space Telescope). CGRO, at
17 tons, was the heaviest astrophysical payload ever flown at the time of its launch on
April 5, 1991 aboard the space shuttle Atlantis. After successful operation for 9 years
it was safely de-orbited and re-entered the Earth’s atmosphere on June 4, 2000. Comp-
ton had four instruments that covered an unprecedented six decades of the electromag-
netic spectrum, from 30 keV to 30 GeV. Of these, three were instruments for pointed
observations, the Oriented Scintillation Spectrometer Experiment (OSSE), the Imaging
Compton Telescope (COMPTEL) and the Energetic Gamma Ray Experiment Telescope
(EGRET). These three instruments cover an unprecedented broad range of energies in
the 0 -rays, from 50 keV to 30 GeV. The fourth instrument was an all sky monitor Burst
And Transient Source Experiment (BATSE) which continuously observes the entire sky
in the hard X-ray and soft gamma-ray range of 20 keV to 600 keV. Figure 2.3 shows
the schematic diagram of the CGRO satellite along with the location of individual in-
struments. The major scientific results from CGRO are, the discovery of an isotropic
distribution of the gamma-ray bursts, discovery of “bursting pulsar” as well as the dis-
covery that the blazar active galactic nuclei as primary source of the highest energy
Fig. 2.3: Schemtic representation of the Compton Gamma Ray Observatory (CGRO)satellite
observatory, with all of its detector systems. (The figure is obtained from
http://cossc.gsfc.nasa.gov/images/epo/gallery/cgro/cgro line labels.jpg)
cosmic
: gamma-rays. CGRO also obtained first map of the Milky Way at the energy of
Al 0 -ray line.
Of the four instruments on board the CGRO, data from two instruments have been
utilized for the work in this thesis, viz. BATSE and OSSE. A brief introduction to these
instruments follows:-
Burst And Transient Source Experiment (BATSE). The Burst And Transient Source
Experiment (BATSE) serves as the all-sky monitor for the Compton Observatory, detect-
ing and locating strong transient sources called 0 -ray bursts as well as outbursts from
other sources over the entire sky. There are eight BATSE detectors, one facing outward
from each corner of the satellite, which are sensitive to gamma-ray energies from 20
keV to 600 keV. Each detector module contains two NaI(Tl) scintillation detectors:
a Large Area Detector (LAD) optimized for sensitivity and directional response, and a
Spectroscopy Detector (SD) optimized for energy coverage and energy resolution. The
eight planes of the LADS are parallel to the eight faces of a regular octahedron, with
the orthogonal primary axes of the octahedron aligned with the coordinate axes of the
CGRO spacecraft. The LAD detector is a disk of NaI scintillation crystal 20 inches in
diameter and one-half inch thick, mounted on a three-quarters inch layer of quartz. A
light collector housing on each detector brings the scintillation light into three 5-inch
diameter photomultiplier tubes. The signals from the three tubes are summed at the
detector. A quarter-inch plastic scintillation detector in front of the LAD is used as an
anti-coincidence shield to reduce the background due to charged particles. The spec-
troscopy detector is an uncollimated NaI(Tl) scintillation detector 5 inches in diameter
and 3 inches thick. A single 5 inch photomultiplier tube is directly coupled to the scin-
tillation detector window. The crystal housing has a 3-inch diameter 0.05 inch thick
beryllium window on its front face in order to provide high efficiency down to 10 keV.
Each of the eight BATSE detector modules sends data to the Central Electronics Unit
(CEU). The CEU produces data for telemetry at a uniform rate of one packet every
2.048 seconds. BATSE detects 0 -ray bursts on-board by examining the count rates of
each of the eight LADs for statistically significant increases above background on each
of three time scales: 64 ms, 256 ms, and 1024 ms. The discriminator rates in channels
2 and 3 (approximately 60 to 325 keV) are used. The background rate is determined for
each detector over a commandable time interval currently set at 17.4 seconds. The sta-
tistical significance required for a burst trigger is set separately for each of the three time
scales, with a quantization of 0.0625 sigma. The software and data analysis of BATSE
is given by Hakkila (1990), while the experiment is outlined by Fishman et al. (1989).
Oriented Scintillation Spectrometer Experiment (OSSE) The Oriented Scintilla-

tion Spectrometer Experiment (OSSE) consists of four NaI scintillation detectors, sen-
sitive to energies from 50 keV to 10 MeV. Each of these detectors can be individually
pointed. This allows observations of a 0 -ray source to be alternated with observations
of nearby background regions, enabling the measurement of background contamination
quite accurately. The synchronization of the four detectors is provided by the central
electronics, which provides the data acquisition timing and coordination of the point-
ing directions. The experiment’s data acquisition and control system incorporates varied
modes of operation depending on the type of information desired during a particular ob-
servation. The primary element of each detector system is the NaI(Tl) portion of a 330
mm diameter NaI(Tl)-CsI(Na) phoswich consisting of a 102 mm thick NaI(Tl) crystal
optically coupled to a 76 mm thick CsI(Na) crystal. Each phoswich is viewed from the
CsI face by seven 89-mm diameter photomultiplier tubes (PMTs), providing an energy
resolution of 8% at 0.661 MeV. Active gain stabilization is used to maintain this en-
ergy resolution by individually adjusting the gain of each of the seven PMTs. Using the
different scintillation decay time constants of NaI(Tl) and CsI(Na), the detector event
processing electronics differentiates the events occurring in the NaI crystal from those
occurring in the CsI, allowing the CsI portion of the phoswich to act as anti-coincidence
shielding for the NaI portion. The instrumental details and related necessary information
are given by Messina et al. (1992), and the in orbit performance details are reported by
Johnson et al. (1993).
An in-depth account of technical details of the various detectors aboard the CGRO is
given in http://cossc.gsfc.nasa.gov/nra/appendix g.html.
2.2.3 Other notable X-ray missions

A few other notable X-ray missions of the recent past and the present are ROSAT, ASCA,
BeppoSAX, CHANDRA, XMM-NEWTON, INTEGRAL. All these are satellite based ob-
servatories dedicated for high energy astronomy. The Roentgen Satellite (ROSAT) was
a collaborative mission from Germany, US and UK, sensitive in the energy range of 0.1
keV – 2 keV.was a collaborative mission from Germany, US and UK, sensitive in the en-
ergy range of 0.1 keV – 2 keV. Operating between June 1991 – February 1999, it featured
an X-ray imaging telescope with the best ( ( arcsec) spatial resolution of its genera-
tion. The Advanced Satellite for Cosmology and Astrophysics (ASCA) was a Japanese
mission (following GINGAand preceding the ASTRO-E II to be launched within a cou-
ple of years) sensitive in 0.1 keV – 10 keV energy range. Operating between February
1993 – March 2001, it consisted a high throughput X-ray optics using CCD detectors
for X-ray astronomy for the first time. The CHANDRAand XMM-NEWTONmissions
are an advanced versions of ROSAT and ASCA respectively. CHANDRA, the third of
NASA’s Great Observatories (originally known as AXAF, it was re-christened after its
launch, after the great Indian astrophysicist, Prof. S. Chandrashekhar), is sensitive in
the energy range of 0.1 keV – 10 keV. Launched on 23 July 1999 and still currently
active, it features an unprecedented imaging capability with spatial resolution of ) 1
arcsec. XMM-NEWTONis a mission from the European space Agency featuring large
collecting area. Launched on 10 December 1999 and is still currently active, it is sensi-
tive in 0.1 keV – 10 keV energy range. BeppoSAXwas an Italian-Dutch mission which
was operational between April 1996 – April 2000. It was the first X-ray mission with a
scientific payload covering more than three decades of energy - from 0.1 to 300 keV -
with a relatively large effective area, medium energy resolution and imaging capabilities
in the range of 0.1-10 keV (CGRO covered five decades of energy in the 0 -ray band).
The International Gamma-Ray Astrophysics Laboratory (INTEGRAL) is the European
Space Agency mission launched on 17 October 2002. It is the successor of CGRO.
These missions are not described here because data from these have not been used in the
present work, but a detailed introduction to this observatories may be found at the web
site http://www.heasarc.gsfc.nasa.gov.
2.3 X-ray astronomical data analyses and techniques

The analysis of X-ray emission from the satellite observatories essentially involve two
basic steps:-
1. Extraction, from the ’low level’ data, of the (meaningful) standard products, viz.
the image (spatial distribution of the electromagnetic flux in a given energy band),
lightcurve (time-series of flux variation in a given energy band) and spectrum
(distribution of the electromagnetic flux as a function of energy).
2. Analysis of these standard products to understand the physical mechanism and

phenomena of the processes giving rise to these electromagnetic emission.
The observations are made as per the proposals of the observers, selected on the
basis of scientific merits by a review committee. The data is immediately made avail-
able to the observer (and his team of collaborators) for scientific analysis, and after a
stipulated time (generally a year or so, but may differ for different observatories), the
data is archived and made available to the general public, anywhere on earth, available
through the internet. The specific software for the extraction of standard products from
the different observatories are also made publicly available through the internet.
2.3. X-ray astronomical data analyses and techniques 39
The telemetry data received from the satellite is in highly instrument specific for-
mat, which is converted into FITS (Flexible Image Transport System) format by the
mission operations centre who also performs some preliminary analysis to verify the
data integrity and goodness, before making it available to the guest observer. The FITS
format, originally developed for the exchange of scientific images, is a standard for any
type of data exchange. It is basically a standard format for describing the format of the
data. Even if the data is available as a FITS format, the underlying structure of the data
is still highly instrument/mission specific. Since X-ray detectors are essentially photon
counting devices, the basic X-ray data usually comprise lists of detected photons, known
as events, stored along with some other attributes. These attributes normally consist of
time-tag of the event (photon arrival), the position (in detector and sky coordinates) and
an energy. Thus each event can be thought of as occupying a position in a 4-dimensional
space, the four dimensions being the time, the u and coordinates and the energy. In case
the X-ray detector is of non-imaging type then, obviously, the two position attributes are
absent. The events may have other instrument specific attributes such as the pattern of
pixels for the CCD detector or grating order for the grating instruments.
The analyst’s first step, that of data reduction, involves the following couple of sub-
steps: 1) scientific filtering of the events, i.e. selection of data events depending on
required scientific criteria, viz. location in the sky, particular time interval, background
noise level, etc., 2) extraction of the standard products from the ’raw’ event data provided
by the concerned mission operation center, using the mission specific softwares.
The analyses of these standard products, in a mission independent way, may be done
by the generic softwares viz. FTOOLS and XANADU packages, etc, made available freely
(in a combined package called LHEAsoft) by the High Energy Astrophysics Science
Archive Research Center (HEASARC) 1 , GSFC, NASA. The HEAsoft package also
contains many mission specific data reduction software, many general utilities to ma-
nipulate FITS files as well as high level mission independent data analysis software for
detailed timing, spectral and imaging analysis. The complete reduction and analysis of
the RXTE data, for the work presented in this thesis, has been done using the LHEAsoft
package (described in the following sub-section). For the CGRO data, the standard prod-
ucts for the OSSE and BATSE instruments are obtained from the CGRO archives, and
hence the detailed data reduction are not explained here. The analysis of the standard
products has been done using the FTOOLS and the XANADU packages.
1
http://heasarc.gsfc.nasa.gov
2.3.1 RXTE data analysis

The ASM monitoring data is reduced by the instrument team at the MIT, and the standard
products (lightcurves in three energy bands plus the combined total energy range) made
available via the web site “http://xte.mit.edu”. The RXTE data, supplied to the observer,
of PCA & HEXTE, is already low-level filtered and converted to FITS by the RXTE GOF
(Guest Observer Facility) at GSFC, NASA. For the PCA and ASM, this is achieved by
the Experiment Data System (EDS). The EDS consists of eight Event Analyzers (EA),
of which six are dedicated to the PCA and two to the ASM. Each EA contains an Intel
80286 processor and associated memory. The EAs can be programmed independently
to process incoming events from the instruments in any of the following modes: 1)
transparent mode, using 1, 2, or 3 EAs, 2) event mode, using 1 or 2 EAs, 3) binned
mode (time and/or energy), 4) burst catcher mode, 5) Fourier transform mode, 6) pulsar
fold mode, 7) autocorrelation mode, 8) arrival time differences histogram mode. Each
RXTE pointed observation has a unique observation ID (also known as ObsId) of the
format “NNNNN-TT-VV-SSX” where:
$ NNNNN is the five proposal number assigned by the GOF
$ TT is the two-digit target number assigned by the GOF. In the case of only one
target in the proposal, the target number may be zero.
$ VV is the two-digit viewing number, assigned by GOF, which tracks the number
of scheduled looks at the target.
$ SS is the two-digit sequence number used for identifying different pointings that
make up the same viewing.
$ X is a special character, which if present, indicates different types of observation
e.g. slew, scan, grid observations or observations with data gaps or segments of a
long observation etc.
The two main instruments PCA and HEXTE are non-imaging instruments. Hence
data from these instruments consist of time-tagged events along with energy. However,
due to very large effective area, particularly for the PCA, it is not possible to transmit
each event separately. Therefore the detected events are analyzed and binned on-board
by the EDS. The PCA can provide the same data in five different modes, two of which
are standard, known as “standard-1” and “standard-2” modes. The former mode data
contains the light curve of the source with 125 ms time resolution but without any spec-
tral information, whereas the latter data contains the 129 channel spectra accumulated
every 16 seconds. Other three modes are proposal specific. These modes basically obtain
data with the different levels of trade-off between the timing and spectral resolutions i.e.
modes with very high time resolution (micro-seconds) have low or no spectral resolution
whereas the modes with moderate time resolution (milli-seconds) have higher spectral
resolution (16 to 128 channels). Similarly HEXTE also provides multiple modes of the
same data. It has one standard mode called the “archive” mode accumulating 64 channel
spectra every 16 seconds and one user specified mode. The detailed descriptions of the
RXTE data as well as all available data modes is given in the RXTE handbook “The ABC
of XTE”2 . The user level software for the reduction of the PCA and HEXTE, as well as
for background estimation, response matrix generation etc. is developed and maintained
by the RXTE-GOF. This software is available as a part the LHEAsoft software from the
HEASARC.
For spectral analysis, typically the “standard-2” mode data from PCA and “archive”
mode data from HEXTE may be used. A standard tool saextrct is available to extract
the spectra from the PCA or HEXTE data. Similarly standard tools are available for gen-
erating the observation specific response matrix as well as background data e.g. pcarsp
and pcabackest for PCA and hxtrsp and hxtback for HEXTE respectively. Nor-
mally only PCA data is used for timing analysis, as the large area provides sufficient
photon count, even at high time resolution ( sub-millisecond), to provide meaningful
statistics. In the present work, a “Single-Bit” mode data is used for timing analysis which

provides the highest time resolution, typically 16 – 125 s but no spectral information.
Such data are normally available in two different energy ranges, 2 – 6 keV and 6 – 15
keV. Again the same tool saextrct is used to extract high time resolution light curve
from the “Single-Bit” mode data. The detailed
: description of RXTE data reduction pro-
cess is given in the “RXTE Cook Book” maintained by the RXTE-GOF at HEARSRC,
NASA.
2.3.2 Timing analysis

The purpose of this analysis is to study the variability of the intensity of emission in
the concerned energy band, with the aim of ascertaining the physical processes giving
rise to the emission. A light curve with appropriate time resolution is the starting point
of the timing analysis. The most common types of variability observed and analyzed
is periodicity and/or stochastic variability. The Fourier analysis is perhaps the most
sturdy and common mode of statistical analysis of the variability (see chapter 4 for a
discussion). The XRONOS package (available with LHEAsoft) is used as the standard
2
Available on-line at http://heasarc.gsfc.nasa.gov/docs/xte/abc/contents.html
timing analysis in this thesis.
2.3.3 Spectral analysis
A spectrum is the distribution of flux as a function of energy. The understanding of

this distribution provides the knowledge of the underlying physical processes giving rise
to the emission. The various emission mechanisms (described in chapter 1) have their
characteristic shape of distribution, and hence any complex spectrum may be explained
by a combination of the concerned processes. The spectral analysis of an X-ray spectra
as observed by the detectors is a complex and involved process, as the interaction of the
X-ray photon with the detector matter by which the X-ray photons are detected causes
distortion in the spectral distribution of the original emission. Hence the knowledge of
the detector response is essential to deconvolve the finally obtained spectra.
Detector response matrix and spectral fitting. The final spectral distribution ob-
tained by the X-ray detector is a convolution of the incident X-ray spectrum with the
response matrix of the spectrometer. In any X-ray detector there is a finite probability
that an incident photon with certain energy will be detected as having some other energy
(due to the statistical nature of the interaction), which is determined by the type of de-
tector and physics of X-ray interaction. Also, interactions like photoelectric absorption
cause additional features like escape peak which distort the spectrum. Further, certain
physical processes involved in the detection process (viz. Comptonization) will cause a
deviation in the statistical distribution of the photons, as a function of energy as only a
fraction of the photon energy is deposited, thereby making the original spectrum an in-
herently unknown quantity which is irretrievable. For each energy of an incident photon,
the detected spectrum has a spread across almost all channels. This final spectrum, may
be obtained by using the response matrix of the detector, which is a two dimensional
matrix which gives the probability that a photon of given energy will be detected in a
given energy channel. Evidently, the proper understanding of the response matrix of
the detector is essential for inferring the incident X-ray spectrum. The X-ray spectrum
is represented by a distribution of photon counts over the pulse height channels, where
the total counts in any given channel is the sum of contribution to that particular chan-
nel from the original photon distribution over all energies (in practice all energies over
·
which the detector is sensitive). By general convention the spectral analysis is carried
2
out in units of count-rate rather than total counts. Thus, the observed count rate in
2
a given channel from the incident spectrum ih] (photon flux density at energy h , in
:
photons cm l s l keV l )3 is given by ·
Â
2
U 7Iâ
2
]
[
h ih_ Æ]ih_ h (2.1)
2
where h_ is the response matrix of the detector which gives the probability of
[ 2
detecting a photon with energy h in the channel · and Æ_ih] is the effective area of the
detector at energy h .
In principle, the incident source spectrum ih_ can be obtained· from the observed
2

spectrum by inverting the above equation 2.1, but in practice the response matrix
2
h_ is generally non-invertible. · Therefore the source spectrum ih_ is, in general,
[
not possibly 2
· obtainable directly from the observed spectrum U . The general practice
is to choose a model spectrum ih] such that it can be described in terms of a few pa-
rameters, ih f f : , convolve the model spectrum with the detector response matrix
[ [
to get the predicted count spectrum
2
and then compare it with the observed count
2
n
spectrum U . A “fit statistic” is computed from the comparison, which enables one
to judge whether the model spectrum describes the observed spectrum properly or not.
The model spectrum which agrees best with the observed spectrum is then assumed to
be the true incident spectrum. This process is known as fitting a model spectrum to the
observed spectrum. : The most common fit statistic in use for determining the “best-fit”
model is the fit which is defined as
2 : ~ 2 :
: 7 U 2
n

Ì Ì

(2.2)
~ 2 2 ~ 2
where is the error for channel I. If C(I) are counts in channel then is
usually estimated by U .
2
The best-fit model spectrum : is generally obtained by determining the “goodness-of-

fit” of the model using the -statistic which provides a well established goodness-of-fit
criterion for a given number of degrees of freedom or dof, : (which is the number of
channels minus : the
: number of model parameters). · When
8
exceeds a critical value,
“reduced ” ( 8
·
) \
(Bevington 1969), the concerned model is to be rejected on
statistical basis. Even if the best-fit model, ih f does pass the goodness-of-fit test, the
[

possibility of any other model ih f fitting the data equally well or better cannot be
[
ruled out. The choice of correct model is a matter of scientific judgment. Thereafter, a
”confidence interval” for the best fit model parameters (Lampton et al. 1976) is computed
to define the acceptable range (error) of the best fit parameter values. In this thesis, the
confidence interval, or the errors for a single parameter is calculated at 90% confidence
(Arnaud and Dorman 2002).
3
X X X
The X-ray spectra can also be shown in units of keV cm e s e keV e or keV cm e s e keV e
V V V V
The XSPEC package. The most commonly used software package : for the general X-

ray spectral fitting using the “best fit” statistic (normally the fit) is the XSPEC (Arnaud
1996), which is a sub-package of the XANADU available as a part of the LHEAsoft
package. XSPEC is a command driven, interactive, completely detector-independent,
X-ray spectral fitting program. To work with data from any detector, all that the XSPEC
needs is the observed spectrum data file (generally identified by a “.pha” extension), the
detector response matrix (generally identified by a “.rsp” extension; in some cases two
separate files, the ancillary response file “.arf” and the redistribution matrix file “.rmf”
are required to specify complete response matrix) and the spectral background noise data
file, all in the standard FITS format. This package contains many built in components for
the source model spectrum such as blackbody spectrum, power-law spectrum, Gaussian
or Lorentzian line, etc, a combination of which may describe the source spectrum. But,
it is also possible to add a custom defined model component into the XSPEC package.
Apart from fitting the model spectrum to the observed spectrum, XSPEC also features
different methods to find out the confidence intervals for the parameters of the best-fit
model.
Calibration issues. The response matrix of a given instrument, in principle, depends

on the physics of the detector and hence can be estimated by purely theoretical con-
siderations. However it is not always feasible to consider each and every detail of the
detector in practice, and an experimental method for determining the correct detector
response matrix is needed. This process of experimentally determining the detector
response, i.e. establishing a relation between the incident and observed X-ray spec-
trum is known as calibration. Generally the calibration process involves shining the
detector with available monochromatic X-ray source and subsequent recording of the
corresponding output spectrum. The final response matrix has to be determined by rea-
sonable interpolation. A point to be noted is that even though the individual detectors
are very well calibrated, their absolute calibration may differ slightly. Hence, combining
data from different detectors is a very contentious issue, as the observation of the same
source, such as Crab, may give slightly different parameter values for different observa-
tories/instruments (Vadawale et al. 2001b). This issue needs very careful consideration
when attempt is made to fit the X-ray spectral data simultaneously from different detec-
tors.
Source model component spectrum The X-ray spectral analysis software package,
XSPEC, contains a wide range of model components. Many different combinations of
these model components are possible to define the assumed source X-ray spectrum. It is
also possible to add new model components to XSPEC. Some of the model components
which are used in this thesis are described below:-
powerlaw. A commonly used spectral component, it is defined as
ih]j7
J "! $#
ih
ì ] l (2.3)

J
%
where D the power-law index and D its normalization are the two parame-
ters of the model. Power-law X-ray spectrum can be generated by many different
physical processes such as synchrotron radiation, shock acceleration etc. But, im-
portantly, if the signal-to-noise ratio of the data is not very good then the shape of
the X-ray spectrum due to most of the other mechanisms can be approximated by
a power-law, and hence it is one of the most commonly used spectral component
in X-ray astronomy.
diskbb. This represents a multi-coloured disc black body spectra from a basic accretion
disc emission (equation 1.15). It has two
J parameters, the temperature at inner disk
radius 4 in keV and normalization which is given by -
ðòñ
:
J N F P
7 ðòñ & ' )( ( *

ï T
¼
Ä
(2.4)
&
where D the distance of the source (units of 10 kpc), D the radius of the
Ä ðøñ
inner edge of the disk (units of km) and D the inclination angle of the system.
The multicolor disc-blackbody is mostly used to model the X-ray spectrum from
GBHCs, particularly during their high / soft states.
CompST. This is the simplest Comptonization model based on the work of Sunyaev
and Titarchuk (1980). This model has three parameters, a temperature FS4 in keV
and optical depth Ô of the Comptonizing plasma · and a normalization which is
analytically described as J
¹
7 , : + (2.5)
· J
, -., 0-./2, 1
/ 3
where D total no. of photons from the source, D distance to the source,
D factor
Q
64 5
, here Y is the spectral index, is the injected photon energy
ód
(units of temperature) and P is the incomplete function. %
CompTT. An improved version of Comptonization (after Titarchuk 1994) taking the
relativistic effects into consideration. This is a tabular model which takes the seed
photon energy as an explicit parameter 4 . Also, there are two different Comp-
ðòñ
tonization geometries possible which are specified by an additional ‘geometry’
parameter. This process is typically used to model the X-ray spectrum of low /
hard states of GBHCs.
gaussian. A simple gaussian line profile given by
7 98 e < / X
Æ_ih]j7
(
¹ ~ : = l -;: : (2.6)
~ V
:
where h D ~ line energy (keV), D line width and D total photons/cm /s in
¹
the line. If 0 then it is treated as a delta function ih_ . =
edge. An absorption edge given by
ih] 7£
for h)h
7 = l $> - :': ?A@ / e Ë for h-h (2.7)
where h D threshold energy and ÔUD absorption depth at the threshold.
In the work leading to this thesis, the RXTE–ASM monitoring data, CGRO–BATSE
monitoring data and GBI monitoring data were obtained as the standard products from
the archives of these observatories available publicly over the world wide web. The
standard timing analysis of these data have been made using the XRONOS software
available as a part of the LHEAsoft package, while any non-standard analysis and
file-manipulation has been done by programs developed by us. For the RXTE–PCA &
HEXTE data analysis, the ‘low-level’ data have been downloaded from the RXTE archives,
the lightcurves and spectra were reduced using the FTOOLS package (whichever current
version that was available during the time), and the standard timing as well as spectral
analysis were done using the XRONOS and XANADU package, available with LHEA-
soft. In this case too, any non-standard analysis and file-manipulation (viz. correcting
the X-ray emission from Cygnus X-3 for binary modulation) has been done using pro-
grams developed by us.
2.4 Radio astronomy

2.4.1 Green Bank Interferometer (GBI) observatory
The present work has used radio observations from the radio telescope Green Bank In-
terferometer. The observation and data reduction was carried out by the GBI team, and
2.4. Radio astronomy 47
Fig. 2.4: The three telescopes of the Green Bank Interferometer.
we only used publicly available radio monitoring data (lightcurves) of the Galactic mi-
croquasars and hence the details of the radio data reduction are not described. The Green
Bank Interferometer (GBI) is located in Green Bank, West Virginia, US and is operated
by the National Radio Astronomy Observatory (NRAO), US. It consists of three radio
telescopes of 26m diameter with a baseline of 2400 m. It operates simultaneously at
two frequencies 2.25 GHz and 8.3 GHz with 35 MHz bandwidth. The sensitivity of the
observatory is 6 mJy (S band, 2.2 GHz) and 10 mJy (X band, 8.3 GHz) B
rms noise
measured in 5 minutes scan. GBI used to monitor almost all interesting radio sources in
the sky few times a day, however, from October 2000 on-wards the monitoring program
has been stopped due to lack of funding. Earlier monitoring data is archived and is freely
available and extensively used in this thesis.
2.4.2 Giant Metrewave Radio Telescope (GMRT)

Although data from this observatory has not been used in this thesis, some preliminary
observations have established the prospect of future study of microquasar systems (viz.
Cygnus X-3) from this observatory, and hence a very brief mention of it is made here.
Located in Khodad, about 80 km from Pune, India, Giant Metrewave Radio Telescope
(GMRT) is the world’s most powerful radio telescope operating in the frequency range
of about 50 to 1500 MHz. Operated by The National Centre for Radio Astronomy,
T I F R, it consists of 30 fully steerable gigantic parabolic dishes of 45 m diameter
each spread over distances of upto 25 km. The number and configuration of the dishes
was optimized to meet the principal astrophysical objectives which require sensitivity at
high angular resolution as well as ability to image radio emission from diffuse extended
regions. Fourteen of the thirty dishes are: located more or less randomly in a compact
central array in a region of about 1 km , and the remaining sixteen dishes are spread
out along the 3 arms of an approximately ‘Y’-shaped configuration over a much larger
region, with the longest interferometric baseline of about 25 km. The array operates in
six frequency bands centred around 50, 153, 233, 325, 610 and 1420 MHz. All these
feeds provide dual polarization outputs. In some configurations (eg. 1420 MHz), dual-
frequency observations are also possible.
Chapter 3
Cygnus X-3: spectral studies
3.1 Why Cygnus X-3?

In chapter 1 it was explained and emphasized that a long term multi-wavelength moni-
toring was mandatory to provide a comprehensive picture of the observational features
of the various microquasars, which may provide a handle to probe the accretion-ejection
connection mechanism of these systems. Ideally, the source should be persistent in the
multi wave bands, covering both quiescent (non-flaring) and flaring states, displaying
both transient ejection events and the steady jets in the low-hard states. As mentioned in
chapter 2, RXTE–ASM, CGRO–BATSE and GBI provide the only continuous monitor-
ing data in the X-ray and radio bands, publicly available, for the period when these three
observatories were simultaneously operational. Fender and Kuulkers (2001) provide a
list of Galactic X-ray binaries which are observable in the X-ray and radio bands, of
which most are the transient type. Collating these sources with those monitored by the
three observatories, we found three X-ray binaries (black hole candidates) viz. Cygnus
X-1, GRS 1915+105 and Cygnus X-3, which are persistent in radio, soft and hard X-ray
bands and for which (quasi) simultaneous data from the three observatories are avail-
able. Of these three sources, Cygnus X-1 doesn’t show the huge transient outflow (burst)
episodes. Whereas GRS1915+105, since its discovery in the last decade, is perhaps the
most studied and reported source. Cygnus X-3, on the other hand, despite being one of
the commonly observed source, is neglected in the literature in the recent times, espe-
cially so in the high energy band in the post RXTE era. The likely reason for this is,
perhaps, the difficulty in explaining the X-ray observational features (viz. spectral, tem-
poral). But given the diverse behavioural pattern in the X-ray as well other wavelengths
displayed by this source, it was imperative that a detailed multi-wavelength study, with
emphasis on the X-ray spectral and temporal evolution, be carried out. The diversity of
49
50 Chapter 3. Cygnus X-3: spectral studies
Fig. 3.1: Chandra image of Cygnus X-3 highlighting the halo (and/or dust) engulfing the system,
which may originate from the wind of the companion.
the X-ray behavioural patterns in this source is surpassed only by GRS 1915+105, which
is extensively reported in the literature.
3.2 General properties of Cygnus X-3

Cygnus X-3 is one of the earliest Galactic X-ray sources to be detected (Giacconi et al.
1967), and therefore obviously is one of the brightest X-ray objects in the sky. There-
after, in a programme to detect radio counterparts of the early X-ray sources, a compar-
atively weak radio counterpart for this source was discovered by Westerbork synthesis
telescope (Braes and Miley 1972) and NRAO Green Bank Interferometer (Hjellming et
al. 1972), during what we understand now as the low state with core jet outflow. Soon,
however, huge radio outbursts were observed by Algonquin Radio Observatory, Ontario,
Canda (Gregory et al. 1972) followed by GBI (Hjellming and Ballick 1972) and other
various radio observatories (Anderson et al. 1972, D’Addario and Stull 1972, Aller and
Dent 1972, Dent et al. 1972, Gary et al. 1972, Branson et al. 1972). During the same time
infra-red (Becklin et al. 1972) and mm wavelength (Pomphrey and Epstein 1972) emis-
sions were also discovered from the source. Using hydrogen–line absorption in front of
3.2. General properties of Cygnus X-3 51
Fig. 3.2: Chandra image of Cygnus X-3 highlighting the elongated X-ray excess, which may
be due to dust from the companion, or material from a previous jet emitted from the
compact object.
the source in the line of sight, Lauque et al. (1972) derived the location of the source in
one of the Galactic arms at a distance of 8 – 11 kpc from earth (see Bonnet-Bidaud and
Chardin 1988, for a review), which was later corroborated by the method of determin-
ing the geometric distance by measuring the apparent delay of the intensity variation of
the radiation of the halo scattered by interstellar dust, due to smearing out in the halo
itself (Predehl et al. 2000, see Figure 3.1). All attempts to detect the optical counterpart
were unsuccessful (see Bonnet-Bidaud and Chardin 1988, for a review), therefore we
can safely conclude that any possible counterpart in the optical has magnitude more than
26.
The X-ray emission displayed a modulation of 4.8 hours (Parsignault et al. 1972,
Sandford and Hawkins 1972), which was also observed in the infra-red band (Becklin
et al. 1973), and this was attributed to the binary orbital motion. At the same time
the source was plagued with a variety of aperiodic variabilities, including outbursts, in
all the bands(Becklin et al. 1974). Since the early days of its observations, the system
was recognized to be of the accreting binary type, with one of the components of the
binary being a compact star, accreting matter from the companion, which loses material
via strong winds (Davidsen and Ostriker 1974, Pringle 1974), with the whole system
engulfed in a spherical shell of gas (Milgrom 1976). This idea was further developed by
Fender et al. (1999a), who suggested that the stellar wind is compressed into a flattened
disc with dimensions much larger than the binary system.
Given the compactness of the system as suggested by the binary period, initially it
was assumed to be a low mass X-ray binary system (LMXB), but the infra-red obser-
vations of broad emission lines of He I and He II, along with the absence of hydrogen,
suggest presence of dense wind and indicate the companion to be a Wolf-Rayet star (van
Kerkwijk et al. 1992), the helium core of a massive star embedded in dense wind (van
Kerkwijk et al. 1996). The measurement of the radial velocity in the infra-red band
and its interpretation as due to binary Doppler shift has led to the derivation of a large
mass function for the source (Schmutz et al. 1996). But this simplistic explanation of
the He II line shift is strongly disputed by the relative phasing of the infra-red and the
X-ray binary modulation, as pointed out by van Kerkwijk (1993) who indeed suggests
that the He line emissions originate in the two temperature wind from the Wolf-Rayet
companion, and the line-shift occurs as the wind recedes and approaches the observer,
as a function of the binary motion. Since the velocity measured is that of the wind, the
motion of the Wolf-Rayet companion remains undetermined and hence the mass func-
tion of the system is unknown. Interestingly, quite recently Stark and Saia (2003) have
suggested an upper limit to the mass of the compact object of 3.6 M . They measured
the Doppler shift of the He-like K & like line of iron XXV (Paerels et al. 2000, Kita-
moto et al. 1994) and ascribing it to originate from very near the surface of the compact
object they provide a conservative estimate of the stellar masses and the separation of
the binary system. Till date, all the models of the infra-red emission from the source
consider the (wind from the) Wolf-Rayet companion to be the only source of radiation
in this band. Recently Heindl et al. (2003) have reported an extended emission in the

X-ray band (using the CHANDRA observatory), which has a dimension of pc at
a distance of ( pc from the source (assuming Cygnus X-3 to be at a distance of 10
kpc). Shown in Figure 3.2, the the lines numbered 1, 2, etc. correspond to different roll
angles, all of which give the same extended structure in space, negating the possibility
of any artefactual manifestation of the point spread function.
Persistent in all the bands in which detected so far, Cyg X-3 does provide a stable
system to study the accretion and ejection phenomena, across various possible states.
The evolution the X-ray spectral features, correlated to the radio emission, provides a
challenge for the current generation of astrophysicists to understand and explain the disc
jet connection in this (class of) object(s).
3.3. General spectral features of Cygnus X-3 53
3.3 General spectral features of Cygnus X-3

3.3.1 A historical perspective
Complex features like absorption edges (Bleach et al. 1972) and broad iron line (from
Ariel-5 satellite – Sandford et al. 1975, which was later resolved to consist of three sepa-
rate iron lines by ASCA– Kitamoto et al. 1994) were observed quite early in this source.
UHURUdetected that the source existed in two different states with different levels of
X-ray emission flux (Leach et al. 1975), which were found to possess different spectral
shapes from the rocket flight observations (Serlemitsos et al. 1975). Efforts were made
to study the binary phase resolved X-ray spectra of the source in both its high (Blissett et
al. 1981) and low state (Becker et al. 1978), with conflicting results. The latter reported
an excess of soft X-ray flux at the phase 0 (minimum) of the binary modulation, which
was not seen in the high state. The issue is not settled till date because of the high ab-
sorption of the soft X-ray flux in the line of sight to the source, caused probably due to
the circumstellar matter that may have been caused by the strong wind ejection from the
Wolf-Rayet companion. Accretion disc corona has also been considered a medium of
X-ray reprocessing (White and Holt 1982) along with the X-ray scattering halo (Molnar
and Mauche 1986).
The first attempts of detailed X-ray spectroscopy was attempted by the EXOSAT (Ra-
jeev et al. 1994) and GINGA (Nakamura et al. 1993) observatories, while observations
from ASCAfurther improved the resolution of the spectra in the soft X-ray regime, facil-
itating the identification of the three iron line features (Kitamoto et al. 1994). Rajeev et
al. (1994) explained the X-ray spectral energy distribution (SED), in both (low) hard and
(high) soft state with a composite model consisting of a disc black body component and
a non-relativistic thermal Comptonization component-CompST (Sunyaev and Titarchuk
1980) along with the line and edge features, from the EXOSAT pointed observations,
covering the 1-100 keV band. They also found a low energy thermal bremsstrahlung
component, more significantly in the (high) soft state. These observations also unam-
biguously resolved two absorption edge components, at 7.1 keV (neutral Fe) and 9.2
keV (iron XXV - XXVI). From the GINGAobservations, Nakamura et al. (1993) ex-
plained the X-ray continuum SED, in both the states, by a composite model consisting
chiefly of a disc black body component and a powerlaw component with an exponential
cut-off. They pay stress on the emission from an X-ray halo, the scattering of the low
energy X-ray emission in the circumstellar medium (Molnar and Mauche 1986), other-
C
wise, they claim, the absorption due to effective hydrogen column density N will be
severely underestimated. Nevertheless, they report very high absorption in the low en-
ergy regime, giving unreasonably high values of the effective hydrogen density column
C
N . While Rajeev et al. (1994) speculated the presence of a Comptonizing cloud along
with a thermal accretion disc, Nakamura et al. (1993) postulated the presence of three
kind of hot gases with differing ionization levels. From the ASCAobservations the broad
ED D
iron K & line emission feature was resolved to be consisting of three separate compo-
FD
nents, a He-like line at ùëù
line at ù .

keV, a H-like line at ù ù ,( keV and a neutral
,. keV (Kitamoto et al. 1994). Also, various other emission lines viz.,
K & line of He-like and H-like ions of S, Ar, Ca, along with Fe, were observed in the data
from this observatory, confirming the presence of a photo-ionized plasma in the system,
which might extend asymmetrically (Kawashima and Kitamoto 1996). The presence of
these line emissions was later confirmed by the CHANDRA observatory (Paerels et al.
2000).
3.3.2 X-ray wide band spectra from RXTE

The X-ray SED typically shows two different states, low (and correspondingly hard)
and high (and correspondingly soft), distinguished by the shape of the X-ray spectra
characterized (Choudhury and Rao 2002), chiefly but not totally, by the presence (or
absence) of multicoloured disk blackbody component and the power-law index (albeit
with the individual model components more complicated than the canonical X-ray states
of classical black hole candidates characterized, chiefly, by Cygnus X-1, see, for eg.
Tanaka and Lewin 1995). To get a broad-band (5–150 keV) spectral picture we used
72 sets of the pointed observations of both the narrow field of view instruments aboard
the RXTE: viz. PCA and the HEXTE. A systematic error of 2% was added to the PCA
Standard 2 data (all PCUs added), which included all the 129 channel PHA data, and it
was simultaneously fit with 64 channel data from only the cluster 0 of the HEXTE, to get
a proper fit (Vadawale et al. 2001b). Of these 72 sets of observations, there are only in
effect 48 independent observations on different days (others are extended observations
on the same day), of which 11 are in the low (hard) state and the rest are in the high
(soft) state. To get a proper idea of the spectral state during the pointed observations,
in table 3.1 we give the flux as observed by three monitoring instruments: viz.1) ASM
aboard the RXTE, in the soft X-ray energy region (2-10 keV), 2) BATSE aboard the
CGRO, in the hard X-ray energy region (20-100 keV), and 3) GBI in the radio band,
of the electromagnetic spectrum. The background noise was removed from the source
PHA file during the fit, and the background PHA file was generated from the model of
background noise for the corresponding epoch of RXTE observation, as provided.
The resolution of the three iron lines (Kitamoto et al. 1994) and the two absorption
edges (Rajeev et al. 1994) are beyond the capability of the PCA, hence we fix the rela-
tive separation of line and edge energies as reported by Rajeev et al. (1994), Nakamura
3.3. General spectral features of Cygnus X-3
Table 3.1: The MJD of the pointed observations of RXTE along with the average flux obtained by ASM (2–10 keV), BATSE
(20–600 keV) and GBI (radio–2.2 & 8.3 GHz).
MJD ASM BATSE GBI GBI spectral MJD ASM BATSE GBI GBI spectral
H
I
H
I
I
(2.2GHz) (8.3GHz) state (2.2GHZ) (8.3GHZ) state
50319 7.495 0.039 - - low 50321 8.160 0.028 - - low
50322 10.298 0.009 - - low 50324 10.291 0.035 - - low
50325 15.279 0.034 - - low 50500 21.141 0.018 0.101 0.098 high
50501 21.29 0.041 0.095 0.090 high 50604 28.722 0.130 0.197 high
50609 22.137 0.001 0.111 0.316 high 50612 17.675 0.042 3.511 2.478 high
50616 26.746 0.005 0.729 0.964 high 50618 20.635 0.034 0.737 0.966 high
50624 32.351 0.056 0.115 0.550 high 50652 7.636 0.055 0.071 0.081 high
50661 8.150 0.051 0.071 0.083 high 50717 10.778 0.038 0.118 0.205 low
50950 5.573 0.043 0.061 0.083 low 50951 6.715 0.052 0.071 0.082 low
50952 5.453 0.046 0.087 0.098 low 50953 5.551 0.051 0.065 0.067 low
50954 5.384 0.058 0.049 0.080 low 51404 19.67 0.087 0.087 high
51585 14.76 0.069 0.223 high 51586 28.59 0.140 0.236 high
51587 26.53 0.044 0.049 high 51588 23.70 0.039 0.073 high
51589 25.98 0.072 0.171 high 51590 31.12 0.137 0.320 high
51592 20.80 0.020 high 51593 19.33 high
51594 19.53 high 51595 19.92 high
51637 31.49 1.211 0.702 high 51638 23.13 0.643 0.644 high
51639 28.41 0.417 0.242 high 51641 20.76 0.498 0.413 high
51642 28.41 high 51643 24.74 high
51644 18.91 high 51646 16.02 0.278 0.124 high
51647 14.11 0.233 0.153 high 51648 13.33 0.229 0.208 high
51650 20.70 0.331 1.759 high 51656 16.54 high
51661 23.48 6.345 3.521 high 51663 19.75 4.536 2.299 high
51575 24.81 high 51576 18.59 high
units: counts s ; photons cm s ; mJy
G
JG
H
JH
JG
I
55
et al. (1993), Kitamoto et al. (1994), and treat the edge energy at 7.1 keV and the nor-
malization of all the lines and edges as the variable parameters in the fit. Throughout
we fix the line width at 80 eV (reasonably accepted value as obtained by Kitamoto et
al. 1994, from ASCAobservations). This mode of tying the line : energies with one edge
energy doesn’t affect the continuum fitting, but brings the value corrsponding to the
best fit parameteres within the acceptable limit. The spectra is fit separately for the
high/soft and low/hard states. The unfolded spectra of an observation in high/soft state
(MJD 50616) is shown in left panel of Figure 3.3, and that of one of the low/hard states
(MJD 50954) is shown in right panel of Figure 3.3. The criteria of choosing these two
as the representative of their respective states are: 1) exposure time, 2) quality of data,
i.e. better background subtraction. Since the resolution of the PCA is poor in the lower
energies and the absorption due to effective hydrogen column is a very sensitive param-
eter in this region, we neglect
:Ì: : data below 5 keV and fix the:Ì: height: of hydrogen column
C
to N value of 1.6 Û
("K l (low/hard state) and 5 Û
("K l (high/soft state), as
reported by Rajeev et al. (1994). Also, the regime below 5 keV is dominated by the
photo-ionization lines (Kawashima and Kitamoto 1996, Paerels et al. 2000), suggesting
reprocessing of the X-ray emission in the hot gas constituting the circumstellar matter,
most probable origin of it is the dense stellar wind of the Wolf-Rayet companion. The
resolution of these low energy features being beyond the capability of RXTE–PCA was
also a reason for neglecting the flux µ 5 keV. A distinct feature of the X-ray spectrum
in the low (hard) state is the shape of the spectrum in the region 10-25 keV, which looks
like a hump like feature. Normally such feautres in X-ray binary blackhole candidates
are taken as signatures of Compton reflection component, but attempts to fit such a re-
flection model provided very high values of the albido. The details of the results of the
X-ray spectral parameter values of the continuum components are given in table 3.2 for
all the low (hard) state observations (11) and some representative high (soft) state ob-
servations (Choudhury and Rao 2002). Subsequently, the background modelling as well
the response matrix of PCA has improved and the picture has changed slightly in the
L
high/soft state ( 3.6).
The high (soft) state continuum emission spectra is best fit by a combination of
multicoloured disc blackbody and CompST (Sunyaev and Titarchuk 1980) components.
(Powerlaw is needed to fit hard X-ray continuum in certain states instead of the mul-
ticoloured disc blackbody, see section 3.6.) Incorporation of any extra component viz.
powerlaw, in addition to the disc blackbody and CompST, doesn’t improve the quality
of the fit. The low hard state continuum emission spectra is best fit by a combination
of CompST and a powerlaw, with the conspicuous absence of the multicoloured (disc)
blackbody component. The X-ray spectra of black hole sources contain a thermal and
3.3. General spectral features of Cygnus X-3 57
Fig. 3.3: A: The spectra of observation on MJD 50616 showing all the components, viz. three
Fe lines, two absorption edges, multicoloured disk blackbody and the powerlaw con-
tinuum. B: The spectra of observation on MJD 50954 showing all the components, viz.
three Fe lines, two absorption edges, compST (Sunyaev and Titarchuk 1980) and the
powerlaw continuum.
a non-thermal part, which are conventionally modeled as a disc black body spectrum
and a power-law (or cut-off power-law) or more realistic models incorporating Comp-
ton scattering from thermal as well as non-thermal electrons (Zdziarski et al. 2001).
Since our aim is to make a wide band description of the spectra to understand the broad
features, we have adopted an analytically simpler model consisting of a disk black-body
and Comptonization from thermal electrons (Sunyaev and Titarchuk 1980 - the CompST
model), plus an additional power law, in the low (as well as hard) state. Iron line fea-
tures, although significant in both states, are more prominent in the low (hard) state.
Absorption edges also form very important spectral features in both the states. Since
the hot circumstellar gas obscures the low energy X-ray emission, the continuum disc
(multicoloured) blackbody emission, contributing in the energy range µ 5 keV needs
more rigorous analysis and better resolved spectra, which is beyond the scope of the cur-
rent analytical procedures and the observatory in consideration. Although CompST is
a non-relativistic model, Vadawale et al. (2002) have shown that CompST model gives
Table 3.2: X-ray spectral parameters of Cygnus X-3 . The results in this table were reported in
Choudhury and Rao (2002), where the then current version of FTOOLS were used
(v5.0). Subsequently better background emission and response matrix have changed
M
the picture little bit for the high (soft) state ( 3.6).
Low/hard state.
L Best fit parameters
:3 for CompST+power law
L :3
MJD kT (keV) % » (d.o.f.) MJD kT (keV) % » (d.o.f.)
50319 4.45 2.47 1.26(86) 50321 4.39 2.51 0.90(108)
ä
50321 4.47 2.61 1.18(108) 50322 4.91 2.51 1.08(108)
ä
50322 4.36 2.70 0.92(108) 50324 4.17 2.67 0.77(108)
50325 5.58 2.45 0.60 (108) 50717 5.09 2.55 0.74(86)
50950 4.97 2.10 1.26(88) 50951 4.74 2.08 1.34(89)
50952 5.02 2.03 1.43(86) 50953 5.06 2.02 1.45(91)
50954 4.87 2.01 1.42(108)
High/soft state. Best fit parameters :3 for diskbb+power law :3
MJD N
kT (keV) % » (d.o.f.) MJD N
kT (keV) % » (d.o.f.)
50604 1.49 2.55 0.53(109) 50609 1.59 2.21 1.19(109)
50612 1.62 2.25 1.22(109) 50616 1.55 2.53 0.65(109)
50618 1.53 2.34 0.83(109) 50624 1.52 2.63 0.60(109)
50500 1.74 2.98 0.55(109) 50501 2.91 3.06 0.54(109)
ä
50501 2.56 2.98 0.59(109)
* extended observation
a functionally correct description of the more elaborate numerical codes, although with
slightly different parameters. Moreover, such composite models have been used earlier
for Cygnus X-3 (Rajeev et al. 1994).
The most striking feature of the X-ray wide-band spectra obtained from RXTE ob-
servations is the unambiguous presence of the non-thermal tail extending beyond 150
keV (Choudhury and Rao 2002), in both the states of X-ray emission. Such features are
commonly observed in the Galactic black hole binary candidates, viz. GRS 1915+105
(Zdziarski et al. 2001), Cygnus X-1 (Zdziarski et al. 2002), GX 339-4 (Wardziński et
al. 2002), and also some neutron star binary systems. For the black hole systems this
power law tail can be interpreted as due to two component (thermal and non-thermal)
Comptonization (Zdziarski 2000), or due to the existance of the bulk motion Comp-
tonization (Chakrabarti and Titarchuk 1995), espcially in the high/soft state. Detailed
X-ray spectral analysis, correlated with the radio emission, will be dealt with in the
following sections, and their theoretical interpretations will be presented in chapter 6.
3.4. Correlation of radio & X-ray emission in Cygnus X-3: Spearman’s Partial Rank Correlation test59
Fig. 3.4: The combined simultaneous lightcurve of Cygnus X-3 in the soft X-ray (2 -12
keV,RXTE–ASM, top panel), hard X-ray (20 - 100 keV, CGRO–BATSE, middle panel)
and the radio (2.2 GHz, GBI, bottom panel). The various ‘states’ of the source are
separated by vertical dashed lines and identified with numbers. The arrows on the top
panel give the start time of RXTE pointed observations and the three inverted arrows
give the days for which wideband X-ray spectral studies are carried out.
3.4 Correlation of radio & X-ray emission in Cygnus X-3:

Spearman’s Partial Rank Correlation test
Cygnus X-3 is one of the brightest radio source ever associated with an X-ray binary
(Waltman et al. 1995). Since its discovery in the radio band (Braes and Miley 1972,
Hjellming et al. 1972), it has been found to be a persisting source in this band, exhibit-

ing regular radio outbursts (Gregory et al. 1972, Hjellming and Ballick 1972). These
large outbursts, which may get brighter than 10 Jy, were found to occur only during
the high (soft) state of X-ray emission (Watanabe et al. 1994). Normally, there is a period
of quenched radio emission, with flux µ 30 mJy, preceding the loud outburst episode
(Waltman et al. 1996, Fender et al. 1997b). Watanabe et al. (1994) also hint at a correla-
tion between the soft X-ray (as observed by ASM aboard the Ginga observatory) and the
quiescent radio emission (as seen by the GBI), during the low-hard state, but a detailed
correlation analysis is not presented, possibly due to the sparse sampling of the source
by the GINGAsatellite during its course of regular monitoring. McCollough et al. (1999)
report a correlation test between the hard X-ray (as observed by the BATSE aboard the
CGRO) and the radio (GBI), in the various states of radio and X-ray emission. They
report 1) anti-correlation between the radio and hard X-ray emission during the quies-
cent period, 2) correlation between the radio and hard X-ray emission during the major
flaring period, and 3) no correlation between the radio and hard X-ray emission during
the minor flaring period. Continuing our mission of the study of long term multiwave-
length (radio & X-ray) monitoring of this source, we perform a Spearman’s Partial Rank
Correlation test, to analyse the association between the radio (2.2 GHz, GBI), soft X-
ray (2 - 12 keV, RXTE–ASM) and hard X-ray (20 - 100 keV, CGRO–BATSE), covering
the period when these three observatories were simultaneously monitoring the source.
The Spearman’s test provides the most comprehensive correlation among three variables
(Macklin 1982), investigating the possible role of the third variable in the correlation
that (may) arise between the other two variables.
3.4.1 Spearman’s Partial Rank Correlation test

The Spearman partial rank correlation test is used to determine the correlation between
two or more variables. The partial rank coefficient is computed from the sampling dis-
tribution which may be derived by analogy with a parametric statistic (Macklin 1982).
For correlation among three variables, say A, B & C, the null-hypothesis is that the cor-
relation between A and B arises entirely from those of C with A and B separately. The
value of the correlation coefficient lies between -1 and 1. The negative value signifies
anti-correlation. The significance level associated with the correlation between A and B,
independent of C, is given by the D-parameter, which is normally distributed about zero
with unit variance if the null-hypothesis, that the A-B relation arises entirely from those
of C with A and B separately, is true (Macklin 1982).
Three variables can be mutually correlated, but independent, in the following way:
Let X, Y, Z be three variables which are completely independent of each other. Define
A, B, C as follows:
A=X+Y
B=Y+Z
C=Z+Y
We can see that A, B, C are mutually correlated, but each correlation is independent
of the third parameter. We have explicitly verified this by taking 200 sets of X, Y, and
Z (each of them a random number) and calculating the correlation coefficient and D
3.4. Correlation of radio & X-ray emission in Cygnus X-3: Spearman’s Partial Rank Correlation test61
( 0.5) while the D parameter -

+
parameter for A, B, and C defined above. We find significant correlation among all three
for all the three cases indicate that each correlation
is independent of the third parameter.
On the other hand, defining the 3 parameters as:
A=X
B=X+Y
C=X+Z
shows a D parameter of 0.3 for B:C correlation showing that this correlation is en-
tirely dependent on the common parameter A.
The Spearman’s Rank Correlation (SRC) coefficient and the realted parameters (D -
parameter & null hypothesis coefficient) were computed for the monitoring data obtained
from RXTE–ASM, CGRO–BATSE and GBI, as explained in the previous section. The
lightcurve of the period covered in this correlation test is shown in Figure 3.4, where
the daily averaged flux level is depicted. Historically, the behaviour of radio emission

in Cygnus X-3 is classified into: 1) quiescent period ( 50–100 mJy), 2) major flaring
( 1Jy) with a preceding quenched state ( 10–20 mJy), and 3) minor flaring (

100–150mJy) with partial quenching state (see Waltman et al. 1995, McCollough et al.
1999). Accordingly we have demarcated four regions in Figure 3.4, region 1 and 4
corresponding to the quiescent state (although region 4 contains two minor flares along
with the long quiescent period), region 2 corresponding to the major flaring state and
region 3 corresponding to the minor flaring state. The number of data points in region
1 (of Figure 3.4) is meagre and hence is not included in the correlation tests. Table 3.3
shows the SRC coefficient, null hypothesis probability and the D-parameter, using 10
day averages of the data for the different periods (and their combination) as demarcated
in Figure 3.4. Reducing the number of days for averaging does not significantly change
the results.
The most interesting result is that the soft X-ray and radio are very strongly cor-
related, with a very high significance
~ level, in region 4 and regions 3 & 4 combined.
It can also be ruled out (at - 5 level) that this correlation is influenced by the third
parameter, the hard X-ray emission. Though these two parameters are correlated even
in the flaring state (region 2) at a much reduced significance level, the correlation tests
in this region could be influenced by the high variability at time scales shorter than the
period chosen for taking the averages (10 days). Hence we concentrate on the results
obtained for region 4 and region 3 & 4 combined. It is also found that the soft X-ray is
anti-correlated with the hard X-ray emission. The anti-correlation between the radio and
hard X-ray emission, though strong, could be influenced by the other two correlations,
particularly when we examine the data for the region 3 & 4 combined. The similarity of
Table 3.3: The Spearman Rank Correlation (SRC) coefficient, null-hypothesis probability and
D-parameter between the radio, soft X-ray and hard X-ray for different periods de-
marcated in Figure 3.4.
SRC coeff. Null Prob. D-Parameter
Region 4
ASM:GBI 0.84 6.3 Û
l Ì 5.2
GBI:BATSE 0.75 9.2 Ñ
1l Ì 2.7
ASM:BATSE 0.74 3.1 Ñ
l Ì 2.2
Region 3
ASM:GBI 0.66 8.3 Û
l x: 2.7
GBI:BATSE 0.43 4.7 Û
l 0.4
ASM:BATSE 0.71 1.9 Û
l x 3.2
Region 2
Q
ASM:GBI 0.56 3.5 Û
l 4.1
GBI:BATSE 0.10 6.3 Û
l : 2.7
ASM:BATSE 0.50 1.1 Û
l 3.8
Region 3 & 4
ASM:GBI 0.83 2.1 Û
l
: O 6.2
Q
GBI:BATSE 0.72 4.7 Ñ
1l 1.5
ASM:BATSE 0.79 4.1 Ñ
l QP 3.8
the behaviour of region 4 and regions 3 & 4 combined suggest that the emission mech-
anism during the minor flaring and the quiescent period are the same, and hence these
two can be clubbed together as one class (see Choudhury et al. 2002). This suggests
that the X-ray spectral behaviour controls the radio emission during this state of low flux
core-jet outflow. Figure 3.5 shows the variation of flux in radio (correlated) and hard
X-ray (anti-correlated) with soft X-ray flux during the low-hard state (region 4 of Fig-
ure 3.4), which clearly demonstrates the simple monotonic dependence of the both with
the soft X-ray (ASM) during the few hundred days when the source was in this state.
This result vindicates the study of McCollough et al. (1999) and (Watanabe et al. 1994).
The positive correlation between the soft X-ray and the radio flux, during the (low) hard
state is the strongest seen in such systems. The strong and significant anti-correlation
between the hard X-ray and both soft X-ray as well as radio is also a unique feature not
reported before for any other binary microquasar system. A detailed spectral analysis to
investigate the apparently contradictory results of the radio – X-ray correlation (soft and
hard) was the next necessary step needed to probe the disc - jet connection in the system.
3.5. X-ray spectral pivoting in the low (hard) state 63
Fig. 3.5: The variation of flux in radio (GBI, 2.2 GHz, shown as stars) and hard X-ray (BATSE,
20-100 keV, shown as circles) with soft X-ray (ASM, 2-10 keV), during the low-hard
state of Cygnus X-3 (region 4 of Figure 3.4). Each data point is an average value for
10 days.
3.5 X-ray spectral pivoting in the low (hard) state
Most observations of Cygnus X-3 by RXTE are done in the high (soft) state, with the aim
of studying the X-ray spectra during the radio flaring episodes. In the low (hard) state,
there are about ten indpendent observations, during this period of simultaneous monitor-
ing (shown in the top panel of Figure 3.4), or otherwise, comprising three major group
of separated observations. We present the X-ray spectra of three such observations that
span the range of observed X-ray and radio fluxes within the precincts of the low (as well
as hard) state. In Figure 3.6 the three unfolded spectra are overlaid on the top panel with
the PHA ratios for the two extreme spectra shown in the bottom panel. Some salient fea-
tures of these spectra are given in table 3.4. As explained in the previous section, during
the low-hard state, encompassing the regions 3 & 4 of Figure 3.4, the continuum spectra
(5 – 150 keV) is best described by the Comptonization of seed photons from a ther-
mal multi-coloured accretion disk by a thermal Comptonizing plasma cloud (CompST
- Sunyaev and Titarchuk 1980, Nakamura et al. 1993, Rajeev et al. 1994) along with
a non-thermal powerlaw emission (Choudhury and Rao 2002, Choudhury et al. 2002).
The iron lines and absorption edges are fit by following the recipe expounded in the pre-
Fig. 3.6: Top panel: The unfolded X-ray spectra of Cygnus X-3 during the radio quiescent
period on three different occasions (1: MJD 50954; 2: MJD 50661; 3: MJD 50717).
Bottom panel: PHA ratio of spectra from the extreme observations.
L
vious section. The best fit values of the electron temperature (kT ) of the Compton cloud
%
and the powerlaw photon index ( » ) for the three spectra are given in Table 3.4. Other
simple models like cutoff powerlaw (along with a powerlaw), broken power-law, etc. do
not consistently fit all the observed spectra with physically feasible parameter values and
give much inferior fits. It can be seen from the figure that there is a systematic change in
the shape of the spectrum with increasing soft X-ray flux (and radio flux, see Table 3.4).
The X-ray spectrum, when the radio flux is low, is quite flat in the 20 – 40 keV region
and the spectral curvature increases with the soft X-ray flux.
The spectral evolution, during the (low) hard state can be clearly intrepreted as a
pivotal behaviour correlated to the radio emission, with the pivot point lying in the energy
range 10 – 20 keV. The total flux in the wide energy band (5-60 keV) in X-ray remains
constant, the increase in soft X-ray (5-12.5 keV) is at the expense of corresponding
decrease in the hard X-ray (12.5-60 keV) flux (Table 2). Thus, in the low-hard state,
the following picture emerges: 1) the physical process causing the soft X-ray emission,
generally attributed to the accretion phenomenon (inflow of matter), is the causal factor
determining the resultant flux of radio emission (outflow of matter), 2) the radio emission
is coming from a rather weak jet (outflow), embedded inside the thermal plasma cloud
3.6. X-ray spectral evolution driving the radio flares: high (soft) state 65
Table 3.4: The observed flux and X-ray spectral parameters of Cygnus X-3 during the three
pointed RXTE observations.
MJD
50717 50661 50954
A.Flux
ASM (cts s l ) : 11.11 8.18 5.37
BATSE (ph cm l È l ) 0.038 0.051 0.058
GBI-2.2GHz (mJy) 115 64 43
GBI-8.3GHz (mJy) 165 73 53
D
B. LBest fit parameters for CompST+power law
D + D
kT (keV)
» % 2.55
D
5.09 .
((
D
4.37
2.19 ,
D
4.87
2.01
+
:3
(d.o.f.) 0.74(86) 1.80(59) 1.42(108)
whose opacity determines the amount of radio flux coming out, 3) the physical processes
giving rise to soft and hard X-ray emissions are competing with each other, resulting
in the anti-correlation. The decreasing opacity of the thermal Comptonizing plasma
cloud plays a major role in the X-ray state transition from low-hard to the high-soft state
(Rajeev et al. 1994), with a corresponding change in the radio emission. Here we show
that it plays a similar role in a very small scale of change within the precincts of the
low-hard state.
Such a pivoting behaviour, with the soft X-ray flux correlated to the radio emission
and the corresponding hard X-ray flux anti-correlated to the soft X-ray flux as well as
radio emission, had not been reported for any microquasar system. This unique observa-
tional feature provides a unique insight into disc - jet connection in such binary systems.
This establishes, with an aura of definiteness, the close connection between the physical
mechanisms that give rise to the accretion and ejection phenomena. This result will be
generalised for other prominent Galactic microquasar systems in chaper 5 and a phen-
emenological interpretation will be attempted in chapter 6. In the following section,
we continue the study of the long term X-ray spectral evolution, correlated to the radio
emission, in the (high) soft state.
Fig. 3.7: The soft X-ray (RXTE–ASM, 2-12 kev) and radio (GBI, 2.2GHz) monitoring of the
source during the X-ray high state. The days for which the X-ray spectra obtained
from the pointed observations using RXTE–PCA are reported here are indicated by
arrows. The insets in the two right hand panels highlight the minor flares. The colour
of the pointed arrow denotes the particular phase of the X-ray spectra, as observed by
the RXTE–PCA. The colour scheme is as follows: purple R radio quiescent phase,
R R
pink pre-radio flare & orange post-radio flare.
3.6 X-ray spectral evolution driving the radio flares: high

(soft) state
The Spearman’s Partial Rank Correlation test doesn’t reveal any appreciable correlation
between the radio and the high energy emissions in the (high) soft state, compared to
the very strong and significant results in the (low) hard states. The chief reason for
this is perhaps two-fold, 1) the dynamical time scale of the variability in this state is
faster than a few days, 2) and more importantly this state incorporates more complex
evolution, which includes the major radio flares preceded by quenched radio emission,
with the X-ray spectral evolution deviating from the linear monotonic pivotal behaviour.
Hence the only procedure available to monitor the radio – X-ray association is to study
the spectral evolution from the pointed observations from RXTE–PCA along with the
daily monitoring in the radio band by GBI. Scouting through the complete date set of
observations made by RXTE–PCA of Cygnus X-3, during the period when RXTE and
GBI were serendipitously monitoring the source, we found four episodes of radio flaring,
minor as well as major, well covered in the X-ray regime. These episodes are depicted
in Figure 3.7, with the arrows depicting the days of RXTE–PCA observation analysed
and reported here.
The X-ray spectra in this state is generally dominated by the thermal multicoloured
disc black body component, along with a hard component best described by a thermal
COmptonizing model, viz. CompST model Sunyaev and Titarchuk (1980), except for
the post flare phase, when the spectral shape hardens in the soft X-ray region, and the
spectral components that fit the spectra best in this phase is the combination of CompST
and power law, identical to the (low) hard state, although the spectral state remains in
the high state. The X-ray spectra may be classified into three phases:-
The radio quiescent phase. The X-ray spectra has a strong disk black body and an
equally strong Comptonising component.
Pre-radio flare. The Comptonising component becomes very small (near vanishing),
resulting in a flare. The flare may result in a time scale of a day or less (minor
ones).
Post-radio flare. The succession of radio flares, both minor as well as major, is stopped
by the change in the X-ray spectrum, with the spectral shape hardening in the
soft X-ray region. During this phase the disk black body component becomes
insignificant, the spectral shape is explained by the similar combination of non-
thermal components, viz. CompST and power law.
The above classification is done by comparing the relative flux of the individual
model components viz., disc black body and CompST, the identification of the post-radio
flare phase is easily achieved by the disappearance of the disck black body component
and its replacement by the powerlaw component (table 3.5). In Figure 3.7 the colour of
the pointed arrow denotes the particular phase of the X-ray spectra, as observed by the
RXTE–PCA. The colour scheme is as follows: purple D radio quiescent phase, pink D
pre-radio flare & orange D post-radio flare. In Figure 3.8, the same colour represents
the total spectrum in the corresponding phase, with the individual components of the
spectra fowllows the following convention: green D disc blackbody, blue D CompST
and red D powerlaw.
The evolution of the X-ray spectra during the each individual flaring episodes, as
covered by the RXTE pointed observations are as follows:-
$ MJD 50490 – 50510. There are essentially two independent X-ray pointing dur-
ing this flare; on MJD 50495, the spectra is in the post-radio flare phase, with the
CompST and the powerlaw component. On MJD 50500 the X-ray spectra is in
the phase of the quiescent phase, with the disc black body reappearing (although
it contributes only 25% of the toal flux, see table 3.5) in place od the powerlaw.
There aren’t any X-ray observation during this episode, although the source ex-
hibits minor radio flares after a few days.
$ MJD 50600 – 50650. There are ten X-ray pointings during this episode, and we
present the spectra of three such occasions, which are adequate to elucidate the
spectral evolution. On MJD 50604 the spectra, described by disc black body &
+
CompST, has only 10% of Comptonizing component and susequently there is a
minor radio flare ( µ mJy) the following day, followed by loud radio flares

( . Jy) after about six days. This state of depleted Comptonizing component
persists with the reoccurance of the minor and major flares, for eg. on MJD 50624
+
the Comptonizing component has only 15% contribution, and the next day there
is a minor flare ( µ mJy). This succession of the flaring episodes come to a
rest after the disc black body component vanishes (MJD 50632) and replaced by a
powerlaw. The spectral components are identicial to those of the (low) hard state,
albeit with high soft X-ray flux, and the source subsequently comes down to the
(low) hard state, with the corresponding pattern of radio emission.
$ MJD 51580 – 51590. During this episode of mnor flaring, four X-ray spectra are
enough to bring forth the pattern vanishing Comptonizing flux preceding a flare.
On MJD 51586 the CompST component contributes
+ to the total flux, and

within the span of a day there is a minor radio flare ( /
mJy), which subsides
Fig. 3.8: The X-ray spectral energy distribution (SED) and the individual continuum compo-
nents, during the radio quiescent, pre-radio flare & post-radio flare phases. The qui-
escent phase has disk black body and Comptonising component at near equal ratio,
the pre-radio flare has vanishingly small Comptonising component, and the post-radio
flare has the disk black body component replaced by a simple power law. The colour
S
coding of the total spectra and their components are as follows: Total spectra purple
R R R
radio quiescent phase, pink pre-radio flare & orange post-radio flare; individ-
ual continuum components S R
green disc blackbody, blueR CompST and red R
powerlaw.
Table 3.5: Model parameters of the continuum components and their flux contributions of the
X-ray SED
Quiescent Radio Emission
MJD Total Flux

T û éëê
(keV) Flux
Disk Black Body
% of Total Flux
T ûSü (keV) U Flux
CompST
% of Total Flux
50500 7.9 1.8 1.9 24.05 23.93 2.22 6.0 75.95
51587 6.6 V 1.44 3.4 51.51 13.39 1.59 3.2 48.49
51588 5.1 1.42 3.7 72.55 21.69 1.39 1.4 27.45
Pre-Radio Flare
MJD Total Flux

T
û éëê (keV) Flux
Disk Black Body
% of Total Flux
T û ü (keV) U Flux
CompST
% of Total Flux
50604 6.6 1.53 6.0 90.91 18.27 4.07 0.6 9.09
50624 6.9 V 1.55 5.9 85.51 18.34 3.31 1.0 14.49
51586 5.3 1.56 4.4 83.02 42.31 3.41 0.9 16.98
51589 4.5 1.53 3.8 84.44 53.53 2.40 0.7 15.56
51646 3.6 1.63 3.3 91.67 54.33 2.53 0.3 8.33
51650 5.8 1.70 5.0 86.21 80.79 9.91 0.8 13.79
Post-Radio Flare
MJD Total Flux

V Flux
Power law
% of Total Flux
T û ü (keV) U Flux
CompST
% of Total Flux
50495 8.5 2.43 4.2 49.41 4.03 8.12 4.3 50.59
50632 9.9 V 2.62 4.0 40.40 5.12 7.05 5.9 59.60
51676 7.9 2.62 2.7 34.18 6.18 6.36 5.2 65.82
X
units: ¨ e'W0X0Y0Z\[]_^e [¯e
V V
quickly and on MJD 51587 the disc black body and the Comptonizing components
are equally strong, with the disc blackbody component gaining strength on the next
day (MJD 51588) with 75% of the flux. Continuing the trend of the disappearing
of the Comptonizing cloud, on MJD 51589 the Comptonizing flux is only £
!,+ ,

and on MJD 51590 there is a minor flare (
mJy). This episode shows that a
minor flare is brought about after the CompST flux gets to £
!,+ or lower.
$ MJD 51630 – 51680. During this episode of minor flares ( I mJy) intermixed
Comptonising components are

+
with huge radio flares (
!. Jy), we show that on MJD 51646 and 51650 the
+ and
+ , with the minor flares occuring
nearly throughout this whole episodes, followed by the huge radio flare on MJD
51652. The sampling of the radio monitoring is sparse here, but trend of the
data shows convincingly that the radio flares subside after the disc black body
component vanishes in the X-ray spectra, replaced by the powerlaw.
The general picture of the X-ray spectral characteristics in the radio flaring states can
3.7. Complete X-ray spectral evolution 71
be summed up as follows:-
1. The radio quiescent emission is marked by the radio emission (2.2 GHz) bordering
around 110 mJy and below. The X-ray spectra has strong CompST component
with the contributing flux of this component amounting (,+ ,+ of the total
flux.
2. The vanishing of the CompST component (flux going below 15%) always pre-

cedes a minor flare, with the radio flux around
!W mJy, suggesting the
ejection of the central Compton cloud resulting in the flare. The stronger the ejec-
tion, the louder the flare.
3. The minor flare may be followed by the filling of the central Compton cloud,
i.e. increase in CompST flux, causing the radio emission to become quiescent.
Otherwise, if the continuous accretion persists with the central cloud unfilled, i.e.
a`
the CompST flux remains low, a major radio flare (2.2 GHz, flux -I
) follows.
4. The continuing series of minor and major flares come to an end only with the
change in the X-ray spectra, i.e. hardening of the soft X-ray band, with the flux
level remaining high. This is the most interesting state of the X-ray spectra with
the shape being best fit by the model of the low (correspondingly hard) state, i.e.
power law and CompST, although the soft X-ray flux remains high. This change
in the X-ray spectra puts a brake in the episodes of radio flaring.
3.7 Complete X-ray spectral evolution

We report here the complete X-ray spectral evolution of the microquasar Cygnus X-
3, associated to the radio emission, for the first time for this source. The (low) hard
state is best characterised by a non-relativistic thermal Comptonizing component, along
with an additional (non thermal) powerlaw component. The radio emission increases
monotonically with the increasing soft X-ray flux, anti-correlated to the hard X-ray flux,
which is beautifully elucidated by the pivoting in the spectra. The absence of a clear
black body component in this state may be attributed to the obsecuring of the source
by the circumstellar material. After the state transition into the (high) soft state, the
disc black body component becomes prominent, and the spectra in this state is normally
characterized by a multicoloured disc black body and CompST (Sunyaev and Titarchuk
1980). The Comptonizing component, characterised by the CompST model, becomes
negligibly small preceding a minor radio flare, and on persistance of this state a major
radio flare follows. The succession of the flares are put to stop by the vanishing of
the disc black body component, which is replaced by the powerlaw component. The
system, at this phase may continue to remain in the ‘high’ state with the disc black body
reappearing, or it may susequently transit to the ‘low’ hard state.
Chapter 4
Cygnus X-3: temporal studies
The study of the temporal behavioural pattern of X-ray sources consists of mainly two
broad classes, the long term variability which are normally studied by monitoring de-
vices viz. ASM aboard the RXTE, and the short term variability studied by the pointed
observations. The long term behavioural pattern of the X-ray variability has been ex-
plained to the complete detail (given the observational capabilities of the current gener-
ation X-ray detecting instruments) in the last chapter. In this chapter we will first review
the long term properties of Cygnus X-3 after performing the correction due to binary
orbital modulation of the X-ray emission, thereafter we will discuss the short term vari-
ability properties of the X-ray emission from this source.
4.1 Binary modulation and correction with a given

ephemeris
The X-ray source Cygnus X-3 was detected to be a binary system within a few years of
its discovery. The binary modulation is seen in the X-ray (Parsignault et al. 1972, Sand-
ford and Hawkins 1972) and the infra-red bands (Becklin et al. 1973). The evolution
of the 4.8 hour binary period of Cygnus X-3 has been studied extensively with the time
derivative ( Á ) measured to be 10 lr (van der Klis and Bonnet-Bidaud 1981, 1982,
¸
1989, Molnar 1988, Kitamoto et al. 1987). Recently, Singh et al. (2002) have extended
the data base of the binary period measurements using the The Indian X-ray Astron-
omy Experiment (IXAE)and archival data from ROSAT, ASCA, BeppoSAX and RXTE.
They found that the binary template obtained from the EXOSATdata adequately explains
D

the recent observations and have derived a value of the period derivative of 5.76 0.24
10 l . Inclusion of second derivative marginally improved the fit, essentially giving
73
74 Chapter 4. Cygnus X-3: temporal studies
b
an upper limit to the second derivative ( = -1.3 1.6 10 l Ì yr l ). In order to note
D
¸
the applicability of the binary correction with the given ephemeris, we have folded the
RXTE–ASM data using the quadratic ephemeris of Singh et al. (2002) from MJD 50410
to MJD 51400 (P = 0.19968443 d; Á = 5.76 10 l ; time of zero phase T = 2440949.892

¸
JD). The time span chosen for the folding corresponds to the data used by Choudhury et
al. (2002) for the detailed X-ray, radio and hard X-ray correlation analysis. The folded
light curve is shown in Figure 4.1 (the errors in the data points are smaller than the sym-
bol size). The template data from the EXOSATobservations (van der Klis and Bonnet-
Bidaud 1989) is also shown in Figure 4.1, after appropriate scaling. The relative phase
of the data and template are not arbitrarily shifted for a proper match: the folded light
curve (with the above ephemeris) is overlaid with the vertically shifted and scaled tem-
plate data. The folded count rate correlates with the template value with a correlation
coefficient of 0.95 (for 64 data points). It can be seen from the figure that there are non-
statistical variations due to the source over and above the binary variation. The ASM
count rate, C, can be expressed as
7£
D c y
. 1
.
D
.ed (4.1)
where V is the template value for a given ~ phase as given in van der Klis and Bonnet-
Bidaud (1989). The errors are nominal 1 errors obtained by assuming that the fluctua-
tions in each phase bin is random in nature.
During this period, Cygnus X-3 shows several X-ray flares and also transition be-
tween low-hard to high-soft states (see Figures 3.3 and 3.8). Considering the large vari-
ation of the X-ray flux during the period covered, the folded light curve agrees with the
template quite well. We have examined the fluctuations in the source counts over and
above the binary variations with respect to binary phase as well as the observation dura-
tion. Though there is some evidence for large flare like variations during the phase 0.2
to 0.5, the rest of the fluctuations occur throughout without any association with binary
phase or time of observations. Note also that we have not done any phase fitting for
this analysis. Hence we conclude that the error in the phase of light curve minimum is
negligible compared to the overall source fluctuations.
Using this information, we can correct each observation for the binary modulation,
qs
as follows. The quadratic ephemeris to get the zero phase for the < cycle of the period
is given as (Singh et al. 2002):
4
ñ
7 4 y qzy ; q : (4.2)
¸
where c = : Á . This equation can be inverted to give the binary phase at any time T
¸ ¸
4.1. Binary modulation and correction with a given ephemeris 75
Fig. 4.1: The X-ray light curve of Cygnus X-3 obtained from RXTE–ASM for the period of MJD
50410 to 51400 folded at the quadratic ephemeris given by Singh et al. (2002). The
template data obtained from EXOSAT(van der Klis and Bonnet-Bidaud 1989) is also
shown, with appropriate scaling for the count rates.
as,
q 54 ³4 54 4 :;
7 Q (4.3)
¸ ¸
For the RXTE–ASM data, we have calculated the phase using equation 4.3 and from the
corresponding template value, corrected for the linear term in equation 4.1. The binary
corrected ASM data is again folded and the folded light curve is shown in Figure 4.2. It
can be seen from the figure that for the long term monitoring data a quadratic ephemeris
can explain the modulation quite well and this ephemeris may be used to correct the
data for the binary variation. However, we caution, that despite the stable profile with
a monotonic change in its period for several years, Cygnus X-3 may show jitters in the
individual binary phase measurements (see Singh et al. 2002). This implies that either (a)
there is an inherent jitter in the minimum of the binary phase or b) it has a stable profile
and short term variability is superimposed on it. We have made the binary correction for
the subsequent analysis assuming that Cygnus X-3 has a stable profile. The implication
of the possible random phase jitter is examined at appropriate places, if required.
We have applied the binary correction to the individual RXTE–PCA observations
and in Figure 4.3 we show both the uncorrected (top panel) and corrected (bottom panel)
Fig. 4.2: The X-ray light curve of Cygnus X-3 obtained from RXTE–ASM for the period of
MJD 50410 to 51400 folded at a binary period of 4.8 hours after the correction for the
quadratic ephemeris given by Singh et al. (2002).
lightcurve of the single longest observation (Obs. Id. 10126-01-01-020 & 10126-01-
01-02) of the source by RXTE. The observation covers about two and a half binary
cycles, punctuated by the necessary breaks due to Earth occultation and various other
data dropouts. The lightcurve is obtained from single bit data (2–6.5 keV) with all the
+
five PCUs on. The binary template (Figure 4.1) is asymmetric with an unusually broad
peak lasting in the Ã phase of the binary period with a gradual rise before
and a steep fall after the peak. It is evident that the correction for binary modulation is
very good for the rising and falling phase of the binary ephemeris, highlighting the small
variations which were otherwise smothered by the binary modulation. Again we empha-
size here that we have not done any fitting for deriving the zero phase, but have used the
quadratic ephemeris to derive it. During the peak the lightcurve shows fluctuations not
correctable by the smooth peak of the template. This, generally random, fluctuation is
an inherent feature of the source present in all the observations, past and present (see van
der Klis & Bonnet-Bidaud van der Klis and Bonnet-Bidaud (1982, 1989).
Perhaps, one of the most noteworthy features of Cygnus X-3 is that the binary pe-
riod has remained consistent for more than 25 years with a linear decay with no second
derivative of the period (Singh et al. 2002). The relative phasing of the X-ray and infra-
4.1. Binary modulation and correction with a given ephemeris 77
Fig. 4.3: The X-ray light curve of Cygnus X-3 obtained from RXTE–PCA pointed observations
(top panel), shown along with light curve (bottom panel) corrected for the binary vari-
ation using the quadratic ephemeris given Singh et al. (2002).
red emissions from this source, coupled by the orbital evolution of the line shift of the
He I emission lines, gives the picture that the binary modulation is due to the orbital mo-
tion of the ionized two temperature wind originating in the companion Wolf-Rayet star
irradiated by the X-rays from the compact object (van Kerkwijk 1993). The minimum
occurs when the cooler part of the wind in the shadow of the Wolf-Rayet companion is in
the line of sight of the observer with the compact object at the superior conjunction. The
asymmetry in the binary modulation profile may be due to 1) eccentricity in the binary
orbit, and/or 2) asymmetric distribution of matter within the system (Elsner et al. 1980).
Given the fact that both the binary period and the template have remained consistent for a
period of 25 years, the presence of any reasonable apsidal motion (Ghosh et al. 1981)
may be ruled out (van der Klis and Bonnet-Bidaud 1989). Hence, eccentricity, if present
in the binary orbit, is negligible to effect any asymmetry in the lightcurve. Therefore it
is more likely that an asymmetric distribution of matter, causing an asymmetric distribu-
tion of optical depth, introduces the asymmetry in the binary template. In this scenario
the orbital period decay is explained by the loss of angular momentum via the wind from
the Wolf-Rayet companion.
Since the early days of its observation, it is believed that in this source the X-ray
emission is extensively reprocessed, either in the stellar wind (Davidsen and Ostriker
1974, Hertz et al. 1978, Becker et al. 1978) or in a cocoon (X-ray halo) surrounding and
extending beyond the binary system (Milgrom 1976, Predehl et al. 2000). There also
exist models of X-ray reprocessing in the accretion disk corona (White and Holt 1982,
Molnar and Mauche 1986) where the low energy photons are upscattered by Comp-
tonization also producing fluorescence of ionized iron (Rajeev et al. 1994). Nakamura et
al. (1993) explain the X-ray spectra as obtained by the GINGA observatory by propos-
ing the presence of three species of ionized gas, fully ionized, almost fully ionized and
nearly ionized, engulfing the binary system. Fender et al. (1999a) suggest a WR type
wind with the geometry of a disc in the binary plane with a size much bigger than the
binary system to be the origin of the binary modulated He emission lines (obtained from
infra-red spectra). This wind originates from the companion which is a WN type Wolf-
Rayet star. Polarimetric study of the K-band lightcurve also suggests a preferential plane
of scattering. In this model the X-ray emission undergoes scattering in this disc-like two
temperature wind (van Kerkwijk 1993), resulting in the asymmetry in the binary tem-
plate.
The residue of the (binary) folded lightcurve Figure 4.2 may be attributed to the long
term variation of the X-ray emission by virtue of the change in wind and/or cocoon mass
distribution, including various state changes from soft (and high) to low (and hard) and
vice-versa, accompanied by the correlated radio flares. One small aspect not considered
so far is the generally random fluctuation in the X-ray lightcurve at the peak of the
binary ephemeris. A very detailed analysis of the nature of this fluctuation may help in
determining the geometrical and physical structure of the accreting system involving the
wind from the companion and the X-ray halo (Predehl et al. 2000) engulfing the system.
Recently, Stark and Saia (2003) have attempted to constrain the binary orbital parameters
by measuring the line shift of Helium & line of Fe XXV and Lyman & lines of Si XIV
and S XVI as a function of the orbital phase. By ascribing the Fe line to originate
very close to the compact object surface and the Si and S line to originate from the
wind from the companion, they constrain the masses of the companion and the compact
object to be µ 7.3 and µ 3.6 , respectively, provided the orbital inclination is
small (i=24 ¶ - this agrees with the limit 0-40 ¶ of Fender et al. 1999a). The phase resolved
spectra of the various iron emission lines (6.4 keV, 6.7 keV & 6.9 keV) and the absorption
edges (7.1 kev & 9.1 keV) during the both (low) hard and (high) soft states may provide
better constraints on the origin of the iron emission lines. The binary modulation of the
emission lines reported by Stark and Saia (2003) need to be fit properly with the binary
template (Singh et al. 2002) to obtain a more definitive mass function of the system. A
long continuous X-ray observation of the source spanning a few orbital periods with very
4.2. Radio X-ray correlation of Cygnus X-3 79
high resolution X-ray spectra will provide a better binary profile which is essential for
constraining the binary parameters and obtaining the geometrical and physical structure
of the system.
Given the consistency of the binary orbital modulation template of the X-ray emis-
sion as obtained from the EXOSATobservations (van der Klis and Bonnet-Bidaud 1989)
for over a( years, we employ this template for the correction of the modulation of X-
ray emission using the ephemeris of Singh et al. (2002) for the timing studies presented
in the following sections.
4.2 Radio X-ray correlation of Cygnus X-3

Cygnus X-3 possesses a binary orbital period of 4.8 hours, therefore statistical inference
gleaned from daily averaged monitoring data, which covers five orbital periods, should
sufficiently be independent of any effects of binary modulation. Nevertheless, in order to
establish the correlation result presented in chapter 3 (where the SRC coefficient reported
was obtained from ten days averaged flux, covering fifty binary orbits), we repeated
the SRC test for the binary corrected X-ray emission (Choudhury et al. 2004). The
correlation coefficient and other parameters (Choudhury et al. 2002, 2003) are given
in table 4.1 for integration time of data points ranging from 1 to 15 days. The null-
hypothesis probability, as a function of bin size, is plotted in Figure 4.4 where it appears
that the correlation becomes stronger for smaller bin sizes due to increase in the number
of degrees of freedom. Hence, evidently, both from table 2 and Figure 4.4, the correlation
time scale is shorter than one day, which is not surprising because for a binary period of
4.8 hours the possible time scales which can come into play, viz. the viscous time scale
of the accretion disc, the variability time scale of the accretion disc corona, and the time
scale for variation of the jet emission must be smaller than a day.
Correlation between soft X-ray and radio emission has been noted in many black
hole binary sources (Gallo et al. 2002) and finding similar correlation in Cygnus X-3
(Choudhury et al. 2002) firmly puts Cygnus X-3 as a good candidate for harboring a
black hole. Choudhury et al. (2003) point out that correlation between the X-ray and
radio emission and the anti-correlation between the soft and hard X-rays are directly
related to a causal connection between the spectral shape in the X-rays and the radio
emission. By correcting the X-ray light curve for the binary variations, we have shown
that the correlation time scale is shorter than a day (Table 4.1 & Figure 4.4). Therefore
it is necessary to analyze the pointed continuous observation in the X-ray wide-band in
order to determine the dynamic time scale of soft and hard X-ray emission. This will
enable us to understand the detailed structure of the accretion disk, and is the focus of
Chapter 4. Cygnus X-3: temporal studies
Table 4.1: The Spearman Rank Correlation coefficient, the null-hypothesis probability for no correlation and the D-parameter
for the effect of the third parameter, among the radio (GBI), soft X-ray (ASM) and hard X-ray (BAT) emission of
Cygnus X-3, in the low-hard state with the observed flux averaged for various bin sizes.
Bin Size No. of Spearman Rank Corr. Coeff. Null-hypothesis Probability D Parameter
(days) points A:G A:B G:B A:G A:B G:B A:G A:B G:B
f
g
f ji ji ji j j j j j j j j j j j
g j j j j j j j j j j j j j j
f
g
h
h
1 638 0.70 -0.50 -0.45 .10 .28 .68
emkl emkl 'mkl cmq l eml eml cmq l mcn l emo l mco l cmq l cmo l cmq l cmp l 17.94 -7.31 -4.37
emkl emhkl 'mgkl cmgkl emgfkl emgfkl cmgfkl mcf kl emgfkl mcf kl cmgkl cmfkl cmfkl cmfkl
m h k l m g k l m l m f k l m f k l m f k l m f k l m f k l m f k l m fh k l m fh k l m f k l m f k l m fh k l
no no no
no o p r
j
2 356 0.73 -0.57 -0.51 .10 .15 .49 13.83 -6.38 -3.22
n p s t q r n
j
3 246 0.76 -0.64 -0.58 .10 .13 .12 11.15 -6.11 -3.08
g k
4 186 0.77 -0.68 -0.60 .36 .12 .27 9.62 -6.26 -2.17
h k gh k fh k g k g k g k g k f k g k f k f k
j
5 149 0.79 -0.69 -0.66 .28 .23 .79 8.73 -4.64 -3.11
j
6 125 0.82 -0.73 -0.68 .49 .41 .20 8.54 -4.75 -2.32
j
7 108 0.82 -0.77 -0.70 .50 .19 .25 7.18 -5.32 -2.06
j
8 94 0.83 -0.77 -0.72 .63 .62 .17 s p o p n t 6.80 -4.60 -2.25
9 84 0.81 -0.83 -0.74 .95 .16 .15 j j j j j j j 5.27 -6.05 -1.74
10 75 0.84 -0.82 -0.73 .26 .56 .99 6.17 -5.12 -1.19
11 69 0.81 -0.86 -0.75 .20 .19 .76 4.33 -6.26 -1.46
12 63 0.87 -0.86 -0.82 .88 .33 .19 5.09 -4.25 -2.26
13 58 0.85 -0.81 -0.74 .24 .74 .31
r o 5.54 -4.26 -1.13
14 54 0.89 -0.87 -0.82 .63 .70 .19 4.90 -4.33 -1.61
ASM:GBI(2.2GHz); ASM:BATSE; GBI(2.2GHz):BATSE
f
h
80
4.3. Power Density Spectrum (PDS) 81
Fig. 4.4: The null-hypothesis probability for not having a correlation between the soft X-ray
(ASM), radio (GBI) and hard X-ray (BATSE) emissions in Cygnus X-3 for various bin
sizes of integration. The first three points in the top panel denote the upper limits.
our future work. A simultaneous radio observation will be imperative in understanding

the detailed mechanism of the disk jet connection in the system, which will provide a
quantitative estimate of the extent of jet power being emitted in the X-ray band.
4.3 Power Density Spectrum (PDS)

X-ray binaries are the cosmic sources which display the smallest scale of temporal vari-
ability in the high energy regime. The fastest variability scale is of the order of mil-
liseconds (for a review see van der Klis 2000). These objects emit X-ray photons at a
u
very large rate (say,
x photons/sec), of which only a small fraction (say, Ý
rl x )
Q
is detectable by our instruments. Hence, event hough the photon production rate at the
2 s
source site may be considered a continuous function of time , the time J series of
s ¿
photon
J arrival times may be expressed in a discrete form
ð
7
[ n ¶ (with
!

! the total number of detected photons), which should be [ ideally [available for anal-
n ¶
ysis. However, the maximum telemetry rate of the instruments limit the registration and
transmission of all these arrival times. Therefore, the data are binned into equidistant
time bins, onboard the satellite observatories. The finally telemetered information con-
J J
sists of a sequence of numbers u P 7k

, where is the total number
[ [ [ ¶ ¶
of time bins, u represents the number of photons detected during the time bin P . Usu-

ally the time bins are equidistant and contiguous. The aim is to reconstruct as much as
2 s
possible about from the measured u .

For standard analysis, the rate of background photons is considered negligible, as it
is not supposed to vary strongly on the rapid time scales considered of interest, and these
variations are uncorrelated to the fluctuations of the source object. If the source doesn’t
s
vary intrinsically, the will be to a high degree of precision randomly distributed. ,
@ ð
so that the~ u : follow Poisson statistics appropriate to a rate ¬7 u wv x
, with a standard
deviation å 7
õ
y .

The standard approach (seeJ van der Klis 1997) is to divide J the time series into
u @
equal-length
J J time segments of time bins [ Fm7I [ [
each, (so that, ideally
7I ), in order to calculate the Fourier transform of each segment:
¶
5'z |{ l u @ = ;: } z @> J
@" | ð { ~ 7k
[ [
(
[
(4.4)
in order to convert this into a power spectrum for each segment

z 5 ( 5 z: J
¸
|
~ 7k
[ [
(
[
(4.5)
5 "|
{
@ l u @ J |
and then to average these power spectra. Here 7 , the number
n
of photons detected in the segment. With this power spectral normalization (Leahy et al.
@ z
1983),: if the u are distributed according to the Poisson distribution, then the follow
the z ¸
distribution ~with 2 degrees of freedom (constituting of ‘noise’ and ‘signal’), so
that v x
¸
79( and 79( . This white noise component with mean level 2 and stan-
dard deviation 2, induced in the power spectrum, due to Poisson fluctuations in the time
series, is called ‘Poisson level’, which can be considered as “background” in the power
spectrum, against which other power spectral features caused by the intrinsic variabil-
ity in the source are observed. In XRONOS (XANADU - LHEASOFT) terminology this
particular normalization of the power spectra is known as d/f normalization.
z 5'z
The physical dimension of the thus defined powers is the same as that of the time z
series: 7 @
7 u . The physical interpretation of the in terms of properties of
z z
¸ ¸
the source are inconvenient. Therefore, in recent times, a more evolved power spectral
c z z
|
normalization has become popular J where the powers are reported as , where
¸
D the “count rate”, i.e. 7
n
4 , where 4 is the duration of the segment. The
are dimensionless and can be interpreted as estimates of the power density near

8
z
the frequency
8
|

~
4 , where is a function of frequency whose integral gives the
8
fractional root-mean-square amplitude of the variability. This is defined as
{ @"{ | l u @
u { l u
"ì dì
| u | @
J
@ |
u (4.6)
[
following directly from Parseval’s theorem. The fractional rms amplitude : due to
: is given by
fluctuations in the given frequency range
8 [÷8
3
: 7Iâ 3
X

(4.7)
8 8
can be interpreted as the functionV whose integral gives the square of the frac-
8

tional rms amplitude of the: variability in the original time series. The physical unit
used for is (rms/mean) /Hz, where ‘rms’ and ‘mean’ both refere to the time series;
8
“rms/mean” is just the dimensionless quantity .
The averaging of the power spectrum is usually performed both by averaging the z
individual power spectra (from different segments) together (averaging the ’s with the
z ~ z ¸
same from the different segments), and by averaging powers at adjacent frequencies

¸ * ) 0

estimates.
¸ [
¼ . The goal is to decrease the standard deviation of the power
· The calculation of the power spectra from segments provide the study of the
evolution of the power spectrum with time. Thence,: one may fit the various functional
shapes to the power spectra using the
8

minimization. To decipher any critical
component in the low frequency end, where large number of individual powers may not
be available for the averaging, other fitting procedures may be applied (Papadakis and
Lawrence 1993), although such methods are not generally applied. A number of non-
Fourier methods are also available for the temporal analysis (Deeter 1984, Deeter and
Boynton 1982).
The power spectral components seen in the X-ray binaries are usually broad (except
for the sharp peak component for the pulsars). The very wide power distribution over
typically several decades in frequency are called ‘noise’, while the localized broad peaks
are called ‘quasi-periodic oscillations’ (see, for eg. van der Klis 1989). Further, the noise
which has a high frequency cut-off or steepening towards high frequency are known as
band limited noise, and may be classified into various classes depending on their cut-off
frequency (van der Klis 1994). These power spectral components are assumed to be
underlined variations which are stochastic in nature. Currently very fervent research is
being carried all over to create a physical model to explain these PDS features.
Fig. 4.5: Left panel: The power density spectrum of observation on MJD 50949–50950 (low
and hard state) showing the monotonic power law behaviour of the red noise, with the
spectrum merging with the back ground white noise at 0.1 Hz. Right panel: The power
density spectrum of observation on MJD 50500–50501 showing the monotonic power
law behaviour of the red noise, with the spectrum merging with the back ground white
noise at 0.1 Hz.
4.3.1 Power Density Spectrum (PDS) of Cygnus X-3

From a historical perspective, the most commonplace study of the temporal properties
of the source has concentrated on the binary modulation and its evolution. A detailed
study of its variability characteristics as a function of frequency and comparison with
other black hole candidate sources has not been attempted so far. van der Klis and
Jansen (1985) report the detection of very low frequency regular oscillations during the
rising phase of the binary and Rao et al. (1991) report regular oscillations in hard X-
rays. Here we present the PDS for both (low)-hard and (high)-soft states of Cygnus X-3,
after correcting for the binary variability (although the binary correction doesn’t have
any significant consequence of the power density spectrum, except maybe in the low
frequency regime).
For the (low)-hard state we consider the set of six pointed observations (RXTE–PCA)
between MJD 50949–50954, one pointing per day (Choudhury and Rao 2002), with
the spectral pivoting correlated to the radio emission (Choudhury et al. 2002). Each
individual pointing lasts for only about half a binary orbit ( *
x s) including a break
of about 2000 s due to passage through SAA (see Figure 4.3), and hence there is not
Fig. 4.6: Power Density Spectrum (PDS) of Cygnus X-3 for the RXTE–PCA pointed observa-
tions shown in Figure 4.3, after correcting for the binary variations. The PDS is gener-
ated separately for three regions of the binary phase namely the fall (phase 0.75 - 1.0 -
top panel), rise (phase 0 - 0.4 - middle panel) and the peak (phase 0.4 - 0.75 - bottom
panel).
adequate coverage for generating a PDS. Therefore we provide the PDS of the combined
lightcurve of the six days pointing (Obs. Ids.: 30082-04-01 – 30082-04-06) in Figure
4.5, left panel. The PDS has a flat spectra in the low frequency region, ) 3
Àl Ø ,
Q
4
followed by a power law dependence till w
Hz, but then the power merges with the
white noise without any exponential cutoff. This feature of lack of significant power
above 0.1–1 Hz is the most interesting aspect of the PDS of Cygnus X-3, and this
feature is consistent with all the obtained PDS of the source from RXTE observations.
As a contrast, in Figure 4.5, right panel, we present the PDS of four observations (Obs.
Ids.: 20099-01-01-010,20099-01-01-01,20099-01-01-020,20099-01-01-02), spanning a
total period little more than one day, combined together, with the source in the high,
and correspondingly soft, state. The observation was carried out on MJD 50500–50501.
The low frequency region shows more power as the amplitude of the binary modulation
during this state is more pronounced in this state, and despite our correction for this, the
modulation manifests itself in the PDS. The higher frequency region shows behaviour
similar to that of the low state with the power hitting the deck of background white noise
level at 0.1 Hz.
To highlight the effect of binary correction on the PDS of this source, we again
consider the longest continuous stretch of pointed observation of this source shown in
Figure 4.3 (Observ. Id. 10126-01-01-020 & 10126-01-01-02, MJD 50322). Figure 4.6
shows the PDS after the binary correction. To explicitly demonstrate that the shape of
residual variability is independent of binary phase, we show in Figure 4.6 the PDS’s
separately for the falling part of the light curve (phase 0.75–1.0 - top panel), rising part
of the light curve (phase 0–0.4 - middle panel), and the peak of the light curve (phase 0.4
–0.75 - bottom panel). Apart from the larger variability due to flares in the rising part
Q
(manifest as increased power below 10 l Hz), the shape of the PDS are similar for all the
Q
three cases. The power-law index in the frequency range 10 l Hz – 0.1 Hz is consistent
with -1.5 and the total rms power is a few per cent. One of the remarkable feature of
the light curve is the negligible power above 0.1 Hz, and this feature is observable in the
PDS obtained from all the lightcurves, spanning both hard and soft states (Figure 4.7). In
fact, as shown in Figure 4.7, the PDS show the same characteristic featureless powerlaw
behaviour in all the states.
: In all the PDS considered above, the spectra is represented in
the units of (rms/mean) /Hz (eqs. 4.6,4.7). This is achieved by running the powspec
program of the XRONOS package (LHEASOFT v5.2) with the normalization parameter
= 2 (or -2).
PDS at low frequencies. Reig et al. (2002) have examined the aperiodic variability
of two micro-quasars Cyg X-1 and GRS 1915+105 at very low frequencies using the
Fig. 4.7: The PDS of Cygnus X-3 over a wide frequency range obtained using RXTE–ASM and
RXTE–PCA at three observation periods.

RXTE–ASM dwell data. At frequencies below 10 l Hz it was found that the PDS is

consistent with an index of -1 and the rms power below 10 l Hz is 21 – 27 %, for these
two sources (Figure 4.10). Since we have argued that Cygnus X-3 is an archetypical
black hole binary (see chapter 3) it is instructive to compare the PDS of Cygnus X-3 at
very low frequencies.
We have derived the PDS of Cyg X-3 at very low frequencies using RXTE–ASM data
using a method similar to that followed by Reig et al. (2002). The results are given in
P D
Figure 4.7, along with the PDS obtained using the pointed observations of RXTE–PCA.
It is found that below 5 10 l Hz, the power-law index is flatter (-0.97 0.17), con-
sistent with that obtained for Cygnus X-1 and GRS 1915+105. The remarkable feature
Fig. 4.8: The three Power Density Spectra of Cygnus X-3 (reported in the previous section) with
the normalization defined by eqs. 4.4 & 4.5, the d/f normalization of XRONOS package,
such that the power of the background white noise level corresponds to the value of 2,
provided the distribution obeys Poisson statistics.
of Cygnus X-3, however, is the low power - about 3.5% - in this frequency range. The
flatter power law at low frequencies ( ) 10 l Hz) and the steeper power-law at higher
Q
frequencies ( - 10 l Hz) intercept at 10 l x Hz signifying a break in the power spec-
trum. Though there is a hump like feature at this frequency, we cannot rule out the
influence of the binary period ( 10 l x Hz) on the shape of this hump. A long uninter-
rupted observation would be required to address this issue.
The distinct feature of the power density spectra of the source is the shifting of the
spectra towards low frequency regime (vis-a-vis the ‘normal’ frequency regime of other
Galactic X-ray binaries), corresponding to very massive black hole systems (Hayashida
et al. 1998, Czerny et al. 2001). To ascertain this fact we obtain the PDS using the nor-
malization described by eqs. 4.4 and 4.5. This is achieved by running the powspec
program of the XRONOS package (FTOOLS v5.2) with normalization parameter = 1
(or -1), the d/f normalization, such that the numerical value of the white noise is 2, pro-
vided the distribution obeys Poisson statistics. In Figure 4.8 we plot the three concerned
PDS in the frequency range 1 – 50 Hz. It is visually evident that power is not discernible
from the background white noise. Fitting a constant (dotted lines in Figure 4.8) for the
three PDS we obtain the value of the mean power in this frequency range to be 1.99, 2.06
& 1.99 for MJD 50322, MJD 50949-50954 & MJD 50500-50501, respectively. These
PDS are obtained without any correction for the dead time error. The PDS for frequency
9
l x Hz, i.e. the features preceding the power law decay, are not obtainable due to
observational constraints of the RXTE–PCA, with the satellite having an orbital period
of 96 minutes and it passes through the South Atlantic Anomaly in every orbit. Berger
and van der Klis (1994) provide the PDS of this source from the EXOSATobservations,
where the frequency dependence of the power in the region 0.05 – 1000 Hz is given.
The overlap region of their PDS with the PDS obtained by us (from RXTE–PCA ob-
servations) are in agreement, and hence we reiterate that this behaviour is an inherent
property of the source and not due to any instrumental or analysis procedural artifact.
Sunyaev and Revnivtsev (2000) have compiled the power density spectra (multiplied
by frequency) of the most common Galactic X-ray binaries, both neutron stars and black
hole candidates (in their low/hard states) (Figure 4.9). The black hole binaries show,
typically, a power law dependence with a positive index in the region of 0.01 – 1
Hz, flat spectra for the next decade of frequency range, followed by a power law decay
(i.e. negative index) of power in the 10 –100 Hz. The PDS of the neutron stars is
generally shifted towards the higher frequency region by an order of magnitude. From
the PDS of Cygnus X-3 (Figure 4.6) it is quite apparent that, from the RXTE–PCA
pointing observations, we are catching the power law decay portion of the PDS, shifted
by a few decades in the lower frequency regime, with the power getting merged with
the background white noise at 0.1 Hz. Due to improved continuous observational
capabilities of RXTE it is possible to extend the PDS to the lower frequency regime
( Í
l x Hz) and hence ascertain from the wide frequency band PDS that, in general,
the power above 0.1 Hz is not discernible from the white noise. One may reconcile the
absence of power in the high frequency regime to the scattering of the X-ray photons in
the wind from the companion Wolf-Rayet star, reducing the amplitude of the fast X-ray
variability (Berger and van der Klis 1994). If we reconcile the absence of power above
0.1 Hz to the reprocessing of the X-ray emission in the circumstellar environment, then
this feature does provide an interesting sidelight to the paradigm which states that the
variability time-scale scales linearly with the mass of the compact object (Hayashida et
Fig. 4.9: Broad band power spectra of X-ray binaries in the low spectral state. This figure is
obtained from Sunyaev and Revnivtsev (2000).
4.4. Time lag between soft and hard X-rays 91
Fig. 4.10: Comparison of the ASM PDS of Cygnus X-1 & GRS 1915+105 with that observed
at higher frequencies. The dashed lines represent the best fit model to the PDS of
Cygnus X-1 at higher frequencies when the source is in the high-soft luminosity state.
The dotted line represents the averaged PDS of Cygnus X-1 in the low-hard state. The
figure is obtained from Reig et al. (2002).
al. 1998, Czerny et al. 2001), and introduce an additional factor, viz. reprocessing in the
dense circumstellar material, where it exists, into the picture. It is also interesting that the
PDS at very low frequencies have about an order of magnitude lower power compared
to other black hole candidate sources Cygnus X-1 and GRS 1915+105 (Figure 4.10).
4.4 Time lag between soft and hard X-rays

In section 4.2 it has been established that the time scale of (anti-)correlation between the
soft and hard X-rays, during the ‘low’ as well as ‘hard’ state, is less than a day. Given the
sparse sampling of the RXTE–ASM as well as the CGRO- BATSE during the day, it is
not feasible to ascertain this time scale more accurately than a day (or more). Therefore
a long pointed observation by the pointing instruments viz. RXTE–PCA and HEXTE
is needed for the observation of such a delay. Most of the observations of Cygnus X-3
by RXTE are TOO observations during the high, state, triggered by the radio behaviour.
Furthermore, the observations during the recent times are of very short durations, at
times as low as 1000 seconds.
Fig. 4.11: Cross correlation between soft and hard X-rays (2 - 7 keV & 20 - 50 keV, respec-
tively). Top panel: Observation carried out on MJD 50321. The hard X-ray flux is
obtained from RXTE–HEXTE. Middle panel: Observation carried out on MJD 50322.
The hard X-ray flux is obtained from RXTE–HEXTE. Bottom panel: Observation car-
ried out on MJD 50949 - 50954. The hard X-ray flux is obtained from RXTE–PCA.
4.4. Time lag between soft and hard X-rays 93
Here we report the results of the search for the (anti-)correlated delay between the
soft and hard X-rays, for three observations, carried out on the MJD 50321 (obs. id.
10126-01-01-010 & 10126-01-01-01), MJD 50322 (obs. id. 10126-01-01-020 & 10126-
01-01-02) and MJD 50949 – 50954 (obs. is. 30082-04-01-00 to 30082-04-06-00). The
duration of the observations on the six days MJD 50949 – 50954 consist of about a
satellite orbit each, hence the analysis of the combined lightcurve is presented here. The
cross-correlation between the fluxes in the 2 - 7 keV and 20 - 60 keV are shown in
Figure 4.11. These cross-correlations are obtained after correcting the lightcurves for
the binary modulation (section 4.1). The soft X-ray lightcurves are obtained from the
RXTE–PCA (pointed) observations while the hard X-ray lightcurves are obtained from
RXTE–HEXTE for MJD 50321 & 50322, and from RXTE– PCA for the days MJD 50949
– 50954. On MJD 50322 an anti-correlated delay of
s is observed, whereas
from the observations on the other two days a delay of * s are seen, vindicating
the results of the SRC test (chapter 3 and section 4.2). Attempts to locate such a direct
measurement of (anti-)correlated delay in the ‘high’ as well as ‘soft’ state has not yet
yielded any positive detection.
Observation of such a direct delay, (anti-)correlated, between the soft and hard X-
ray is measured for the first time for this source in particular, and Galactic microquasars
in general. A physical interpretation of this extremely important observational result is
currently beyond the scope of this research, but this feature can be the cornerstone of the
building block of a complete theory providing the explanation of the physical mechanism
and the geometrical structure of this source, which will further provide the necessary
direction to build a general understanding of these Galactic microquasars in general.
Current trends indicate that this delay may be the manifestation of the dynamical viscous
time scale in the accreting flow of the disc.
Chapter 5
Disc-jet connection in
microquasars: low (hard) states
The characterizing feature of the Galactic microquasars is the radio emission, along
with the presence of the accretion of mass onto a compact object analogous to the Active
Galactic Nuclei (AGNs). In the introductory section of this thesis (chapter 1), the various
theoretical paradigms which attempt to explain the accretion phenomenon were stressed
upon. Further, the (quasi) simultaneous observations of X-ray binaries in the radio and
X-ray bands has led to the notion that the presence of radio jets is ubiquitous in sources
with black holes or low magnetic field ( µ
G) neutron stars as compact objects
(see Fender 2001b, Fender and Kuulkers 2001). Though superluminally moving radio
jets are detected in several microquasars (Mirabel and Rodriguez 1994, Tingay et al.
1995, Hjellming and Rupen 1995, Hjellming and Johnston 1988), which are invariably
associated with huge radio flares, only recently it has been realized that non-thermal
radio emission is a common feature during relatively quiet phases. Fender (2001b,a)
made a detailed calculation of the energetics during such quiet phases and argues that
the non-thermal emission forms a substantial part (5%—50%) of the energy budget.
Compact radio jets are indeed observed in the low-hard state of several microquasars viz.
Cygnus X-1 (Stirling et al. 2001), GRS 1915+105 (Dhawan et al. 2000), 1E 1740.7-2942
(Mirabel et al. 1992), and GRS 1758-258 (Rodriguez et al. 1992). Further, the spectral
analysis of the radio emission from X-ray transient blackhole candidates, GS 2023+38,
GRO J0422+32 and GS 1354-64 (Fender 2001b) and two persistent X-ray blackhole
candidates GX 339-4 (Corbel et al. 2000) and XTE J1550-564 (Corbel et al. 2001) are
interpreted to originate from synchrotron emitting, compact conical jets. With a plethora
of uncoordinated high energy and radio observations giving rise to a very confused and
foggy picture of the disc - jet connection across the various microquasar (X-ray binary)
94
95
systems, a concise systematic study of the long term radio:X-ray correlation, in the lines
of the analysis carried out for Cygnus X-3 in chapter 3 was the essential need of the
hour. Our effort was to find a possible generalization of the pivoting behaviour the X-
ray emission, analogous to that of Cygnus X-3, correlated to the radio emission, during
the (low) hard state (see chapter 3 for the detailed analysis). While the generalization of
the X-ray spectral evolution during the radio flaring states is intended to be the guiding
direction of future research work in this field.
Fender and Kuulkers (2001) have compiled an extensive list of X-ray binaries, both
neutron stars and black hole candidates, for which simultaneous X-ray and radio obser-
vations have been made. The Green Bank Interferometer, West Virginia, operated by
NRAO, provides data for a number of X-ray sources that were monitored during its sev-
eral years of operation. Collating these sources with those monitored by RXTE– ASM
and CGRO–BATSE we found two X-ray binaries (black hole candidates) viz. Cygnus
X-1 and GRS 1915+105, along with Cygnus X-3, which are persistent in radio, soft and
hard X-ray bands and for which (quasi) simultaneous data from the three observatories
are available. In this chapter, we report the results of our analysis for the two persistent
X-ray binaries, microquasars and black hole candidates, Cygnus X-1 and GRS 1915+105
for which quasi-simultaneous radio and X-ray data are available from GBI (2.2 GHz),
RXTE–ASM (2 – 12 keV), and CGRO–BATSE (20 – 100 keV). We select data during
periods when there are no radio flares and the source is bright both in radio and hard
X-rays. We examine radio X-ray correlations in such hard states and show that these
correlations are similar to those found in the low-hard states of well studied sources like
GX 339-4. We complement our analysis of this state by giving a qualitative self con-
sistent picture of the steady X-ray non-flaring (radio) states of these sources, along with
Cygnus X-3 and GX 339-4. In the next chapter (chapter 6) we will attempt to provide
a phenomenological picture of the accretion - ejection connection in these Galactic mi-
croquasars, using the Two Component Advective Flow (TCAF) model of Chakrabarti
(1996a) transition among the X-ray as well radio emission.
A brief introduction to the these sources are as follows (for Cygnus X-3 refer to
chapter 3):-
$ GRS 1915+105. The X-ray binary GRS 1915+105 was first detected in 1992
(Castro-Tirado et al. 1992), and since then has been observed in the X-ray, radio
and infra-red bands (see Belloni 2002, for a review). From the K-band monitoring
of the source (Greiner et al. 2001a) it has been classified as a low mass X-ray
binary system with an orbital period of 33 days with a 10 black hole as the
compact object (Greiner et al. 2001b). Muno et al. (2001) classify the radio emis-
sion into three classes, radio faint, radio steep and radio plateau. This source is
96 Chapter 5. Disc-jet connection in microquasars: low (hard) states
extremely variable in nature and has been classified into several variability classes

(Belloni et al. 2000), of which the class is the closest analogue to the canonical
low-hard states of Galactic black hole sources (Rao et al. 2000).
$ Cygnus X-1. This source, the first to be X-ray detected in the Cygnus region
(Bowyer et al. 1965), is the archetypical black hole candidate (Herrero et al. 1995),
whose optical counterpart, O9.7Iab super giant HDE 226868, was among the ear-
liest to be identified for an X-ray binary (Bolton 1972, Webster and Murdin 1972).
A persistent source in X-ray, radio (Braes and Miley 1971) and optical, it shows
a binary modulation with a period of 5.6 d in all the bands (Pooley et al. 1999,
Brocksopp et al. 1999). The radio emission is weak, generally around 15 mJy,
varying between 10 to 25 mJy in the low-hard X-ray state getting considerably
weaker in the high-soft state of X-ray emission (Brocksopp et al. 1999).
$ GX 339-4. This X-ray binary system was discovered in 1971 (Markert et al.
1973b,a) whose radio counterpart was first reported by Sood and Campbell-Wilson
(1994). It is considered to be a blackhole candidate (Samimi et al. 1979) because
of the similarity of the X-ray spectral characteristics and short term variabilities
with those of Cygnus X-1 (Tanaka and Lewin 1995), especially in the low-hard
state (Corongiu et al. 2003).
$ V404 Cygni. This is an X-ray transient whose luminosity can increase by a factor
of
during outburst. Simultaneous observations, during the decay phase
following an outburst, in the radio (Han and Hjellming 1992) and X-ray (Kitamoto
et al. 1990) revealed that the source never entered the soft state during this phase
from this source to be k ,(

+D
(Zycki et al. 1999). In the quiescent phase the ROSAT–PSPC measured the flux

;
qs
È l (Wagner et al. 1994, for recent
observations see Kong et al. 2002).
The sources above (plus Cygnus X-3) dealt with in this chapter, with apparently
diverging behavioural patterns, span the whole gamut of multifarious characteristic fea-
tures of the Galactic microquasar systems. Hence, any generalized distinguishing at-
tribute of the disc-jet connection in these systems may be taken as a normal trait for this
class of cosmic sources. In the following sections we provide the existing radio - X-ray
association in these objects, prior to our study, followed by our statistical analysis of
the monitoring data and the X-ray spectral evolution, all in the low-hard states of these
systems. For the sake of completeness of the analysis and the inference, we present the
main results of Cygnus X-3 in a precis form (for details refer to chapter 3). The existing
features of the radio:X-ray correlation of the prominent persisting sources in the (low)
hard states, prior to our work are as follows:-
5.1. Radio:X-ray correlation of the persistent sources: hard states 97
$ Cygnus X-3. Radio (GBI, 2.2GHz, monitoring data) and hard X-ray (CGRO–
BATSE, 20 – 100 keV, monitoring data) show anti-correlation during the hard
state, no correlation during the minor flaring state and correlation during the flar-
ing state (McCollough et al. 1999). Normally the radio flares occur during the
high state (Watanabe et al. 1994).
$ GRS 1915+105. This source exhibits quite complicated behaviour. It is highly
variable in nature, with the radio emission always present in the hard steady X-
ray state (Muno et al. 2001). Very generic picture gives no correlation between the
radio emission and both soft X-ray emission and wide-band (1 - 200 keV) flux, but
anti-correlation between hard X-ray (20 - 200 keV) flux and the radio emission,
in the hard state (Rau and Greiner 2003). During high radio emission the X-ray
spectrum is softer.
$ Cygnus X-1. There exists mild correlation between radio (GBI, 2.2 GHz) and soft
X-ray (RXTE–ASM, 2-12 keV) and very poor correlation between radio and hard
X-ray (CGRO–BATSE, 40 – 140 keV) (Brocksopp et al. 1999) in this object. The
radio emission is suppressed during the high-soft state.
$ GX 339-4. In this Galactic black hole binary system the radio, soft X-ray as well
as hard X-ray are very strongly correlated in the low -hard state (Corbel et al.
2003). Whereas, the radio emission is suppressed (quenched) during the high -
soft state (Corbel et al. 2000).
$ V404 Cygni. In this source, strong correlation between radio and soft X-ray emis-
sion (RXTE–ASM, 2 – 12 keV) is seen during the low-hard states, from various
quasi-simultaneous observations.It shows similar radio:X-ray emission character-
istics as GX 339-4 (Gallo et al. 2003).
5.1 Radio:X-ray correlation of the persistent sources: hard

states
We repeat the statistical procedures employed to perform the Spearman Partial Rank
Correlation test for Cygnus X-3, in order to obtain the Spearman Rank Correlation
(SRC) coefficient (Macklin 1982) of the radio (GBI), soft and hard X-rays (RXTE–ASM
& CGRO–BATSE) emissions in GRS 1915+105 and Cygnus X-1, during the period
when these three observatories were simultaneously operational in regularly monitoring
these sources. Here we report the results of the correlation analysis for data averaged
over different time intervals viz., 1, 5 and 10 days as well as correlation between X-ray
hardness ratio (ratio of observed BATSE count rate to that of ASM count rate) and radio
(ASM count rate is taken as the third parameter) for the duration in which the respective
sources were in a relatively long term steady quiescent hard state. Table 5.1 gives the
complete result of the Spearman Rank Correlation (SRC) test for the correlation among
1) the radio, soft X-ray and hard X-ray, and 2) the hardness ratio in X-ray and radio
(GBI) for GRS 1915+105 and Cygnus X-1 along with Cygnus X-3, during the steady
quiescent hard state of X-ray emission, for fluxes in different bands averaged over 1, 5
and 10 days.
5.1.1 Cygnus X-3

A detailed description of the SRC test and the results are given in chapter 3 along with
the X-ray spectral fitting correlated to the source, where a pivoting behaviour of the X-
ray spectra is inferred. Here, in table 5.1 the result of the Spearman Rank Correlation
test for fluxes averaged over 1, 5 and 10 days, during the radio quiescent state (Figure1
Choudhury et al. 2002, see Figure 3.4 chapter 3, this thesis), also corresponding to X-
ray quiescent and hard state.Though the value of the SRC coefficient increases with
increasing bin time, it should be noted that the number of degrees of freedom decreases
and hence the best correlation result is obtained for one day averaging (as seen by the
value of the null hypothesis probability). Hence the correlation time scale is shorter than
a day. The X-ray hardness ratio shows a very strong and significant anti-correlation with
the radio emission. Interestingly this anti-correlation is stronger and more significant
than the anti-correlation of the hard X-ray flux with radio (or soft X-ray). This suggests
spectral bending being correlated with the radio emission, amply demonstrated in the top
panel of Figure 5.2 (adapted from Figure 3 Choudhury et al. 2002, Figure 3.6, chapter
3 this thesis). The pivoting of the spectrum around 12 keV, as obtained from pointed
RXTE observations, explains the anti-correlation between the soft (2 – 10 keV, RXTE–
ASM) and hard X-ray (20 – 100 keV CGRO–BATSE) emission.
Table 5.2 gives the results of the spectral fitting, along with the soft X-ray (ASM),
hard X-ray (BATSE) and radio (2.2 & 8.3 GHz, GBI) flux, of two observations corre-
sponding to the extreme behavior of the source within the precincts of the hard state.
In chapter 3, more spectral data were presented which showed a systematic change in
the wide band spectrum correlated with the radio flux. The available spectra showed
a pivoting behavior around 12 – 15 keV. This fact, coupled with the strong correlation
between the soft X-ray and radio as well as that anti-correlation of the hard X-ray flux
with radio (and soft X-ray flux) strongly suggest that the spectral shape governed by a
pivoting behavior at around 12 keV is responsible for the observed correlations.
Fig. 5.1: The combined simultaneous light curve of GRS 1915+105 in the soft X-ray (2 – 12
keV, ASM, top panel), hard X-ray (20 – 100 keV, BATSE, middle panel) and the radio
(2.2 GHz, GBI, bottom panel). The low-hard states, selected for the present analysis
(see text), are separated by vertical dashed lines and identified with numbers.
5.1.2 GRS 1915+105

Figure 5.1 gives the daily averaged light curves of GRS 1915+105 in the soft X-ray
(ASM, top panel), hard X-ray (BATSE, middle panel) and the radio (2.2 GHz, GBI,
bottom panel) during the period when all the three instruments were simultaneously

monitoring the sources. As per the classification of Belloni et al. (2000), the class is
the closest analogue to the canonical low-hard states of Galactic black hole sources (Rao

et al. 2000). Belloni et al. (2000) identify three stretches of long duration classes and
two of them have simultaneous BATSE and GBI observations. These two periods are
demarcated by numbers in the top panel of Figure 5.1, and are used for the correlation
analysis. The results of the SRC test, given in Table 5.1 are similar to that of Cygnus
X-3. The radio and soft X-ray fluxes are well correlated. The anti-correlation of the hard
X-ray flux with both radio and soft X-ray flux is not as strong as in the case of Cygnus
X-3. The correlation time scale, as can be concluded from the strength of the correlation,
is one day or less. The correlation between the X-ray hardness ratio and radio flux again
gives results similar to those of Cygnus X-3, suggesting a spectral pivoting correlated to
the radio emission. Hence, it is evident that the radio X-ray correlation behavior in the
steady long term hard state of this source is similar to that of Cygnus X-3.
Chapter 5. Disc-jet connection in microquasars: low (hard) states
Table 5.1: The Spearman Rank Correlation (SRC) coefficient, null-hypothesis probability and D-parameter among 1) the radio,
soft X-ray and hard X-ray fluxes and 2) the hardness ratio of X-ray (ratio of BATSE to ASM flux), the radio and the
soft X-ray fluxes, for Cygnus X-3, GRS 1915+105 and Cygnus X-1 in the low-hard state with the observed fluxes
averaged for 1, 5 and 10 days.
Cygnus X-3 GRS1915+105 Cygnus X-1

SRC Null Prob. D-Par. SRC Null Prob. D-Par. SRC Null Prob. D-Par.
1 day avg. flux no. of data points: 532 no. of data points: 108 no. of data points: 268
ASM:GBI 0.68 0 15.9 0.61 6.8 0.29 1.4
6

GBI:BATSE -0.43 -3.7 -0.27 0.004 -1.8 0.33 0.16 1.4

ASM:BATSE -0.48 -6.7 -0.23 0.018 -0.8 0.70 13.1

RATIO:GBI -0.65 0 -3.9 -0.62 -4.1 0.01 0.93 2.5

5 day avg. flux
ASM:GBI
no. of data points: 149
0.76 8.1 0.71

4.4
0.53 2.0

¡

GBI:BATSE -0.61 -2.5 -0.33 0.067 -1.3 0.53 2.0

ASM:BATSE -0.68 -5.3 -0.23 0.22 -0.1 0.72 5.6

RATIO:GBI -0.71 -2.0 -0.65 -1.5 0.03 0.84 1.6

10 days avg. flux no. of data points: 75 no. of data points: 19 no. of data points: 33
ASM:GBI 0.83 6.2 0.67 0.002 3.1 0.58 1.4

GBI:BATSE -0.72 -1.5 -0.23 0.32 -0.8 0.61 2.0

ASM:BATSE -0.79 -3.8 -0.13 0.60 0.2 0.72 3.4

£
RATIO:GBI -0.85 -2.1 -0.58 0.009 -0.5 0.11 0.55 1.4

100
Fig. 5.2: Top Panel: The unfolded spectra of Cygnus X-3 in the low-hard state (radio quiescent
period) on two occasions (1: MJD 50954; 2: MJD 50717). Bottom Panel: The spectra
of GRS 1915+105 in low-hard state on two occasions (1: MJD 50421; 2: MJD 50737).
Two unfolded spectra for GRS 1915+105 corresponding to GBI 8.3 GHz fluxes of 17
and 77 mJy respectively are overlaid in the bottom panel of Figure 5.2. Table 5.2 gives
the details of the soft X-ray (ASM), hard X-ray (BATSE) and radio (2.2 & 8.3 GHz, GBI)
flux along with the best fit parameters of the spectral fitting (Vadawale et al. 2001b). It
is interesting to note that the wide-band spectra at extreme radio emissions shows a
cross-over at higher energies (20 keV) compared to Cygnus X-3 and, by association,
we suggest that a spectral pivoting occurring at higher energies is responsible for the
observed correlations. A possible reason for a weaker anti-correlation between hard X-
ray (20 – 100 keV) flux and soft X-ray (2 – 12 keV) flux (and radio) compared to that
in Cygnus X-3 is that in the case of Cygnus X-3 the soft X-ray (ASM) and hard X-ray
(BATSE) energy ranges are, correspondingly, below and and above the pivot energy of
around 12 keV, whereas for GRS 1915+105, the pivot energy is at higher energies of
around 20 keV, and the spectrum is relatively harder.

Rau and Greiner (2003) have made a detailed study of all the state observations of
GRS 1915+105 based on the analysis of four years of pointed RXTE–PCA and HEXTE
observations. They find no correlation between radio and soft X-ray flux (1 – 20 keV),
however, they find an anti-correlation between radio and hard X-ray (20-100 keV) flux,
Table 5.2: The observed fluxes and X-ray spectral parameters of Cygnus X-3 and GRS
1915+105 during two pointed RXTE observations, corresponding to the extreme be-
haviour of the sources within the precincts of the respective low-hard states.
Cygnus X-3
MJD 50717 50954
Flux
ASM (cts s e ) X 11.11 5.37
BATSE (ph cm e ¯e )
GBI-2.2 GHzV (mJy)
[ 0.038
115
0.058
43
GBI-8.3GHz (mJy) V 165 53
Best fit parameters CompST+powerlaw
V ÐXª
kT ü (keV) ¤ ¥
¦
5.09 ¨!© ¤
¦
4.87 ¨© ¨
© ¤ §9§
2.55 ¨© ¤
¨
2.01 ¨© ¨
(d.o.f.) 0.74(86) 1.42(108)
GRS 1915+105
MJD 50421 50737
Flux
ASM (cts s e ) X 38.59V 34.96
BATSE (ph cm e ¯e )
GBI-2.2 GHzV (mJy)
[ 0.140
29
0.068
42
GBI-8.3 GHz (mJy) V 17 77 X
Best fit parameters V
« e'¬¬ «««« ¬®¬®W¯
diskBB+CompST(+powerlaw )
ce¬ ¬ «« ®« WW±
kT ü (keV) V
V ª A°
20.76 c X 4.89 c
¤ e'¬¬ ¬¬®²³
é 1.28 c e Ë 1.99 c e
'¬
kT (keV)
V VV
© ÐX
(d.o.f.)
– e
1.14(121)
2.49
1.55(121)
e ¨© ¨
©
X There is no observation on MJD 50421 for GRS 1915+105, we are quoting values for MJD 50422
The powerlaw component is present for MJD 50737, the radio loud state
V
and also find that the slope of hard power law spectrum correlates positively with the
radio flux in the low hard state of the source with observations during high radio emission
showing a softer spectrum. The latter results agree with our findings presented here. Rau
and Greiner (2003) also report spectral pivoting occurring between 20 – 30 keV in X-
ray spectra in hard state of GRS 1915+105. The results of Rau and Greiner (2003) are
consistent with our findings, except that the radio and soft X-ray fluxes are not correlated
in their data, whereas we find a good correlation between radio and soft X-ray flux. Our
results are based on the ASM data which is not very sensitive to X-rays above 10 keV,
whereas Rau and Greiner (2003) use the flux up to 20 keV using a model fit to the joint
PCA and HEXTE observations. The lack of correlation between radio:soft X-ray could
be due to the spectral pivoting around 20 keV, because of which the soft X-ray flux
from RXTE PCA data will comprise of both correlated and anti-correlated fluxes thus
weakening or averaging out the correlation.
5.1.3 Cygnus X-1
Figure 5.3 gives the daily averaged light curves of Cygnus X-1 in the soft X-ray (ASM,
top panel), hard X-ray (BATSE, middle panel) and the radio (2.2 GHz, GBI, bottom
panel) during the period when all the three instruments were simultaneously monitoring
the sources. The radio emission is weak, generally around 15 mJy, varying between 10
to 25 mJy in the low-hard X-ray state getting considerably weaker in the high-soft state
of X-ray emission (Brocksopp et al. 1999). In Figure 5.3, the region 1 as demarcated in
the top panel of the figure denotes the period of the long term low-hard state of the X-
ray emission. Since our emphasis is on the study of the long-term correlated radio:X-ray
behavior of this source in low-hard state, we consider only this period for the SRC test.
The results of the SRC test for Cygnus X-1 are given in Table 5.1. The table shows that
pattern of the correlation between X-ray and radio emission for Cygnus X-1 seems to be
different than for Cygnus X-3 and GRS 1915+105. The soft X-ray and radio fluxes are
not as well correlated as for the other two sources, specially for 1 day averages, being
only 0.29 compared to 0.68 and 0.56 for Cygnus X-3 and GRS 1915+105 respectively.
Nevertheless, the correlation is significant at the level of about one part in 10 . The
anti-correlation of hard X-ray (40 – 140 keV, BATSE) flux with the soft X-ray (2 – 10
keV) as well as radio (2.2 GHz, GBI) fluxes found for Cygnus X-3 and GRS 1915+105
is not present in Cygnus X-1. Instead, the SRC test shows that the hard X-ray positively
correlates with the both soft X-ray and radio emission. Also the X-ray hardness ratio
does not show any correlation with radio flux in the case of Cygnus X-1, unlike for
Cygnus X-3 and GRS 1915+105.
Fig. 5.3: The combined simultaneous light curve of Cygnus X-1 in the soft X-ray (2 – 12 keV,
ASM, top panel), hard X-ray (40 – 140 keV, BATSE, middle panel) and the radio (2.2
GHz, GBI, bottom panel). The low-hard states, selected for the present analysis (see
text), are separated by vertical dashed lines and identified with numbers.
Brocksopp et al. (1999) report a value of the SRC coefficient for soft X-ray:radio flux
correlation in low-hard state of 0.3 for 1 day average, after removing the mean orbital
light curve, which is close to 0.29 found by us for the same correlation, without removal
of the orbital modulation effects. Brocksopp et al. (1999) point out that loose correlation
of radio and soft X-ray fluxes (for their 1 day averages) may be partly due to the possible
offset between the radio and X-ray long period ( 142 days) light curves. They also give
the scatter plot of ASM, BATSE and radio fluxes.
The strength of the SRC between the radio and the both soft and hard X-ray are
similar (Table 5.1), showing moderately strong correlation, whereas the soft and hard
X-ray flux show a very strong positive correlation, the SRC coefficient being 0.70 for
1 day averages. Further, ASM and BATSE observations are at different times during
the day implying that intra-day variability is relatively weak compared to variability on
longer time scales as both fluxes are strongly correlated over longer time scales. Clearly,
the similarity between hard X-ray:radio and soft X-ray: radio flux correlation is because
of strong correlation between hard and soft X-ray fluxes. As mentioned earlier, the radio
emission in Cygnus X-1 is quite weak (around 15 mJy) varying between 10 and 25 mJy
in the low-hard state. As the GBI observations have an error of nearly 4 mJy, a detailed
wide-band spectral analysis for two extreme values of radio flux, for finding the relation
between the shape of the X-ray spectra and the radio emission and the pivoting behavior
and the pivot energy, as done for Cygnus X-3 and GRS 1915+105, is difficult to be
carried out for Cygnus X-1.
Zdziarski et al. (2002) have shown that the long term variability of the X-ray emis-
sion from this source in hard state comprises of two types of spectral variability, one cor-
responds to the change in the shape of the spectra (with spectral shape pivoting around
80 keV) with change in soft X-ray flux and the other corresponds to the change in total
flux, with the spectra simply moving up and down parallel to each other with a constant
shape, in the whole X-ray broad band. This may explain the lack of correlation between
the X-ray hardness ratio and the radio flux as well as the comparative weakness of the
strength of the SRC correlation between ASM and radio flux. Zdziarski et al. (2002)
have analyzed the various correlations among the fluxes of the three energy channels of
ASM (1.5 – 3, 3 – 5 and 5 – 12 keV) along with the 20 – 100 keV and 100 – 300 keV
flux of BATSE and the corresponding specific spectral index of these bands. They con-
clusively show that in the low-hard state of the X-ray emission, over long periods, the
change in the spectral shape occurs with a pivoting around 50 – 90 keV. This explains
that the BATSE flux, being dominated by the lower energy photons is very strongly cor-
related to the ASM flux. The lack of anti-correlation between the X-ray flux hardness
ratio and the radio emission may also be explained by this fact, as the lower energy of
the BATSE flux, below the pivot point, is correlated to the radio emission along with the
ASM flux.
Fig. 5.4: The X-ray pivoting behaviour

during the low-hard state in
Cygnus X-1. The pink and or-
ange points correspond to the
high-soft state. Figure ob-
tained from Zdziarski et al.
(2002)
5.2 Uniform behaviour of X-ray spectral shape with radio

emission
The results presented in the last section will be discussed along with similar results for
GX 339-4 reported by Corbel et al. (2000, 2003). The compact object in GX 339-4 is
also believed to be a black hole (Hynes et al. 2003). Corbel et al. (2000) find that for
GX 339-4, the radio flux is strongly correlated with both soft (see Figure 5.5) and hard
X-ray, covering the range 3-200 keV, in low-hard state, similar to the results obtained by
us for Cygnus X-1.
If we consider the results of the radio, soft X-ray and hard X-ray correlation analysis
for these four sources, at first glance no consistent picture of the correlated variabil-
ity pattern emerges. However, it is immediately noticeable that Cygnus X-3 and GRS
1915+105 have similar overall correlation pattern. For both, the soft X-ray flux is cor-
related with the radio flux, and the hard X-ray flux is anti-correlated with the both radio
and soft X-ray flux. Additionally, the hardness ratio is also strongly anti-correlated with
the radio flux. Wide band X-ray spectral analysis in the hard state for both the sources
at different radio flux levels suggests pivoting of the spectrum around 10 – 25 keV cor-
related with the radio emission.
It can also be noticed that correlation among the radio, soft and hard X-ray fluxes for
Cygnus X-1 and GX 339-4 is similar. Both show a positive correlation among the radio,
soft X-ray and hard X-ray fluxes. For Cygnus X-1 the hardness ratio is not correlated
with radio flux. Further, Zdziarski et al. (2002) find a pivoting of the X-ray spectrum
of Cygnus X-1 at higher energy of around 50-90 keV. Similarly, the wide band X-ray
to 0 -ray spectral analysis of GX 339-4 (Wardziński et al. 2002) has shown that there
is a pivoting in the spectrum at energies 300 keV in the low-hard state of the source
(Figure 5.6).
At this stage it will be worthwhile to note other similarities in the X-ray and radio
emission characteristics of Cygnus X-3 and GRS 1915+105 vis a vis Cygnus X-1 and
GX 339-4. Most notable are that the first two sources are the strongest and most variable
radio sources amongst the Galactic X-ray binaries whereas both Cygnus X-1 and GX
339-4 are amongst the comparatively weak and steady radio sources. On the other hand,
both Cygnus X-1 and GX 339-4 have a very hard X-ray spectrum compared to Cygnus
X-3 and GRS 1915+105.
Thus both X-ray sources with softer X-ray spectrum have a lower pivot energy and
Cygnus X-1 with a much harder X-ray spectrum has a much higher pivot energy, indi-
cating that the pivot energy is directly related to spectral shape. The correlation between
hard X-ray flux and radio and soft X-ray fluxes observed in Cygnus X-1 and GX 339-4
5.2. Uniform behaviour of X-ray spectral shape with radio emission 107
Fig. 5.5: The X-ray - radio scatter diagram of GX 339-4. The X-ray data obtained from RXTE–
ASM monitoring and the radio observations obtained from various observatories. The
distinctive feature is the correlation between radio and the soft X-ray flux in the low-
hard state, and the suppression of the radio emission in the high-soft state. The figure
is obtained from Corbel et al. (2000)
Fig. 5.6: The wide-band X-ray spectra of GX 339-4. top panel:The pivoting of the X-ray spec-
tra above 300 keV. Bottom panel: The individual spectral components of the X-ray
SED. The figure is obtained from Wardziński et al. (2002)
5.2. Uniform behaviour of X-ray spectral shape with radio emission 109
can then be explained because hard X-ray flux (20 – 100 keV for Cygnus X-1, 20-200
keV for GX 339-4) is around or below the pivot energy in these two sources and will,
therefore, be correlated with the soft X-ray and thereby the radio fluxes. Thus, it is quite
evident that X-ray fluxes below and above the pivot energy are anti-correlated for these
X-ray sources and the reported differences between radio, soft X-ray and hard X-ray
correlation amongst these X-ray sources is an instrumental artifact where the ASM and
BASTE energy ranges are fixed and the pivot energy varies from source to source. This
is supported by finding of Zdziarski et al. (2002) who report an anti-correlation between
1.5 – 3.0 keV and 100 – 300 keV flux and find a very weak correlation between 1.5 –
3.0 keV and 20 – 100 keV flux for Cygnus X-1 in the low-hard state. Thus the X-ray –
radio behavior of these X-ray sources are consistent in terms of correlation between soft
X-ray, hard X-ray and radio emissions reported here.
5.2.1 The X-ray soft state and suppressed radio emission

In the previous section we have shown that the X-ray radio emission characteristics in
the hard states of the highly variable sources Cygnus X-3 and GRS 1915+105 are similar
to those seen in the low-hard states of the well studied black hole sources Cygnus X-1
and GX 339-4, once we assume different pivot energy correlated to the radio emission.
Here we explore whether the suppressed radio emission seen in the high-soft states of
Cygnus X-1 (Brocksopp et al. 1999) and GX 339-4 (Corbel et al. 2000) are seen in these
two sources.
We must caution that the identification of various spectral states using monitoring
data is fraught with difficulties of flaring emissions which could be quite delayed in the
various emission bands.
Corbel et al. (2000) show that with the X-ray state transition from low-hard to high-
soft the radio emission evolves from a jet like synchrotron emission to quenched emis-
sion in GX 339-4. This state is generally preceded by a low-hard X-ray (with correlated
radio emission) and followed by an X-ray off state (with radio off). For a comparative
analysis we plot the radio (GBI, 2.2GHz) and soft X-ray (ASM–RXTE, 2-12 keV) scat-
ter diagram in Figure 5.7 for Cygnus X-3 (top panel), GRS 1915+105 (middle panel)
& Cygnus X-1 (bottom panel), for the non flaring states which includes hard as well as
soft states. We have attempted to distinguish the high and low states by the soft X-ray
flux and denoted them by open and filled symbols, respectively. For Cygnus X-3 the
major flares are excluded, and in the process we have excluded the (very low) quenched
radio emission immediately preceding the major flares. For GRS 1915+105, too, the
data for the radio flares are excluded. It is evident that even for these two sources the
radio emission is suppressed in the high state, analogous to the canonical high-soft state.
Fig. 5.7: The radio (GBI, 2.2GHz) and soft X-ray (ASM–RXTE, 2-12 keV) emission scatter
diagram of Cygnus X-3 (top panel), GRS 1915+105 (middle panel) and Cygnus X-1
(bottom panel) for the long-term, steady, hard (filled squares) as well as soft (open
circles) states, after removing the data for the flaring states.
5.3. Universal correlation and its origin 111
Cygnus X-3 shows a very systematic behaviour, with the radio positively correlated to
the soft X-ray in the hard state, until it transits to the soft state, where the radio emission
is negatively correlated to the soft X-ray emission. For GRS 1915+105 the transition
into the soft state with suppressed radio emission is not that drastic but definitely pro-
nounced. Cygnus X-1 shows a more scattered association between the radio and X-ray
emission, but the suppression of the radio emission with higher ASM flux is evident.
Hence, it can be comfortably claimed that the suppression of the radio emission with
the X-ray state transition is a generally consistent feature of the X-ray binary systems
(BHCs), irrespective of their individual spectral characteristics. Therefore a consistent
picture of the accretion-ejection picture is emerging from the observational analysis of
sources with apparently very diverse behavioural patterns.
5.3 Universal correlation and its origin

Thus far we have shown that the four sources, viz. Cygnus X-3, GRS 1915+105, Cygnus
X-1 and GX 339-4, all show a consistent picture of accretion-ejection mechanism, with
the radio emission correlated to the X-ray spectral pivoting in the low state and suppres-
sion of the radio emission in the high state following the X-ray state transition. The most
notable feature, however, is the strong positive correlation between the radio and soft X-
ray flux in the hard state of all the sources. This is similar to the correlation reported for
GX 339-4 (Hannikainen et al. 1998, Corbel et al. 2000, Gallo et al. 2002) and V404 Cyg
(Gallo et al. 2002). We explore below whether the observed correlation is an universal
phenomena among the Galactic black hole candidate sources (Choudhury et al. 2003).
In Figure 5.8, we show a scatter plot of the radio flux against the soft X-ray flux for
Cyg X-1 (plus sign), Cyg X-3 (filled circles) and GRS 1915+105 (open circles), all in
their corresponding low-hard and their analogous states. The data are normalized to a
distance of 1 kpc, with the assumed distances of 2 kpc, 8.5 kpc & 12.5 kpc, respectively
for the above three sources. For the soft X-ray flux (based on RXTE-ASM data) 75 ASM
counts s l is taken as the observed Crab flux. Individual data points are the average value
9 8 P
for a bin size of 5 days. Recently, Gallo et al. (2002) have detected a correlation between
) and X-ray flux (S » ) of the form S Ê
the radio flux (S Ê

ó Ìð ¶ ó Ìð ¶
= k S» for GX 339-4
and V404 Cyg, all the way from the quiescent level to close to high-soft state transition.
This relationship is also shown in Figure 5.8 as a dotted line (for GX 339-4) and dashed
line (for V404 Cyg). The extents of the lines correspond to the data used for the fit in
Gallo et al. (2002).
The remarkable feature of Figure 5.8 is that a simple relation seems to hold for all
the black hole sources over close to 5 orders of magnitude variation in the luminosity,
Fig. 5.8: A plot of radio flux at 2.2 GHz (based on GBI data) normalized to a distance of 1 kpc
against the soft X-ray flux in 2 – 12 keV (based on RXTE–ASM data) normalized to
Crab at 1 kpc for Cygnus X-1, Cygnus X-3 and GRS 1915+105, all in their correspond-
ing low-hard states. The data are averaged for 5 days. The power-law fit (with an index
of 0.7) reported for GX 339-4 and V404 Cyg by Gallo et al. (2002) are shown as dotted
and dashed lines, respectively. The continuous line is a linear fit to the combined data
of Cygnus X-1 and GRS 1915+105.
5.3. Universal correlation and its origin 113
in the low-hard state. The data points for Cyg X-1 fall just below the correlation found
for GX 339-4, whereas, the data points for GRS 1915+105 lie on the line extrapolated
from the correlation derived for V404 Cyg. The data points for Cyg X-3 are parallel
to those of GRS 1915+105. It should be noted here that Cyg X-3 shows a very strong
orbital modulation which is most probably due to the obscuration of the accretion disk
by a cocoon of matter surrounding the accretion disk. Hence the ‘true’ X-ray emission
from the accretion disk should be larger than the observed one and the data points should
move closer to the extrapolated line from GX 339-4 and V404 Cyg. Since Cyg X-3 is
suspected to be a micro-blazar (Fender and Kuulkers 2001), it is also quite possible that
the radio emission is over-estimated due to strong beaming and Doppler boosting. It is
also noteworthy that GRS 1915+105, the most massive stellar mass black hole known,
has the highest intrinsic X-ray emission.
9 8 P
Fitting individual data points of the sources with a function of the form S Ê =k
ó Ìð ¶
S» gives the value of the constant term as 54 mJy, 235 mJy and 1376 mJy, respectively
for Cyg X-1, GRS 1915+105 and Cyg X-3. These values should be compared to those
obtained for GX 339-4 and V404 Cyg, 124 mJy and 254 mJy, respectively. The indi-
vidual data points, however, are also consistent with a linear relation and the continuous
line in the Figure 5.8 is a linear fit to the combined data of Cyg X-1 and GRS 1915+105.
Although we cannot completely rule out the possibility that the individual corre-
lations of these sources may exist due to processes unrelated to one another, the fact
that the data points for sources close to their ‘off’ states (GX 339-4 and V404 Cyg)
occupy one extreme of observations while sources with repeated radio flares (Cyg X-3
and GRS 1915+105) occupy the other extreme strongly suggests of a common physical
mechanism in operation in all the sources. One natural consequence to be explored is
whether the radio and the X-ray emissions are emitted directly from the same source
region. Since there are very strong observational evidences for the radio emission to be
of synchrotron origin, it is tempting to assume that the X-ray emission too is emitted by
the same process, but at the base of the jet. There are evidences for X-ray synchrotron
emission being responsible for the X-ray spectrum of some black hole sources like XTE
J1118+480 (Markoff et al. 2001) and part of the spectrum in sources like GRS 1915+105
(Vadawale et al. 2001b). However, for several black hole sources wide-band X-ray spec-
trum has been extensively studied and the spectral shape is inconsistent with a simple
synchrotron emission. Further, the X-ray:radio correlation appears to be valid from ‘off’
state to low-hard and the associated analogous states, and all the way up to the inter-
mediate state, and it is extremely unlikely that the bulk of the X-ray emission is due to
synchrotron emission in all these states.
Recently Markoff et al. (2003) have suggested jet synchrotron emission as a possi-
ble way to explain the broadband (including X-ray) features of GX 339-4. To explain
the observed correlation, their model predicts that the X-ray emission is mostly due
to synchrotron emission (with a power-law spectral shape). Since the wide band hard
X-ray/ low energy 0 -ray spectra of black hole sources in their hard states like GX 339-
4 (Wardziński et al. 2002), GRS 1915+105 (Zdziarski et al. 2001), and Cygnus X-1
(Zdziarski 2000) require thermal-Compton emission to explain the spectral shape, we
explore below alternate models where the X-ray emission is primarily due to accretion
disk emission.
5.4 X-ray spectral shape as the “driver” of the radio emission

The wide band X-ray spectral shapes of Galactic X-ray binaries with black holes as
compact objects show a systematic and predictable behavior, particularly in well stud-
ied sources like Cygnus X-1 (Zdziarski 2000) and GX 339-4 (Wardziński et al. 2002),
where, in the low-hard state, the spectral energy distribution peaks at 100–300 keV,
with the emission being dominated by thermal-Compton emission from a population of
hot electrons. For sources which go to the ‘off’ state the X-ray spectrum is quite similar
to the low-hard state, but at a very low intensity. In the intermediate state the soft X-ray
flux increases and in the high-soft state the spectrum is dominated by thermal emission
from the accretion disk. Although it is quite convenient to assume that the mass accretion
rate is responsible for such systematic changes, though there is no strong evidence for
this (see Homan et al. 2001, for a different behaviour of XTE J1550-654). Nevertheless,
one can safely conclude that some unspecified ‘accretion parameters’ causally affect the
wide band X-ray spectral shape of black hole X-ray binaries.
The evidences presented in this thesis strongly suggests that the very same ‘accretion
parameters’ must be causally responsible for the radio emission, rather than the amount
of soft X-ray emission, provided they account for the suppressed radio emission in the
soft state. Such a hypothesis neatly explains the behavior of Cygnus X-3, which is quite
bright in all the three energy ranges and hence the observational uncertainty is quite low.
It can also be noticed from table 5.1 that the most significant correlation for Cyg X-3 is
between the radio emission and the ratio of hard X-ray flux to soft X-rays. Though such
an explanation is not very clear in other sources, all the available observations are consis-
tent with this. Since the radio emission too shows increasing emission from ‘off’ state to
low-hard state (in GX 339-4 and V404 Cyg), correlated behavior in low-hard state (the
above two sources and Cygnus X-1), and high radio emission in an intermediate state
very close to the high state (in GRS 1915+105 and Cygnus X-3) and suppressed radio
emission in the high state (Cygnus X-3, GRS 1915+105, Cygnus X-1 & GX 339-4), we
5.5. Summary: the generalized picture of the accretion - ejection mechanism in the ‘low’ - hard state of Galactic m
speculate that the soft X-ray intensity determines the spectral shape and the accretion
disc condition in these sources, which in turn determines the amount of radio emission
(in the X-ray quiescent hard state of these sources). The transition of the Cygnus X-3
system into flaring state and their corresponding behavior has been discussed in detail
in chapter 3, while that of GRS 1915+105 is left as a field of future study, given the
varied amount variability exhibited by the source, and the vast amount of archived data
available, which will require a very extensive program of analysis and study.
5.5 Summary: the generalized picture of the accretion -

ejection mechanism in the ‘low’ - hard state of Galactic
microquasars
In this chapter we have analyzed the (quasi) simultaneous observations on GRS 1915+105
and Cygnus X-1 using the RXTE-ASM, BATSE-CGRO and GBI data and made a de-
tailed study of correlation between radio and X-ray fluxes. Based on this analysis along
with discussion of earlier published results on Galactic microquasars we find that:-
$ A correlation exists between the soft X-ray and radio emission of GRS 1915+105

based on the data during the long state (associated to the low-hard state). The
hard X-ray emission is anti-correlated with both radio and soft X-rays. There is a
spectral pivoting at around 20 keV, correlated with the radio flux.
$ Comparing these results with those of Rau and Greiner (2003) who found a strong

correlation between radio emission and the X-ray spectral index in the states,
we conclude that the X-ray and radio emission characteristics of GRS 1915+105
are similar to those of Cyg X-3 (Choudhury et al. 2002). The only difference lies
in the values of the pivot energy of the X-ray spectra, which is around 12 keV in
Cyg X-3 and around 20 keV in GRS 1915+105.
$ A three way correlation among soft X-ray, hard X-ray and radio emission has
been found in the low-hard state of Cyg X-1, confirming the results of Brocksopp
et al. (1999). Comparing this result with those of Zdziarski et al. (2002) who have
found that soft X-ray and hard X-ray above 100 keV are anti-correlated and also
that there is a spectral pivoting at around 50 – 90 keV, we conclude that the X-
ray:radio behavior of Cyg X-1 is similar to that of Cyg X-3 and GRS 1915+105,
but for the fact that the pivoting energy is at a higher value.
$ The X-ray:radio properties of Cyg X-1 are quite similar to that of GX 339-4, where
a 3-way correlation between soft X-ray, hard X-ray and radio emission has been
reported (Corbel et al. 2000, 2003). Though an anti-correlation between soft X-
ray/radio with hard X-rays has not been reported in this source, we note that the
X-ray spectrum during the low-hard states also show a pivoting behaviour at high
energies 300 keV (Wardziński et al. 2002).
$ The radio emission is suppressed for Cygnus X-1 & GX 339-4 in their high-soft
state and similarly for Cygnus X-1 and GRS 1915+105 in their high states (with
associated softer spectra). Therefore, all these four sources with apparent diverse
X-ray and radio properties show very similar behavioural pattern encompassing
the long term steady non-flaring state.
$ Compiling the soft X-ray and radio observations of the above sources (GRS 1915+105,
Cyg X-3, and Cyg X-1) with the published correlation in GX 339-4 and V404 Cyg
(Gallo et al. 2002), we find that all the sources show a monotonic increase of radio
emission with the soft X-ray emission, spanning a 5 orders of magnitude variation
in their intrinsic luminosities. Cyg X-3 deviates from a single relation by about an
order of magnitude which can be reconciled if 1) the observed X-ray intensity is
an under-estimate because of obscuration and/or 2) the observed radio intensity is
an over-estimate because of beaming and Doppler boosting.
$ If a common physical phenomena is responsible for such an uniform relation span-
ning across ‘off’ state to intermediate state, we argue that both radiations (X-ray
and radio) are unlikely to be originating from a single mechanism like synchrotron
emission.
Thus we succeed in forming a generalized picture of the disc - jet connection in

the Galactic microquasars systems, spanning across (apparently) very diverse types of
variability as well as intrinsic high energy and radio emission features. The next logical
step would be to construct a concise phenomenological model explaining these obser-
vational features and, at the least, putting some constraints on the various physical and
geometrical paradigms that exist in the literature today.
Chapter 6
Two Component Accretion Flow

model
In the introduction to this thesis (chapter 1) the various existing theoretical paradigms for
the accretion phenomena were introduced, with a brief mentioning of the Two Compo-
nent Accretion Flow (TCAF) model (Chakrabarti 1996a) involving the concept of bulk
motion Comptonization. In the subsequent chapters it has been established, with partic-
ular emphasis on the X-ray binary system Cygnus X-3, that the accretion and ejection
need to be considered in an associated manner to form a unified picture of the physical
and geometrical structure of the systems. In this chapter an attempt is made understand
and explain the observational features detailed out in the chapters 3 & 5 in the light of
the TCAF paradigm in order to provide a unified phenomenological picture, as ordained
L
by the aim of this thesis ( 1.6).
Bulk Motion Comptonization. Comptonization, in its most elementary treatment as

the cause of the high energy emission (powerlaw tail), is considered as upscattering of
the soft photons from the disc by inverse Comptonization by a corona of hot electrons
with a Maxwellian (thermal) distribution, with a temperature of 50-100 keV, but this
formulation is found to be inadequate. In the bulk-flow Comptonization scenario seed
photons gain energy from collisions with radially infalling electron’s in a geometrically
thick accretion disc in the inner orbits around the compact object (for a short review,
see Chakrabarti and Chattopadhyay 2002). Chakrabarti and Titarchuk (1995) showed
that this infalling matter, which needs to enter the event horizon at the velocity of light
(Chakrabarti 1996a), is capable of upscattering of the photon even if it is ‘cool’, for the
relativistic velocity of the bulk motion (instead of the thermal motion) can Comptonize
the seed photons to the said high energy bands. The accretion flow is advective and only
117
118 Chapter 6. Two Component Accretion Flow model
Fig. 6.1: Schematic (cartoon) diagram of the TCAF paradigm, highlighting the Keplerian accre-
tion disc and the sub Keplerian accretion flow around the compact object, along with
the CENBOL (see text for details).
a few photons scattered from this cold but relativistic electrons escape and produce a
powerlaw spectrum.
6.1 Two Component Accretion Flow (TCAF)

The TCAF model (Chakrabarti and Titarchuk 1995), which is based on the most general
advective flow solutions (see, for eg. Chakrabarti 1990, 1996b,c,d), predicts that the ac-
cretion flow onto the compact object (blackhole) consists of two components, namely, a
Keplerian and a sub-Keplerian component. The Keplerian component consists of a geo-
metrically thin and optically thick Keplerian accretion disk, where the accreting matter
has high viscosity as well as angular momentum, analogous to the SS disc solution of
Figure 1.7. The sub-Keplerian flow, which is necessary for the matter to fall into the
blackhole horizon, consists of matter with lower viscosity and angular momentum, sur-
rounding the thin accretion disk as well as occupying the space of the inner orbits of the
truncated disc. The geometry of the truncated disc is dictated by the fact that the matter
entering the horizon has the velocity of the light, and hence the flow cannot be Keplerian
anywhere near the last stable orbit. Instead, the disc is truncated a a transition radius,
6.1. Two Component Accretion Flow (TCAF) 119
´
Ê £
! , beyond which the flow becomes nearly free fall, with possible manifes-
tation of bulk motion Comptonization. Due to the lack of an agent to transfer the angular
momentum outwards, the freely falling sub-Keplerian flow has (nearly) constant angu-
lar momentum resulting in a strong centrifugal force, which becomes comparable to the
gravitational force after some distance. Thus the centrifugal force balances the gravita-
tional attraction and causes a barrier in the sub-Keplerian flow, the Centrifugal pressure
supported Boundary Layer (CENBOL). Nevertheless, further close to the black hole the
gravity overcomes the CENBOL and the matter enters the event horizon supersonically.
The CENBOL may form a standing or a scintillating shock (Molteni et al. 1996, Ryu
et al. 1997, for a detailed quantitative treatment see Chakrabarti 1996a). The shock,
if present, heats up the post shock region which forms a puffed up hot cloud (Figure
6.1). The soft photons from the disc are energized to hard X-rays by inverse Compton
scattering. Therefore the total high energy spectrum is explained by the disc blackbody
from the Keplerian component plus the Comptonized component, which may include
the non-thermal population with bulk motion. The canonical states of X-ray emission
may be explained by the TCAF model as states defined by the accretion rate, both in the
t
Keplerian disc P Á as well as in the sub-Keplerian cloud P Á . The observational feature

of this model is the wide-band X-ray spectra with multi-coloured disc blackbody plus
a strong hard (powerlaw) component. The angular momentum of the inflowing matter
slows down the radial velocity, increasing the optical depth, which causes the slope of
the high energy tail spectra to increase (Chakrabarti et al. 1996, Ebisawa et al. 1996).
¢ ¢
P P P
µ
Low-hard state. When Á
º .Á (where Á D mass accretion at Ed-
dington limit) Ê /
, beyond which the flow is Bondi-type, sub-keplerian
and hot. This hot, Comptonizing cloud produces hard X-ray by Comptonizing the soft
thermal disc (seed) photons.
> As Ê is quite large in this state and the inner disk tem-
Q x
perature varies as l (see Longair 1994), the inner disk temperature is quite low in
this sate. Also, certain features like presence of quasi-periodic oscillations seen in some
systems may be explained by the oscillation of the shock front of the CENBOL (Figure
6.2). The low energy and low angular momentum flow behaves like a spherical Bondi
inflow and Parker winds (Parker 1960). Or else, the steady inflow and outflow may not
possess shock but the inflow is hot due to slowing down at the centrifugal barrier. In
addition, transonic viscous flows are also possible. At low viscosity the shock is obvi-
ously weak and the optically thin (sub-Keplerian) flow joins the Keplerian disc far out.
The extent of the sub-Keplerian flow may decide the hardness of the X-ray wide-band
spectra, with the slope of the omnipresent powerlaw component depending on various
physical factors.
Fig. 6.2: The accretion flow as per the TCAF paradigm in the low-hard state, with and without
the shock barrier. Panel a: Shock is formed, QPOs may or may not be present. Panel
b: Shock is not formed, QPOs are not expected. In both cases wind may be formed at
low rate. (The figure is obtained from Chakrabarti 2000)
High-soft state. When P Á

+P ¢
P
¢
Á (where Á
D mass accretion at Eddington
limit) temperature of the inner accretion disk is high resulting in a strong multi-
coloured disk blackbody type emission. The cooling of the Compton cloud is very effi-
cient due to the enhanced flux of thermal seed photons, and the disc may extend close to
the last stable orbit. The presence of the hard X-ray tail in the spectra may be explained
by the bulk motion Comptonization of the seed photons by the energetic matter in the
Compton cloud possessing a bulk velocity as the cloud pushes into the blackhole hori-
zon at relativistic speeds. The building blocks of the accretion-ejection flow structure
contains, in addition to the ones in the low-hard state, solutions with the inflow pass-
ing through the inner sonic point, as well as all the possible combinations of inflow and
outflow possessing or not possessing shocks (Chakrabarti 2000).
6.2 Outflow of mass

Outflows exist in all sorts of astronomical objects, stellar or otherwise. The primary
difference between the outflow from accreting X-ray binaries and the ordinary stellar
sources lies in the fact that for the former the mass outflow consists completely of the
6.2. Outflow of mass 121
Fig. 6.3: The accretion flow as per the TCAF paradigm in the high-soft state, with and without
the bulk motion Comptonization. Panel c: Bulk motion Comptonization is present,
with shocks (if formed) cooled down in accretion, without any QPO or significant
wind outflow. Panel d: Flaring outflow state where shocks of intermediate strength
may form but since the outflow rate is high it may be periodically cooled to produce
strange behaviours. (The figure is obtained from Chakrabarti 2000)
mass from the inflow. The approach to find the hydrodynamical or magnetohydrodynam-
ical pressure effects leading to the formation of the jets by treating the outflow separating
from the inflow has been found to be inadequate. In the TCAF paradigm it is possible to
study, self consistently, both the mass inflow and outflow, as the same set of equations
govern the mass flow, with the wind type solution guiding the outflow and the accretion
type solution deciding the inflow, provided the existence of the CENBOL is accepted.
One may strongly assert that whether or not the shock actually forms, the inner dense hot
region with (quasi) spherical Bondi-like flow does exist (Abramowicz 1998), even for
different physical effects considered (for eg., Chang and Ostriker 1985 showed that pre-
heating of the gas could produce standing shocks at large distances, and Kazanas and El-
lison 1986 mentioned that pressure due to pair plasma could produce standing shocks at
smaller distances around black hole as well). Therefore it is imperative to investigate the
possible role of this region in the formation of¢ outflows. Since the gas will be hot in this

region, for larger accretion rates (P Á
P Á ) the cooling of this region by the Comp-
tonization from the seed (disc) photons is capable of producing outflows in the low-hard ¢
states, although it doesn’t require the very high accretion rate (P Á
P Á ) of the
ðøñ
so called ‘twin-exhaust’ model (Begelman et al. 1984) which requires a similar physical
and geometrical structure. Chakrabarti (1999) details the model which assumes a CEN-
BOL of size
with the matter passing through a sonic point using the pre-determined
funnel where rotating pre-jet matter is accelerated (Chakrabarti 1984), and the outflow
rate is analytically completely from the inflow rate alone. Analogous to the solar case
where photons from stellar surface deposit momentum to the solar winds (at least upto
the sonic point), the hard photons deposit the momentum to the outflowing wind close to
the blackhole and keeps the flow roughly isothermal (at least upto the sonic point). With
a near Eddington luminosity the outflow could be
lr ¬ year l (i.e. roughly half the
Eddington rate for a stellar mass star). Thus, if the flow is compressed and heated at the
centrifugal barrier it would also radiate enough to keep the flow isothermal (at least upto
q L
the sonic
as l
Q :
while the photon density
·¶
> point) provided the efficiencies
q
are identical.:
falls off as l , hence the
¶
The electrondensity
q L q
> : falls off
G increases

with the size of the region, resulting in a less number of electron’s per photon, making
the process of momentum transfer more efficient, near a blackhole.
Assuming the free-falling conical polytropic inflow and the isothermal outflows, Das
and Chakrabarti (1999) estimate the ratio of outflowing and · the inflowing rate to be
-½¼ ¬ /
>
Q :
Á
¹¸
7a 7 º _» +
¶
=
l
Ë
l X (6.1)
Á
ðòñ

º ðòñ
where
º _»
¶
and
º ðòñ
are the solid angles of the outflow and inflow respectively, is the
compression ratio of the in-flowing matter which is a function of the flow parameters
6.2. Outflow of mass 123
Fig. 6.4: Left Panel: Ratio

¾À Á ¿
rate and the inflow rate as a function of the compression ratio
of the gas at the dense region boundary. Right panel: The variation of the ratio of the
Â
polytropic constant in the strong shock limit is shown. (The figure is obtained from
Chakrabarti 1999)
such as specific energy and angular

· momentum (Chakrabarti 1990) and f is given by
: qmy
7

7
(
q

(6.2)
a
(
¹¸
0.052 and 0.266 for 0 7
[+
n is the polytropic constant = 50öW
l 0 being the adiabatic index. When
¶ º _»

. and . respectively. Assuming a thin inflow
º
ðòñ
,

and outflow 10 " conical angle, the ratio

¸
becomes 0.0045 and 0.023 respectively,
provided that the outflows are normally concentrated near the axis, i.e. are collimated,
whereas the inflow is near the equatorial plane, which gives éëê 7 ÃÅÃ ÄÇÆ ?

. . Thus, the
ù
outflow rate is found to depend only on the compression ratio and the collimating
ÃÅÃ ÄÇÆ ?
property of the outflow é ê and not on the sonic point location, the size of the shock,
or the outward radiation force (also the centrifugal force has been ignored) (see Figure
6.4). All these will come into play in the case for a complete general relativistic theory.
Note that (in the Figure 6.4) if the denser region does not from ( £
) then the outflow
does not form. Therefore, the outflow is originating in the hot, compressed region. In
situations where the assumption that the flow is isothermal (at least) upto the first sonic
point is dropped, the outflow becomes a function of many parameters depending on the
equation of state, but still qualitatively the solution for the outflow is still similar to the
one shown (Figure 6.4)(numerical results are given in Das and Chakrabarti 1999).
The left panel of Figure 6.4 shows the behaviour of the outflow for any generic
¶¶
compression ratio while the right panel shows the same outflow for the polytropic index
q
for w7
l
7(
qzy
· strong shock case only (Chakrabarti 1999). The most

, i.e. the
obvious feature
average strength at ú '( +
of the solutions is that the outflow rate is peaked when the shock is of
, with the rate falling off very rapidly on either
side. If the shock strength is on the higher side ( -( ) the object is in the low-hard
state and in the high-soft state on the other side ('D
z pure soft state, 'D 4–7
pure hard state). Usually the outflowing jets are observed to be hollow, hence they
must be externally supported (by ambient medium pressure or magnetic hoop stress) as
the collimation of these is not explicitly considered in the treatment given above.
6.3 The magnetized TCAF model

Although magnetic fields are considered to be omnipresent in most of the cosmic sources,
in blackhole systems the origin of these is trifle difficult to explain in contrast to the neu-
tron star and white dwarf systems, where they are generated within the compact object
itself. The standard approach is to amplify any stochastic magnetic field present to the
value required by equipartition. This may employ the use of azimuthal shear in the ac-
cretion disc. The value of the magnetic field may vary from
–
G very close
to %
G at about 1 AU distance from the blackhole to of mG at 500–1000 AU
distance (in the radio lobes) (Fender et al. 1997a, Rodrı́guez and Mirabel 1999). These
values when interpolated close to the black hole assuming a simple
dependence,

give the same order of magnitude value for the magnetic field. The equipartition value
P
of magnetic field in the inner accretion disk is expected to be about
–
G. Thus it
is essential to incorporate the magnetic field in any model for an accurate description of
the blackhole environment (Vadawale 2003).
The introduction of magnetic field in the TCAF paradigm (Nandi et al. 2001) based
on the earlier works on magnetic activity in thick accretion disks (Chakrabarti and
D’Silva 1994, D’Silva and Chakrabarti 1994) shows that it is possible for the entire
Compton cloud to be ejected (by the so called magnetic rubber-band effect by Nandi et
al. 2001, see Figure 6.5). The stochastic magnetic (seed) field is supposed to originate
in the accreting matter and toroidal magnetic fields are formed due to the shearing of
the stochastic field by strong azimuthal velocities in the accretion flow. These flux tubes
may rise to the surface of the accreting flow due to buoyancy. After crossing the shock
front (CENBOL) the tubes experience drastic increase in the temperature resulting in
the enhancement of the buoyant force (in comparison to the radial magnetic tension and
drag) which ejects the flux tubes perpendicular to the accretion flow. Such ejection of
the magnetic flux tube takes place at the Alfven velocity which can be very large for
6.4. Phenomenological picture of accretion and ejection connection 125
Fig. 6.5: A schematic diagram of the accretion disk near a black hole in the presence of magnetic
ÈÉ
field. It includes a shock ( ) and a sub-Keplerian and a Keplerian disk with boundary
ÈÊ2Ë
at . Stochastic magnetic fields are sheared and amplified as they leave a Keplerian
disk. In a hot sub-Keplerian flow these toroidal flux tubes are catastrophically ejected
evacuating the post shock region . (The figure is obtained from Nandi et al. 2001).
the post-shock region. Depending on the combination of the various parameters of the
accretion flow, if the number of such ejected magnetic flux tube crosses some critical
value, it might lead to the complete evacuation of the central Compton cloud.
6.4 Phenomenological picture of accretion and ejection

connection
Presently we will discuss the applicability of the TCAF framework, developed in the pre-
ceding sections, to provide a phenomenological picture of the physical and geometrical
structure of Cygnus X-3 (chapter 3) and other microquasars (chapter 5).
6.4.1 Cygnus X-3

Cygnus X-3 is an X-ray binary with a compact object whose nature has been not yet been
identified. Although being one of the brightest radio sources in the sky, there has been
only one isolated report of an apparent superluminal motion in the radio band (Newell
et al. 1998). One of the primary aim of this thesis was to glean some simple X/0 -ray
properties to ascertain the nature of the compact object. During the early days of wide-
band X/0 -ray spectroscopy the hard powerlaw tail seen in the blackhole systems were
believed to be signatures of absence of a hard surface, and hence the presence of an event
horizon (Tanaka 2000). In Cygnus X-3 a similar hard tail is seen (Figure 6.6) till 1 MeV
(and perhaps beyond). But, subsequent observations of similar spectral characteristics
in neutron star X-ray binary systems (see, for eg. Barret et al. 2000, 2003) by RXTE and
BeppoSAX negated the possibility of identifying the nature of the compact object by
such a simple observational feature (see Barret 2001, 2004, for reviews). Despite of
the presence of the powerlaw tail in the neutron star systems, the overall spectral shape,
evolution and correlation with radio emission do suggest the compact object to be a
likely blackhole candidate.
Fig. 6.6: The hard powerlaw tail in the

Ì
wide-band X/ -ray spectrum of
Cygnus X-3 as observed by
CGRO– OSSE.
Low (hard) state of X-ray emission: Pivoting in X-ray spectra. In chapter 3 a de-
tailed account of the statistical test of the correlation among the radio, soft and hard
X-rays were given. Cygnus X-3 is the only source that shows such a strong correla-
tion between soft X-ray and radio, and it is also the only source to distinctly show the
anti-correlation between the soft and hard X-rays, during the low-hard state. The quies-
cent state of Cygnus X-3 has persistent (flat spectrum) radio emission (60 – 100 mJy).
This state in Cygnus X-3 is likely to be similar to the “plateau”radio state seen in the
most active micro-quasar GRS 1915+105 which shows flat spectrum radio emission for
extended durations and the radio emission is identified with a compact jet of size 10
AU (Dhawan et al. 2000). The spectral changes in association with the radio emis-
sion is also quite similar to GRS 1915+105 (Choudhury et al. 2003, Vadawale et al.
2001a). Recently it has been suggested that X-ray emission from BHCs GRS1915+105
(particularly during the “plateau” state) and RXTE J1118+480 could be arising from
synchrotron emission from the base of the jet (Vadawale et al. 2001a, Markoff et al.
2003, 2001). Therefore it leads to a speculation that some of the X-ray flux in Cygnus
X-3 also could be arising from synchrotron emission from the base of the jet. This
can explain the correlation between soft X-ray and radio fluxes but fails to explain the
anti-correlation between soft and hard X-ray fluxes. It is quite likely that the soft X-ray
emission is due to the accretion disc (directly or indirectly) and the observed correla-
tion is due to a connection between the accretion disk and the jet emission. The X-ray
emission from Cygnus X-3 is highly obscured and the bulk of the X-ray emission be-
low 5 keV is due to the emission-line dominated photo-ionized plasma surrounding the
compact object (Paerels et al. 2000, Kawashima and Kitamoto 1996). Hence there is
no clear evidence for the disc blackbody emission, commonly seen as an X-ray spectral
component in the soft X-ray region in other BHCs. Our spectral analysis above 5 keV
has identified two spectral components and these, by analogy with other BHCs, can be
identified with thermal/non-thermal Comptonization (Zdziarski et al. 2001, Gierliński et
al. 1999) occurring in the source (or due to X-ray synchrotron emission - see above). If
we assume that the region of the Comptonization is confined to a small region near the
compact object, we can qualitatively explain the observed correlations under the TCAF
described above, in which the Compton spectrum originates from a region close to the
compact object, confined within the CENBOL. At low accretion rate, the CENBOL is
far away from the compact object, the spectrum is harder with lower outflow (Das and
Chakrabarti 1999). On increasing the accretion rate the CENBOL comes closer to the
compact object with greater outflow, giving rise to increased radio emission. Though
this model qualitatively explains the observed correlations, we must add here that the
thermal Compton model is only an approximation and a correct Comptonization model
requires an accurate description of the geometry of the emission region.
High (soft) state of X-ray emission: Disappearing of Compton cloud. The spectral
evolution detailed out in chapter 3 gives a complete picture of the observational features
spanning all the possible states of X-ray as well as radio emission. The disappearance of
the CompST component in the observations preceding the radio flares suggest the evac-
uation of the central region around the compact object, where the accretion flow is hot
and quasi-spherical. This observational feature provides a direct evidence of the geomet-
rical structure favouring the truncated disc scenario, with the inner orbits having been
replaced by Bondi type flow (albeit with comparatively higher angular momentum, as-
suming the presence of CENBOL). Disappearance of the Compton cloud has been seen
in GRS 1915+105, during the outflows in the low state (Vadawale et al. 2003, 2001a). It
is argued a series of such dips, i.e. evacuation of Compton cloud, give rise to superlumi-
nal flares (Naik et al. 2001). Therefore, one may claim that the general evolving picture
of the geometrical structure is zeroing onto the truncated disc (sombrero) paradigm. The
most intriguing feature of the X-ray spectra of Cygnus X-3 is, perhaps, the post-radio
flare state, where the spectral shape is best fit by model components identical to those
of the low (hard) state, but the soft X-ray flux is higher, comparable to the high (soft)
state. The low temperature of the CompST parameter fit value may suggest the presence
of bulk motion Comptonization, which hinders the formation of jet, as matter is acceler-
ated (advectively) into the horizon. Thus, at a qualitative level, the TCAF paradigm may
be used to explain the observational features of the microquasar X-ray binary system,
Cygnus X-3.
Pivoting in the X/0 -ray spectra of other microquasars. Analogous to the explana-
tion for Cygnus X-3, a similar phenomenological picture may be drawn to explain the
X-ray radio correlation tests and the X-ray spectral pivoting (at the various energies)
of the different types of microquasars, viz. Cygnus X-1, GRS 1915+105, GX339-4, in
addition to Cygnus X-3. Though there are models describing the accretion disk emis-
sion (Zdziarski 2000) or jet emission (Markoff et al. 2003), there are very few models
which self-consistently solve the accretion and ejection phenomena seen in black hole
sources. Since our findings suggest a close connection between these two phenomena,
we attempt below to qualitatively explain the X-ray radio association using the TCAF
paradigm. The X-ray spectral shape in various ‘states’ of the various sources essentially
depends on the location of the CENBOL. Analogous to the special case of Cygnus X-3,
at low accretion rates the CENBOL is far away from the compact object and the X-ray
spectrum is dominated by a thermal-Compton spectrum (if bulk motion Comptonization
is absent). In the transition state, the CENBOL comes closer to the compact object and
can sometimes give rise to radial shocks, causing intense quasi-periodic oscillations, as
seen in GRS 1915+105. In the high state, the increased accretion rate produces copi-
ous photons in the accretion disc which cool the Compton region, giving rise to very
intense disk blackbody emission along with bulk motion Comptonization (a power-law
in hard X-rays with a photon index of 2.5). At some critical accretion rates, the state
transitions could be oscillatory as seen in GRS 1915+105 (Chakrabarti and Manickam
2000).
The behavior of TCAF disks and the outflow has been stated in detail above (Das and
Chakrabarti 1999, Chakrabarti 1999), where the outflow rate is found to be a monotonic
function of the compression ratio, , of the gas at the shock region. In this scenario, at
low accretion rates, the CENBOL is far away from the compact object, a weak shock can
form with low compression ratio, giving low and steady outflow. If this outflow gives rise
to radio emission, one can expect a relation between the radio emission and the X-ray
emission. In this state (off state to low-hard state), an increased accretion rate increases
the overall amount of energy available to the Comptonizing region and hence increasing
the X-ray emission. The CENBOL location would be pushed inward, increasing the
compression ratio (and hence increasing the radio emission) and also can increase the
temperature and optical depth of the Comptonizing region, thus giving rise to a pivoting
behavior at hard X-rays as seen in Cygnus X-1 and GX 339-4 (50 – 90 and - 300 keV,
respectively). At increased accretion rate, the CENBOL can come closer to the compact
region, giving the spectral and radio properties as seen in GRS 1915+105 and Cygnus
X-3. For a given accretion rate the compression ratio, after reaching a critical value (with
the shock region coming correspondingly closer to the event horizon), causes the source
to transit into the high-soft state state, for which the radio emission is progressively
suppressed (Chakrabarti 1999). This model qualitatively explains all the observed X-ray
spectral and radio properties of Galactic black hole sources presented in chapter 5.
6.4.2 Hybrid Comptonization

Although the observational features of the X/0 -ray spectra, their evolution with the ra-
dio emission, can be qualitatively explained by the TCAF paradigm, in the current state
of affairs it is imperative to discuss other possible alternatives for the accretion as well
ejection mechanisms. The hybrid Comptonization model of Zdziarski (2000) provides
a viable alternative, with a truncated disc (sombrero) geometry, which may explain the
X-ray spectral behaviour of the microquasars, Cygnus X-1 in particular (Zdziarski et al.
2002). The hot Comptonizing component consist of thermal as well non-thermal popu-
lation of matter. Selected electron’s from the thermal distribution may be upscattered to
relativistic energies, possibly in the reconnection events. These non-thermal relativistic
electron’s Compton upscatter the disc (seed) photons, forming the high energy tail. The
relativistic electron’s also transfer some of their energy via Coulomb scattering to the
thermal electron’s, heating them to a temperature much above the Compton tempera-
ture. The radiation of the corona is also partly Compton-reflected in the disc. In the soft
state, the geometry of the system is different, with a patchy-corona above a standard op-
tically thick disc. The soft (seed) photons from the disc are (again) Compton upscattered
in the flares, and the emission from the flares is partly Compton-reflected from the disc.
This framework may also qualitatively explain the pivoting of the high energy spectra,
but it doesn’t concern the radio outflow as an inherent feature of the accretion system.
Nevertheless, features of this paradigm may provide a consistent building block in the
Fig. 6.7: The geometry of the hybrid Comptonization model. Top panel: the accretion in the
high state, with the disc extending nearly upto the last stable orbit giving the strong
disc blackbody component. The Comptonizing region consist of small active clouds
scattered over the disc. Lower Panel: the accretion in the low state, with the trun-
cated disc having the inner orbits replaced by the Comptonizing cloud. (The figure is
obtained from Zdziarski et al. 2002)
mechanism of the accretion physics in this class of objects.

Chapter 7
Summary and conclusions
This thesis was aimed at providing a unified, consistent set of observational features of
Galactic microquasars, leading to the development of a phenomenological model taking
the diverse observational characteristics into account, and thus provide a first step to
form a detailed physical theory of the accretion-ejection mechanism in these systems. In
the following section we provide a brief summary of our analysis and the conclusions
derived from them, with a view for the future direction of work.
7.1 Summary and conclusions

7.1.1 Cygnus X-3
The salient features of the general X-ray spectral studies of Cygnus X-3 are given as
follows:-
$ The X-ray spectra exhibits two distinct states, low (as well as hard) and high (as
well as soft), with a high energy tail (extending upto 10 MeV and perhaps beyond)
in both these states.
$ There is very high inherent absorption in the source which obliterates any signa-
ture of the disc blackbody (thermal) emission in the low state. The complicated
and peculiar continuum spectral shape is best fit by a combination of Comptoniz-
ing component (CompST) and a powerlaw, in this state.
$ The continuum spectra of the high (as well as soft) state is best fit by a combination
of disc blackbody component plus Comptonizing component (CompST) in most
situations.
131
132 Chapter 7. Summary and conclusions
$ The three iron lines are not resolvable by RXTE– PCA, and hence no attempt is
made to derive any physical conjectures from the iron line emissions, in this thesis.
The salient features of the long term multi-wavelength monitoring of Cygnus X-3 are
given as follows:-
$ The soft X-ray (2-12 keV, RXTE– ASM) and radio (2.2 GHz, GBI) are very
strongly correlated in the low (as well as hard) state. In the high (as well as soft)
state the long term association is not that prominent from the daily monitoring
data.
$ The soft (2-12 keV, RXTE– ASM) and hard X-rays (20-100 keV, CGRO– BATSE)
are anti-correlated in the low (as well as hard) state, the long term association is
not so prominent in the daily monitoring data in the high (as well as soft) state.
$ Correspondingly, the radio and hard X-ray emissions are anti-correlated in the low
(as well as hard) state.
$ The hardness ratio of the soft and hard X-rays is also anti-correlated to the radio
emission, more strongly so than the hard X-ray emission.
$ The time scale of all these correlations is less than a day, revealing that the dynam-
ical time scale need to observed by the pointing instruments for long durations.
$ The anti-correlation of the soft and hard X-rays is explained by the pivoting of
the wide-band X/0 -ray spectra correlated to the radio emission, in the low (hard)
state.
The TCAF paradigm, which deals with the inflow and the outflow in a unified frame-
work, provides a qualitative understanding of the low (hard) state properties of the X-ray
as well as the radio emission and their association in the source.
In the high (soft) state a more detailed study was required to understand the radio:X-
ray correlated emission and a complete X-ray spectral evolution was analyses for this
state. This state can be further subdivided into three phases, quiescent, pre-radio flare
and post-radio flare periods. The salient features of the X-ray:radio association in this
state is given below:-
$ The radio quiescent emission is marked by the radio emission (2.2 GHz) bordering
around 110 mJy and below. The X-ray spectra has strong CompST component,
the emission due to this component is comparable to that of the multicoloured disc
black body component. The fraction of CompST flux amounts to £(,+ ,+
of the total X-ray flux (5-60 keV).
7.1. Summary and conclusions 133
$ A minor flare is always preceded by the vanishing of the CompST component

(flux going below 15% of the total X-ray flux, 5-60 keV). The minor radio flares

have flux around £
! mJy. This phenomenon suggests the ejection of the
central Compton cloud resulting in the flare. Also, the extent of reduction of the
CompST component is loosely correlated to the maximum flux of the flare, i.e.
the stronger the ejection of the cloud the louder the flare.
$ The dynamical time scale of the decrease of CompST flux and the observation of
the minor flare in the radio may be less than a day. Interestingly, less extent of the
decrease of CompST component causes quicker, but milder, flares.
$ The minor flare may be followed by the filling of the central Compton cloud,
i.e. increase in CompST flux, causing the radio emission to become quiescent.
Otherwise, if the continuous accretion persists with the central cloud unfilled, i.e.
a`
the CompST flux remains low, a major radio flare (2.2 GHz, flux -I
) follows.
$ The continuing series of minor and major flares come to an end only with the
change in the X-ray spectra, i.e. hardening of the soft X-ray band, with the flux
level remaining high. This is the most interesting state of the X-ray spectra with
the shape being best fit by the model of the low (correspondingly hard) state, i.e.
power law and CompST, although the soft X-ray flux remains high. This change
in the X-ray spectra puts a brake in the episodes of radio flaring.
The above sequence strongly suggest the evacuation of a central hot Compton cloud
resulting in the massive outflows visible in the radio bands, in this system. Thus, from an
observational perspective, a complete X-ray spectral evolution associated with the radio
emission, covering all the possible states of X-ray and radio emissions, was obtained for
the first time for this source. The TCAF model may be employed with a qualitative basis
to explain the physical mechanism and the geometrical structure of the system. A brief
outline of the phenomenologically obtained picture is given below:-
$ In the low (hard) state, the geometrically thin and optically thick Keplerian accre-
tion disc is truncated far off from the last stable orbit, the inner region is filled by
the Bondi-type quasi spherical sub-Keplerian inwards flow. The two regions are
divided by the CENBOL.
$ The soft thermal disc black body photons are Comptonized by the hot plasma in
the central Compton cloud, which may or may not possess a bulk motion. These
reprocessed photons may get scattered (reflected) from the optically thick disc
which may be Comptonized, in a feedback mechanism.
$ Perturbations in the accretion disc may cause the CENBOL to move in or out,
i.e. changing the (radial) dimension of the accretion disc with a correspondingly
opposite change in the dimension of the Comptonizing cloud. This feature is man-
ifested in the anti-correlation between the 2-12 keV and the 20-100 keV photons,
best visualized by the pivoting of the X-ray wide band spectra.
$ Although the exact physical mechanism of the jet (outflow) formation still eludes
the current theoretical concepts, in the TCAF paradigm the mass of outflow is
found to be depended upon the compression ratio of the gas of the two components
(Keplerian and sub-Keplerian) in the accretion flow. The outflow, in the low (hard)
state is a direct monotonic function of the ratio, which increases as the accretion
rate increases, till the state transition, where the outflow peaks and subsequently
falls (monotonically) in the high (soft) state.
$ In addition to the steady core jet, visible in the low (hard) state and quenched
or quiescent in the high (soft) state, the ejection of the central Compton cloud
in the during the high (soft) state causes flaring events (in the radio band). With
the accretion disc extending comparatively near the last stable orbit (due to high
accretion rate), the Compton cloud finds it difficult to exist in a steady state and
hence are ejected out, in bursts of minor and major flares.
$ The sequence of ejection of the central Compton clouds is brought to a stop by
a change in the X-ray spectra. In this phase, the spectral shape is that of the low
(hard) state, but the soft X-ray flux is the same as that of high (soft) state (as per
the prevelant classification schemes in the contemporary literature this may be
classified as the ‘very high state’). The exact physical features of this state is not
known, but tentatively one may speculate the onset of bulk motion Comptonization
in this phase, which prevents the formation of outflow (burst, flare etc.) with the
inflowing matter advected at great velocities into the event horizon.
$ Following this phase, the system most likely transits into the low (hard) state in a
span of few days (week or so), and the whole cycle starts again. But it has also
been seen to start the minor flaring episodes without going into the intervening
episodes without transiting into the low (hard) state.
The temporal analysis of the source, done for the first time in the post RXTE era,
yielded the following results:-
$ The binary modulation period as well as the template was found to be consis-
tent over period of 25 years. This rules out the possibility of any asymmetry in
the orbital trajectory (resulting in apsidal motion) in the system. Therefore, one
may consider some asymmetry in the X-ray cocoon (or X-ray halo) engulfing the
system, which is a site of reprocessing the X-ray emissions.
$ The residue of the (binary) folded lightcurve (Figure 4.2) may be attributed to the
long term variation of the X-ray emission by virtue of the change in wind and/or
cocoon mass distribution, including various state changes from soft (and high) to
low (and hard) and vice-versa, accompanied by the correlated radio flares.
$ At the peak of the binary ephemeris the X-ray emission is seen to fluctuate ran-
domly, this feature is unexplained and need a long term observation and analysis.
Given the smoothness of the template in the rising as well as the falling phase, and
the random fluctuation in the peak, consistent over decades, does suggest that the
randomness is inherent feature of the X-ray emission near the compact object, and
otherwise that region is eclipsed by the companion massive Wolf-Rayet star and
its wind.
$ The power density spectrum (PDS) shows a uniform smooth powerlaw behaviour,
irrespective of the state of X-ray emission, quite in contrast to the general patterns
in other sources of the same category. Although some periodic fluctuations have
been seen in the source in past, the Fourier spectrum of the lightcurves obtained
by RXTE–PCA does not reveal any (quasi) periodic structures.
$ The PDS merges with the white noise at a very low frequency of 0.1 Hz, in contrast
to the general behaviour of the Galactic X-ray binary systems. This may (or may
not) be due to the reprocessing of the emission from the inner disc by the cocoon
(or halo) engulfing the system.
$ The most interesting result of the timing analysis is the anti-correlated delay be-
tween the soft and hard X-rays in the low (hard) state. This feature needs to be
studied in detail and depth and may provide the foundation for a quantitative mod-
eling of the physical structure in the source.
7.1.2 Generalized picture of disc-jet connection in Galactic microquasars

The long term multi-wavelength study of Cygnus X-3 was extended to the other Galactic
microquasars, and the following unified results were obtained:-
$ A correlation exists between the soft X-ray and radio emission of GRS 1915+105

based on the data during the long state (associated to the low-hard state). The
hard X-ray emission is anti-correlated with both radio and soft X-rays. There is a
spectral pivoting at around 20 keV, correlated with the radio flux.
$ Comparing these results with those of Rau and Greiner (2003) who found a strong

correlation between radio emission and the X-ray spectral index in the states,
we conclude that the X-ray and radio emission characteristics of GRS 1915+105
are similar to those of Cygnus X-3. The only difference lies in the values of the
pivot energy of the X-ray spectra, which is around 12 keV in Cygnus X-3 and
around 20 keV in GRS 1915+105.
$ A three way correlation among soft X-ray, hard X-ray and radio emission has
been found in the low-hard state of Cyg X-1, confirming the results of Brocksopp
et al. (1999). Comparing this result with those of Zdziarski et al. (2002) who have
found that soft X-ray and hard X-ray above 100 keV are anti-correlated and also
that there is a spectral pivoting at around 50 – 90 keV, we conclude that the X-
ray:radio behavior of Cyg X-1 is similar to that of Cyg X-3 and GRS 1915+105,
but for the fact that the pivoting energy is at a higher value.
$ The X-ray:radio properties of Cyg X-1 are quite similar to that of GX 339-4,
where a 3-way correlation between soft X-ray, hard X-ray and radio emission has
been reported (Corbel et al. 2000, 2003). Though an anti-correlation between soft
X-ray/radio with hard X-rays has not been reported in this source, we note that

the X-ray spectrum during the low-hard state in this source too shows a pivoting
behavior at high energies 300 keV (Wardziński et al. 2002).
$ The radio emission is suppressed for Cygnus X-1 and GX 339-4 in their high-soft
state and similarly for Cygnus X-1 and GRS 1915+105 in their high states (with
associated softer spectra). Therefore, all these four sources with apparent diverse
X-ray and radio properties show very similar behavioural pattern encompassing
the long term steady non-flaring state.
$ Compiling the soft X-ray and radio observations of the above sources (GRS 1915+105,
Cyg X-3, and Cyg X-1) with the published correlation in GX 339-4 and V404 Cyg
(Gallo et al. 2002), we find that all the sources show a monotonic increase of radio
emission with the soft X-ray emission, spanning a 5 orders of magnitude variation
in their intrinsic luminosities. Cyg X-3 deviates from a single relation by about an
order of magnitude which can be reconciled if 1) the observed X-ray intensity is
an under-estimate because of obscuration and/or 2) the observed radio intensity is
an over-estimate because of beaming and Doppler boosting.
Fig. 7.1: ASM monitoring Cygnus X-3 during the period of GMRT observations.
$ If a common physical phenomena is responsible for such an uniform relation span-

ning across ‘off’ state to intermediate state, we argue that both radiations (X-ray
and radio) are unlikely to be originating from a single mechanism like synchrotron
emission.
$ Finally, we invoke a Two Component Advective Flow (TCAF) model (Chakrabarti
1996a) to explain the accretion-ejection behaviour in these systems in the steady
hard as well as soft states.
Thus, in this thesis we were able to provide a unified picture of the observational
properties and features and provide a phenomenological understanding of the accretion
- ejection mechanism of the Galactic microquasar systems, and also provide constraints
on the geometrical structure of the accretion process along the way, in accordance to the
L
‘aim’ specified in 1.6.
7.2 Future directions

To further the research and analysis process started in this thesis, the following steps
may be considered in continuing the mission of studying and understanding the Galactic
microquasar systems:-
$ Cygnus X-3. Simultaneous observation of the source in the radio, infra-red and
X-ray band, to unambiguously determine the jet energetics & the extent of jet
component in the X-ray emission, during the low-hard states.
$ Cygnus X-3. Simultaneous observation in radio and X-ray bands covering the
radio flaring events. Determine the complete physical structure.
$ Cygnus X-3. Detailed study of the soft X-ray emission to determine the repro-
cessing of the X-ray emission in the circum-stellar material.
$ Cygnus X-3. Quantitative application of accretion theories (TCAF or any other) to
explain the anti-correlated delay of hard X-rays w.r.t. soft X-rays, hence develop
a quantitative model for the accretion-ejection connection of the source.
$ Microquasars. A long term program of simultaneous observation in radio and
X-ray bands. Generalize the features of Cygnus X-3 to the other sources.
$ Microquasars & AGNs. Develop a unified picture of the accretion and ejection
across the scale of 6 orders of magnitude.
Some preliminary observations of Cygnus X-3 provided a lightcurve with the flux
varying in the range 55-70 mJy at 1.4 GHz. These observations were made during the
low state of the source. Therefore a groundwork has been made for the multi-wavelength
observations of the source using the observatory, and future detailed studies will unravel
the mystery of the accretion-ejection mechanism, and possibly the nature of the compact
object in the system.
Bibliography
Abramowicz, M. A. (1998). Theory of Black Hole Accretion Discs, ed.: M. A. Abramow-

icz, G. Bjornsson & J. E. Pringle. Cambridge Contemporary Astrophysics. Cambridge
University Press.
Abramowicz, M. A., B. Czerny, J. P. Lasota and E. Szuszkiewicz (1988). Slim accretion
disks. ApJ 332, 646–658.
Aller, H. and W. A. Dent (1972). Large outburst in Cygnus X-3 at 8 GHz. Nature Phys-
ical Science 239, 121–123.
Anderson, B., R. G. Conway, R. J. Davis, R. J. Peckham, P. J. Richards, R. E. Spencer
and P. N. Wilkinson (1972). Observations at 408 MHz of the Cyg X-3 Radio outburst.
Nature Physical Science 239, 117–118.
Arnaud, K. A. (1996). XSPEC: The First Ten Years. In: ASP Conf. Ser. 101: Astronom-
ical Data Analysis Software and Systems V. Vol. 5. pp. 17+.
Arnaud, K. and B. Dorman (2002). Xspec – An X-ray spectral fitting pack-
age, Users Guide for version 11.2.x. HEASARC, Laboratory for High En-
ergy Astrophysics NASA Greenbelt, MD 20771, available electronically at
http://heasarc.gsfc.nasa.gov/docs/software/lheasoft/xanadu.
Balbus, S. A. and J. F. Hawley (1991). A powerful local shear instability in weakly
magnetized disks. I - Linear analysis. II - Nonlinear evolution. ApJ 376, 214–233.
Barret, D. (2001). The broad band x-ray/hard x-ray spectra of accreting neutron stars.
Advances in Space Research 28, 307–321.
Barret, D. (2004). Accretion flows around stellar mass black holes and neutron stars.
ArXiv Astrophysics e-prints pp. astro--ph/0401100.
Barret, D., J. F. Olive and T. Oosterbroek (2003). Simultaneous BeppoSAX and Rossi
X-ray timing explorer observations of 4U 1812-12. A&A 400, 643–647.
Barret, D., J. F. Olive, L. Boirin, C. Done, G. K. Skinner and J. E. Grindlay (2000). Hard
X-Ray Emission from Low-Mass X-Ray Binaries. ApJ 533, 329–351.
Becker, R. H., J. L. Robinson-Saba, S. H. Pravdo, E. A. Boldt, S. S. Holt, P. J. Serlemit-
sos and J. H. Swank (1978). A 4.8 hour periodicity in the spectra of Cygnus X-3. ApJ
139
140 BIBLIOGRAPHY
224, L113–L117.
Becklin, E. E., G. Neugebauer, F. J. Hawkins, K. O. Mason, P. W. Sandford, K. Matthews
and C. G. Wynn-Williams (1973). Infrared and X-ray variability of Cygnus X-3. Na-
ture 245, 302–304.
Becklin, E. E., G. Neugebauer, F. J. Hawkins, K. O. Mason, P. W. Sandford,
K. Matthews, D. Packman, B. Schupler, A. Stark and C. G. Wynn-Williams (1974).
Infrared, radio, and X-ray observations of Cygnus X-3. ApJ 192, L119–L124.
Becklin, E. E., J. Kristian, G. Neugebauer and C. G. Wynn-Williams (1972). Discovery
of infrared emission from the radio source near Cygnus X-3. Nature Physical Science
239, 130–131.
Begelman, M. C., R. D. Blandford and M. J. Rees (1984). Theory of extragalactic radio
sources. review of Modern Physics 56, 255.
Belloni, T. (2002). GRS 1915+105: ten years after. In: New Views on Microquasars.
pp. 287–+.
Belloni, T., M. Klein-Wolt, M. Méndez, M. van der Klis and J. van Paradijs (2000). A
model-independent analysis of the variability of GRS 1915+105. A&A 355, 271–290.
Berger, M. and M. van der Klis (1994). High time resolutions observations of Cygnus
X-3 with EXOSAT. A&A 292, 175–182.
Bevington, P. R. (1969). Data Reduction and Error Analysis for the Physical Sciences.
McGraw-Hill Book Company.
Bjornsson, G., M. A. Abramowicz, X. Chen and J. P. Lasota (1996). Hot Accretion Disks
Revisited. ApJ 467, 99.
Blaes, O. (2003). Accretion discs, jets and high energy phenomena in astrophysics, ed.
V. Beskin, G. Henri, F. Menard, G. Pelletier and J. Dalibard. NATO Advanced Study
Institute 2003, Les Houches Session LXXVIII. EDP Sciences; Springer-Verlag, in
cooperation with NATO Scientific Affair Division.
Blandford, R. D. and M. C. Begelman (1999). On the fate of gas accreting at a low rate
on to a black hole. MNRAS 303, L1–L5.
Bleach, R. D., E. A. Boldt, S. S. Holt, D. A. Schwartz and P. J. Serlemitsos (1972).
X-Ray Spectra of Discrete Sources in Cygnus. ApJ 171, 51.
Blissett, C., , K. O. Mason and J. L. Culhane (1981). Ariel V observation of a 4.8-hr
periodicityin the ‘high-state’ X-ray spectrum of Cygnus X-3. MNRAS 194, 77–84.
Bolton, C. T. (1972). Identification of Cygnus X-1 with HDE 226868. Nature 235, 271–
273.
Bonnet-Bidaud, J. M. and G. Chardin (1988). Cygnus X-3, a critical review.. Physics
Reports 170, 326–404.
Bowyer, S., E. T. Byram, T. A. Chubb and H. Friedman (1965). Cosmic X-ray sources.
Science 147, 394–398.
BIBLIOGRAPHY 141
Braes, L. L. E. and G. K. Miley (1971). Detection of radio emission from Cygnus X-1.
Nature 232, 246.
Braes, L. L. E. and G. K. Miley (1972). Radio detection of Cygnus X-3. Nature 237, 506.
Branson, N. J. B. A., A. H. M. Martin, G. G. Pooley, A. C. S. Readhead, J. R. Shakeshaft,
A. Slingo and P. J. Warner (1972). Observations of Cygnus X-3 at the Mullard Radio
Observatory. Nature Physical Science 239, 133–34.
Brocksopp, C., R. P. Fender, V. Larionov, V. M. Lyuty, A. E. Tarasov, G. G. Pooley,
W. S. Paciesas and P. Roche (1999). Orbital, precessional and flaring variability of
Cygnus X-1. MNRAS 309, 1063–1073.
Castro-Tirado, A. J., S. Brandt and N. Lund (1992). GRS 1915+105. In: International
Astronomical Union Circular. pp. 2–+.
Chakrabarti, S. K. (1984). Active Galactic Nuclei, ed.: J. Dyson. Mancheser University
Press.
Chakrabarti, S. K. (1986). Rotating wind solution and the acceleration of the cosmic
radio jets. ApJ 303, 582–588.
Chakrabarti, S. K. (1990). Theory of transonic astrophysical flows. Singapore: World
Scientific Publication, 1990.
Chakrabarti, S. K. (1996a). Accretion processes on a black hole.. Physics Reports
266, 229–390.
Chakrabarti, S. K. (1996b). Global solutions of viscous transonic flows in Kerr geometry
- I. Weak viscosity limit. MNRAS 283, 325–+.
Chakrabarti, S. K. (1996c). Grand Unification of Solutions of Accretion and Winds
around Black Holes and Neutron Stars. ApJ 464, 664–+.
Chakrabarti, S. K. (1996d). Solutions of Two-dimensional Viscous Accretion and Winds
in Kerr Black Hole Geometry. ApJ 471, 237–+.
Chakrabarti, S. K. (1999). Estimation and effects of the mass outflow from shock com-
pressed flow around compact objects. A&A 351, 185–191.
Chakrabarti, S. K. (2000). Advective Accretion Flows: Ten Years Later. ArXiv Astro-
physics e-prints [astro-ph/0007254].
Chakrabarti, S. K. and I. Chattopadhyay (2002). Bulk motion comptonization - a sure
sign of black holes. In: The Ninth Marcel Grossmann Meeting. pp. 2253–+ [astro-
-ph/0012531].
Chakrabarti, S. K. and L. G. Titarchuk (1995). Spectral Properties of Accretion Disks
around Galactic and Extragalactic Black Holes. ApJ 455, 623–+.
Chakrabarti, S. K. and L. G. Titarchuk (1996). Spectral Properties of Accretion Disks
around Galactic and Extragalactic Black Holes: Erratum. ApJ 467, 474–+.
Chakrabarti, S. K. and P. Bhaskaran (1992). On the origin, acceleration and collimation
of bipolar outflows and cosmic radio jets. MNRAS 255, 255–260.
142 BIBLIOGRAPHY
Chakrabarti, S. K. and S. D’Silva (1994). Magnetic activity in thick accretion disks and
associated observable phenomena. 1: Flux expulsion. ApJ 424, 138–148.
Chakrabarti, S. K. and S. G. Manickam (2000). Correlation among Quasi-Periodic Os-
cillation Frequencies and Quiescent-State Duration in Black Hole Candidate GRS
1915+105. ApJ 531, L41–L44.
Chakrabarti, S. K., D. Titarchuk, L. Kazanas and K. Ebisawa (1996). Observational
signature of the ‘boundary layer’ of galactic and extragalactic black holes.. A&AS
120, C163+.
Chang, K. M. and J. P. Ostriker (1985). Don’t Know. ApJ 288, 428.
Chen, X., M. A. Abramowicz, J. P. Lasota, R. Narayan and I. Yi (1996). Unified descrip-
tion of accretion flows around black holes. ApJ 443, L61.
Choudhury, M., A. R. Rao, S. V. Vadawale, A. K. Jain and N. S. Singh (2004). Binary
corrected X-ray light curve of Cygnus X-3 : implications for the timing properties of
the compact binary system. A&A 420, 665–671.
Choudhury, M., A. R. Rao, S. V. Vadawale and A. K. Jain (2003). Correlated Radio:X-
Ray Emission in the Hard States of Galactic Microquasars. ApJ 593, 452–462.
Choudhury, M., A. R. Rao, S. V. Vadawale, C. H. Ishwara-Chandra and A. K. Jain
(2002). Disk-jet connection in Cygnus X-3. A&A 383, L35–L38.
Choudhury, M. and A. R. Rao (2002). X-ray Spectroscopy of Cygnus X-3. JApA 23, 39–
+.
Coppi, P. S. (1992). Time-dependent models of magnetized pair plasmas. MNRAS
258, 657–683.
Corbel, S., M. A. Nowak, R. P. Fender, A. K. Tzioumis and S. Markoff (2003). Radio/X-
ray correlation in the low/hard state of GX 339-4. A&A 400, 1007–1012.
Corbel, S., P. Kaaret, R. K. Jain, C. D. Bailyn, R. P. Fender, J. A. Tomsick, E. Kalemci,
V. McIntyre, D. Campbell-Wilson, J. M. Miller and M. L. McCollough (2001). X-
Ray States and Radio Emission in the Black Hole Candidate XTE J1550-564. ApJ
554, 43–48.
Corbel, S., R. P. Fender, A. K. Tzioumis, M. Nowak, V. McIntyre, P. Durouchoux and
R. Sood (2000). Coupling of the X-ray and radio emission in the black hole candidate
and compact jet source GX 339-4. A&A 359, 251–268.
Corongiu, A., L. Chiappetti, F. Haardt, A. Treves, M. Colpi and T. Belloni (2003). The
X-ray spectrum of the black hole candidate GX339-4 in a low state. A&A 408, 347–
352.
Czerny, B., M. Nikołajuk, M. Piasecki and J. Kuraszkiewicz (2001). Black hole masses
from power density spectra: determinations and consequences. MNRAS 325, 865–874.
D’Addario, L. R. and M. A. Stull (1972). Observations of Cygnus X-3 at 2.8 cm with 17
arc s beam. Nature Physical Science 239, 120–121.
BIBLIOGRAPHY 143
Das, T. K. and S. K. Chakrabarti (1999). Mass outflow rate from accretion discs aroung
compact objects. Classical and Quantum Gravity 16, 3879+–.
Davidsen, A. and J. P. Ostriker (1974). The Nature of Cygnus X-3: a Prototype for
Old-Population Binary X-Ray Sources. ApJ 189, 331–338.
Deeter, J. E. (1984). Techniques for the estimation of red power spectra. II Evaluation of
alternative methods. ApJ 281, 482–491.
Deeter, J. E. and P. E. Boynton (1982). Techniques for the estimation of red power
spectra. I - Context and methodology. ApJ 261, 337–350.
Dent, W. A. andKapitzky, J. E., B. G Leslie, G. Kojoian, M. L. Meeks, H. H. Danforth,
J. J. Kollasch, E. J Chaisson, D. F. Dickinson, L. E Goad and C. J. Lada (1972). 15.5
GHz observations at the Haystack observatory of the Cygnus X-3 outburst. Nature
Physical Science 239, 126–127.
Dhawan, V., I. F. Mirabel and L. F. Rodrı́guez (2000). AU-Scale Synchrotron Jets and
Superluminal Ejecta in GRS 1915+105. ApJ 543, 373–385.
Done, C. (2001). Galactic black hole binary systems. Advances in Space Research
28, 255–265.
Done, C. (2002). Accretion flows in X-ray binaries. Royal Society of London Philosoph-
ical Transactions Series A 360, 1967–+.
Done, C. and S. Nayakshin (2001a). Observational Signatures of X-Ray-irradiated Ac-
cretion Disks. ApJ 546, 419–428.
Done, C. and S. Nayakshin (2001b). Testing models of X-ray reflection from irradiated
discs. MNRAS 328, 616–622.
Done, C., J. S. Mulchaey, R. F. Mushotzky and K. A. Arnaud (1992). An ionized accre-
tion disk in Cygnus X-1. ApJ 395, 275–288.
D’Silva, S. and S. K. Chakrabarti (1994). Magnetic activity in thick accretion disks and
associated observable phenomena. 2: Flux storage. ApJ 424, 149–157.
Ebisawa, K., L. Titarchuk and S. K. Chakrabarti (1996). On the Spectral Slopes of Hard
X-Ray Emission from Black Hole Candidates. PASJ 48, 59–65.
Eggum, G. E., F. V. Coroniti and J. I. Katz (1985). Jet production in super-Eddington
accretion disks. ApJ 298, L41.
Elsner, R. F., P. Ghosh, W. Darbro, M. C. Weisskopf, P. G. Sutherland and J. E. Grindlay
(1980). Observations of Cygnus X-3 with the Einstein /HEAO 2/ X-ray Observatory
The period derivative and the asymmetric X-ray light curve. ApJ 239, 335–344.
Fender, R. (2001a). Energetics of jets from X-ray binaries. Astrophysics and Space Sci-
ence Supplement 276, 69–77.
Fender, R. (2003). Jets from X-ray Binaries. ArXiv Astrophysics e-prints [astro-
ph/0303339].
Fender, R. P. (2001b). Powerful jets from black hole X-ray binaries in low/hard X-ray
144 BIBLIOGRAPHY
states. MNRAS 322, 31–42.

Fender, R. P. and E. Kuulkers (2001). On the peak radio and X-ray emission from neu-
tron star and black hole candidate X-ray transients. MNRAS 324, 923–930.
Fender, R. P., G. G. Pooley, C. Brocksopp and S. J. Newell (1997a). Rapid infrared flares
in GRS 1915+105: evidence for infrared synchrotron emission. MNRAS 290, L65–
L69.
Fender, R. P., M. M. Hanson and G. G. Pooley (1999a). Infrared spectroscopic variability
of Cygnus X-3 in outburst and quiescence. MNRAS 308, 473–484.
Fender, R. P., S. J. Bell Burnell, E. B. Waltman, G. G. Pooley, F. D. Ghigo and R. S.
Foster (1997b). Cygnus X-3 in outburst: quenched radio emission, radiation losses
and variable local opacity. MNRAS 288, 849–858.
Fender, R. P., S. T. Garrington, D. J. McKay, T. W. B. Muxlow, G. G. Pooley, R. E.
Spencer, A. M. Stirling and E. B. Waltman (1999b). MERLIN observations of rela-
tivistic ejections from GRS 1915+105. MNRAS 304, 865–876.
Fishman, G. J., C. A. Meegan, R. B. Wilson, T. A. Parnell, W. S. Paciesas, G. N. Pendle-
ton, H. S. Hudson, J. L. Matteson, L. E. Peterson, T. L. Cline, B. J. Teegarden and B. E.
Schaefer (1989). The BATSE Experiment for the GRO - Solar Flare Hard X-Ray and
Gamma-Ray Capabilities. BAAS 21, 860–+.
Frank, J., A. R. King and D. J. Rine (1992). Accretion power in astrophys. Cambridge
University Press.
Fukue, J. (1982). Jets from a Geometrically Thick Disk. PASJ 34, 163.
Gallo, E., R. P. Fender and G. G. Pooley (2002). On the correlation between radio and
X-ray flux in low/hard state black holes. In: New Views on Microquasars. pp. 201–+.
Gallo, E., R. P. Fender and G. G. Pooley (2003). A universal radio-X-ray correlation in
low/hard state black hole binaries. MNRAS 344, 60–72.
Gary, B., E. T. Olsen and P. W. Rosenkranz (1972). Radio observations Cygnus X-3 and
the surrounding. Nature Physical Science 239, 128–130.
George, I. M. and A. C. Fabian (1991). X-ray reflection from cold matter in active galac-
tic nuclei and X-ray binaries. MNRAS 249, 352–367.
Ghosh, P., R. F. Elsner, M. C. Weisskopf and P. G. Sutherland (1981). The asymmetric
4.8 hour X-ray modulation of Cygnus X-3 - Model light curves and inferred orbital
parameters. ApJ 251, 230–245.
Giacconi, R., P. Gorenstein, H. Gursky and J. R. Waters (1967). An X-Ray Survey of the
Cygnus Region. ApJ 148, L119.
Gierliński, M., A. A. Zdziarski, J. Poutanen, P. S. Coppi, K. Ebisawa and W. N. Johnson
(1999). Radiation mechanisms and geometry of Cygnus X-1 in the soft state. MNRAS
309, 496–512.
Gierlinski, M., A. A. Zdziarski, C. Done, W. N. Johnson, K. Ebisawa, Y. Ueda, F. Haardt
BIBLIOGRAPHY 145
and B. F. Phlips (1997). Simultaneous X-ray and gamma-ray observations of CYG X-

1 in the hard state by GINGA and OSSE. MNRAS 288, 958–964.
Gregory, P. C., P. P. Kronberg, E. R. Seaquist, V. A. Hughes, A. Woodsworth, M. R.
Viner and D. Retallack (1972). Discovery of giant radio outburst from Cygnus X-3.
Nature 239, 440–443.
Greiner, J., F. J. Vrba, A. A. Henden, H. H. Guetter and C. B. Luginbuhl (2001a). K band
monitoring of GRS 1915+105 during 1999-2000. Astrophysics and Space Science
Supplement 276, 105–106.
Greiner, J., J. G. Cuby and M. J. McCaughrean (2001b). An unusually massive stellar
black hole in the Galaxy. Nature 414, 522–525.
Haardt, F. and L. Maraschi (1991). A two-phase model for the X-ray emission from
Seyfert galaxies. ApJ 380, L51–L54.
Haardt, F. and L. Maraschi (1993). X-ray spectra from two-phase accretion disks. ApJ
413, 507–517.
Hakkila, J., Ed.) (1990). BATSE software for the analysis of the gamma ray burst spatial
distribution.
Han, X. and R. M. Hjellming (1992). Radio observations of the 1989 transient event in
V404 Cygni (=GS 2023+338). ApJ 400, 304–314.
Hannikainen, D. C., R. W. Hunstead, D. Campbell-Wilson and R. K. Sood (1998).
MOST radio monitoring of GX 339-4. A&A 337, 460–464.
Hannikainen, D., D. Campbell-Wilson, R. Hunstead, V. McIntyre, J. Lovell, J. Reynolds,
T. Tzioumis and K. Wu (2001). XTE J1550-564: a superluminal ejection during the
September 1998 outburst. Astrophysics and Space Science Supplement 276, 45–48.
Harmon, B. A., C. A. Wilson, G. J. Fishman, V. Connaughton, W. Henze, W. S. Paciesas,
M. H. Finger, M. L. McCollough, M. Sahi, B. Peterson, C. R. Shrader, J. E. Grindlay
and D. Barret (2004). The Burst and Transient Source Experiment (BATSE) Earth
Occultation Catalog of Low-Energy Gamma-Ray Sources. ApJS pp. to appear (ApJS
preprint doi:10.1086/421940), astro--ph/0404453.
Hartmann, L. (1998). Accretion Processes in Star Formation. Cambridge Astrophysics
Series; 32. Cambridge University Press.
Hayashida, K., S. Miyamoto, S. Kitamoto, H. Negoro and H. Inoue (1998). Central
Black Hole Masses in Active Galactic Nuclei Inferred from X-Ray Variability. ApJ
500, 642–+.
Heindl, W. A., J. A. Tomsick, R. Wijnands and D. M. Smith (2003). Extended Emission
from Cygnus X-3 Detected with Chandra. ApJ 588, L97–L100.
Herrero, A., R. P. Kudritzki, R. Gabler, J. M. Vilchez and A. Gabler (1995). Fundamental
parameters of galactic luminous OB stars. II. A spectroscopic analysis of HDE 226868
and the mass of Cygnus X-1.. A&A 297, 556–+.
146 BIBLIOGRAPHY
Hertz, P., P. C. Joss and S. Rappaport (1978). X-ray transfer in binary systems - A Monte
Carlo study. ApJ 224, 614–624.
Hjellming, R. M. and B. Ballick (1972). Unusual radio events in Cygnus X-3. Nature
239, 443–446.
Hjellming, R. M. and K. J. Johnston (1988). Radio emission from conical jets associated
with X-ray binaries. ApJ 328, 600–609.
Hjellming, R. M. and M. P. Rupen (1995). Episodic Ejection of Relativistic Jets by the
X-Ray Transient GRO:J1655-40. Nature 375, 464–+.
Hjellming, R. M., M. Hermann and E. Webster (1972). Radio observations of Cygnus
X-3. Nature 237, 507.
Homan, J., R. Wijnands, M. van der Klis, T. Belloni, J. van Paradijs, M. Klein-Wolt,
R. Fender and M. Méndez (2001). Correlated X-Ray Spectral and Timing Behavior
of the Black Hole Candidate XTE J1550-564: A New Interpretation of Black Hole
States. ApJS 132, 377–402.
Hynes, R. I., D. Steeghs, J. Casares, P. A. Charles and K. O’Brien (2003). Dynamical
Evidence for a Black Hole in GX 339-4. ApJ 583, L95–L98.
Jahoda, K., J. H. Swank, A. B. Giles, M. J. Stark, T. Strohmayer, W. Zhang and E. H.
Morgan (1996). In-orbit performance and calibration of the Rossi X-ray Timing Ex-
plorer (RXTE) Proportional Counter Array (PCA). In: Proc. SPIE Vol. 2808, p. 59-70,
EUV, X-Ray, and Gamma-Ray Instrumentation for Astronomy VII, Oswald H. Sieg-
mund; Mark A. Gummin; Eds.. pp. 59–70.
Johnson, W. N., J. D. Kurfess, W. R. Purcell, S. M. Matz, M. P. Ulmer, M. S. Strickman,
R. J. Murphy, D. A. Grabelsky, R. L. Kinzer, G. H. Share, R. A. Cameron, R. A.
Kroeger, M. Maisack, G. V. Jung, C. M. Jensen, D. D. Clayton, M. D. Leising, J. E.
Grove and C. S. Dyer (1993). Initial results from OSSE on the Compton Observatory.
A&AS 97, 21–25.
Kato, Y., M. R. Hayashi and R. Matsumoto (2004a). Formation of Semirelativistic Jets
from Magnetospheres of Accreting Neutron Stars: Injection of Hot Bubbles into a
Magnetic Tower. ApJ 600, 338–342.
Kato, Y., S. Mineshige and K. Shibata (2004b). Magnetohydrodynamic Accretion Flows:
Formation of Magnetic Tower Jet and Subsequent Quasi-Steady State. ApJ 605, 307–
320.
Kawashima, K. and S. Kitamoto (1996). Photoionized Plasma in Cygnus X-3. PASJ
48, L113–L116.
Kazanas, D. and D. C. Ellison (1986). The central engine of quasars and active galactic
nuclei Hadronic interactions of shock-accelerated relativistic protons. ApJ 304, 178.
Kitamoto, S., H. Tsunemi, H. Pedersen, S. A. Ilovaisky and M. van der Klis (1990).
Outburst, identification, and X-ray light curve of GS 1354 - 64 (= MX 1353 - 64?,
BIBLIOGRAPHY 147
Centaurus X-2?). ApJ 361, 590–595.

Kitamoto, S., K. Kawashima, H. Negoro, S. Miyamoto, N. E. White and F. Nagase
(1994). Resolving the Cygnus X-3 iron K line. PASJ 46, L105–L108.
Kitamoto, S., S. Miyamoto and W. Matsui (1987). 4.8-Hour modulation of X-rays from
Cygnus X-3. PASJ 391, 259–285.
Knoll, G. F. (2000). Radiation Detection and Measurement. John Wiley & Sons, Inc.
Kompaneets, A. S. (1957). Soviet Physics 4, 730.
Kong, A. K. H., J. E. McClintock, M. R. Garcia, S. S. Murray and D. Barret (2002). The
X-Ray Spectra of Black Hole X-Ray Novae in Quiescence as Measured by Chandra.
ApJ 570, 277–286.
Konigl, A. (1989). Self-similar models of magnetized accretion disks. ApJ 342, 208.
Kuncic, Z., A. Celotti and M. J. Rees (1997). Dense, thin clouds and reprocessed radia-
tion in the central regions of active galactic nuclei. MNRAS 284, 717–730.
Lampton, M., B. Margon and S. Bowyer (1976). Parameter estimation in X-ray astron-
omy. ApJ 208, 177–190.
Lauque, R., J. Lequeux and Nguyen-Quang-Rieu (1972). Determination of the distance
of Cygnus X-3 by 21-cm absorption. Nature Physical Science 239, 119–120.
Leach, R. W., S. S. Murray, E. J. Schreier, H. D. Tananbaum, M. P. Ulmer and D. R.
Parsignault (1975). Further observations of Cygnus X-3 with the UHURU satellite.
ApJ 199, 184–188.
Leahy, D. A., W. Darbro, R. F. Elsner, M. C. Weisskopf, S. Kahn, P. G. Sutherland
and J. E. Grindlay (1983). On searches for pulsed emission with application to four
globular cluster X-ray sources - NGC 1851, 6441, 6624, and 6712. ApJ 266, 160–170.
Levine, A. M., W. Cui, R. Remillard, H. Bradt, D. A. Smith, R. Shirey and E. H. Morgan
(1996). RXTE ASM Characterization of Temporal/Spectral Behavior of Galactic X-
ray Sources. Bulletin of the American Astronomical Society 28, 1316–+.
Lightman, A. P. and A. A. Zdziarski (1987). Pair production and Compton scattering
in compact sources and comparison to observations of active galactic nuclei. ApJ
319, 643–661.
Lightman, A. P. and T. R. White (1988). Effects of cold matter in active galactic nuclei -
A broad hump in the X-ray spectra. ApJ 335, 57–66.
Longair, M. S. (1994). High Energy Astrophysics. 1994 ed.. Cambridge University Press.
Macklin, J. T. (1982). An investigation of the properties of double radio sources using
the Spearman partial rank correlation coefficient. MNRAS 199, 1119–1136.
Magdziarz, P. and A. A. Zdziarski (1995). Angle-dependent Compton reflection of X-
rays and gamma-rays. MNRAS 273, 837–848.
Markert, T. H., C. R. Canizares, G. W. Clark, W. H. G. Lewin, H. W. Schnopper and
G. F. Sprott (1973a). Observations of the Highly Variable X-Ray Source GX 339-4.
148 BIBLIOGRAPHY
ApJ 184, L67+.

Markert, T. H., G. W. Clark, W. H. G. Lewin, H. W. Schnopper and G. F. Sprott (1973b).
Highly variable X-ray source.. In: International Astronomical Union Circular. pp. 1–
+.
Markoff, S., H. Falcke and R. Fender (2001). A jet model for the broadband spectrum of
XTE J1118+480. Synchrotron emission from radio to X-rays in the Low/Hard spectral
state. A&A 372, L25–L28.
Markoff, S., M. Nowak, S. Corbel, R. Fender and H. Falcke (2003). Exploring the role
of jets in the radio/X-ray correlations of GX 339-4. A&A 397, 645–658.
Martı́, J., J. M. Paredes and M. Peracaula (2001). Development of a two-sided relativistic
jet in Cygnus X-3. A&A 375, 476–484.
McCollough, M. L., C. R. Robinson, S. N. Zhang, B. A. Harmon, R. M. Hjellming,
E. B. Waltman, R. S. Foster, F. D. Ghigo, M. S. Briggs, G. N. Pendleton and K. J.
Johnston (1999). Discovery of Correlated Behavior between the Hard X-Ray and the
Radio Bands in Cygnus X-3. ApJ 517, 951–955.
Messina, D. C., R. A. Cameron, W. N. Johnson, R. A. Kroeger, J. D. Kurfess, M. S.
Strickman, C. H. Starr, D. A. Grabelsky, S. M. Matz, W. R. Purcell and M. P. Ul-
mer (1992). Data Types, Reduction Techniques, and Analysis Tools for the Comp-
ton Observatory OSSE Instrument.. Bulletin of the American Astronomical Society
24, 1194–+.
Milgrom, M. (1976). A cocooned X-ray binary model for CYG x-3. A&A 51, 215–218.
Mirabel, I. F. and L. F. Rodriguez (1994). A Superluminal Source in the Galaxy. Nature
371, 46–+.
Mirabel, I. F. and L. F. Rodriguez (1998). Microquasars in our Galaxy.. Nature 392, 673–
676.
Mirabel, I. F., L. F. Rodriguez, B. Cordier, J. Paul and F. Lebrun (1992). A double-
sided radio jet from the compact Galactic Centre annihilator 1E140.7 - 2942. Nature
358, 215–217.
Molnar, L. A. (1988). Interpretation of the period derivative of Cygnus X-3. ApJ
331, L25–L28.
Molnar, L. A. and C. W. Mauche (1986). Effects of the X-ray scattering halo on the
observational properties of Cygnus X-3. ApJ 310, 343–353.
Molteni, D., D. Ryu and S. K. Chakrabarti (1996). Numerical Simulations of Standing
Shocks in Accretion Flows around Black Holes: A Comparative Study. ApJ 470, 460–
+.
Molteni, D., G. Lanzafame and S. K. Chakrabarti (1994). Simulation of thick accretion
disks with standing shocks by smoothed particle hydrodynamics. ApJ 425, 161–170.
Muno, M. P., R. A. Remillard, E. H. Morgan, E. B. Waltman, V. Dhawan, R. M. Hjellm-
BIBLIOGRAPHY 149
ing and G. Pooley (2001). Radio Emission and the Timing Properties of the Hard
X-Ray State of GRS 1915+105. ApJ 556, 515–532.
Naik, S., P. C. Agrawal, A. R. Rao, B. Paul, S. Seetha and K. Kasturirangan (2001).
Detection of a Series of X-Ray Dips Associated with a Radio Flare in GRS 1915+105.
ApJ 546, 1075–1085.
Nakamura, H., M. Matsuoka, N. Kawai, A. Yoshida, S. Miyoshi, S. Kitamoto and K. Ya-
mashita (1993). Unified model fitting to variable X-ray spectra of Cygnus X-3. MN-
RAS 261, 353–365.
Nandi, A., S. K. Chakrabarti, S. V. Vadawale and A. R. Rao (2001). Ejection of the inner
accretion disk in GRS 1915+105: The magnetic rubber-band effect. A&A 380, 245–
250.
Narayan, R. and I. Yi (1995). Advection-dominated Accretion: Underfed Black Holes
and Neutron Stars. ApJ 452, 710–+.
Newell, S. J., M. A. Garrett and R. E. Spencer (1998). Apparent superluminal expansion
in Cygnus X-3. MNRAS 293, L17–L22.
Paerels, F., J. Cottam, M. Sako, D. A. Liedahl, A. C. Brinkman, R. L. J. van der Meer,
J. S. Kaastra and P. Predehl (2000). High-Resolution Spectroscopy of the X-Ray-
photoionized Wind in Cygnus X-3 with the Chandra High-Energy Transmission Grat-
ing Spectrometer. ApJ 533, L135–L138.
Papadakis, I. E. and A. Lawrence (1993). Improved Methods for Power Spectrum Mod-
elling of Red Noise. MNRAS 261, 612–+.
Parker, E. N. (1960). The Hydrodynamic Treatment of the Expanding Solar Corona..
ApJ 132, 175–+.
Parsignault, D. R., H. Gursky, E. M. Kellogg, T. Matilsky, S. Murray, E. Schreier,
H. Tananbaum and Giaconni R. (1972). Observation of Cygnus X-3 by Uhuru. Na-
ture Physical Science 239, 123–125.
Pomphrey, R. B. and E. E. Epstein (1972). Cygnus X-3: 3.3 mm observations. Nature
Physical Science 239, 125–126.
Pooley, G. G., R. P. Fender and C. Brocksopp (1999). Orbital modulation and longer
term variability in the radio emission from Cygnus X-1. MNRAS 302, L1–L5.
Poutanen, J. (1998). Theory of Black Hole Accretion Discs, ed.: M. A. Abramowicz,
G. Bjornsson & J. E. Pringle. Cambridge Contemporary Astrophysics. Cambridge
University Press.
Poutanen, J. and P. S. Coppi (1998). Unification of Spectral States of Accreting Black
Holes. Physica Scripta T77, 57–59.
Poutanen, J. and R. Svensson (1996). The Two-Phase Pair Corona Model for Active
Galactic Nuclei and X-Ray Binaries: How to Obtain Exact Solutions. ApJ 470, 249–
+.
150 BIBLIOGRAPHY
Predehl, P., V. Burwitz, F. Paerels and J. Trümper (2000). Chandra measurement of the
geometrical distance to Cyg X-3 using its X-ray scattering halo. A&A 357, L25–L28.
Pringle, J. E. (1974). Discovery of giant radio outburst from Cygnus X-3. Nature
247, 21–22.
Rajeev, M. R., V. R. Chitnis, A. R. Rao and K. P. Singh (1994). Generalized spectra
model for 1-100 keV X-ray emission from Cygnus X-3 based on EXOSAT data. ApJ
424, 376–394.
Rao, A. R., J. S. Yadav and B. Paul (2000). Rapid State Transitions in the Galactic Black
Hole Candidate Source GRS 1915+105. ApJ 544, 443–452.
Rao, A. R., P. C. Agrawal and R. K. Manchanda (1991). Hard X-ray observations of
Cygnus X-3. A&A 241, 127–130.
Rau, A. and J. Greiner (2003). Comptonization and reflection of X-ray radiation and the
X-ray-radio correlation in the chi -states of GRS 1915+105. A&A 397, 711–722.
Reig, P., I. Papadakis and N. D. Kylafis (2002). The aperiodic variability of Cyg X-1 and
GRS 1915+105 at very low frequencies. A&A 383, 202–209.
Rodrı́guez, L. F. and I. F. Mirabel (1999). Repeated Relativistic Ejections in GRS
1915+105. ApJ 511, 398–404.
Rodriguez, L. F., I. F. Mirabel and J. Marti (1992). The radio counterpart of the hard
X-ray source GRS 1758-258. ApJ 401, L15–L18.
Ross, R. R., A. C. Fabian and A. J. Young (1999). X-ray reflection spectra from ionized
slabs. MNRAS 306, 461–466.
Rothschild, R. E., P. R. Blanco, D. E. Gruber, W. A. Heindl, D. R. MacDonald, D. C.
Marsden, M. R. Pelling, L. R. Wayne and P. L. Hink (1998). In-Flight Performance
of the High-Energy X-Ray Timing Experiment on the Rossi X-Ray Timing Explorer.
ApJ 496, 538–+.
Rybicki, G. B. and A. P. Lightman (1979). Radiative processes in astrophysics. New
York: A Wiley-Interscience Pub., 393 p.
Ryu, D., S. K. Chakrabarti and D. Molteni (1997). Zero-Energy Rotating Accretion
Flows near a Black Hole. ApJ 474, 378–+.
Samimi, J., G. H. Share, K. Wood, D. Yentis, J. Meekins, W. D. Evans, S. Shulman, E. T.
Byram, T. A. Chubb and H. Friedman (1979). GX339-4 - A new black hole candidate.
Nature 278, 434–436.
Sandford, P. W. and F. H. Hawkins (1972). X-ray observations of Cygnus X-3 by Coper-
nicus. Nature Physical Science 239, 135.
Sandford, P. W., K. O. Mason and J. Ives (1975). Observation of a line feature in the
X-ray spectrum of Cygnus X-3. MNRAS 173, Short Commun. 9P–14P.
Schmutz, W., T. R. Geballe and H. Schild (1996). CYG X-3: Evidence for a Black Hole..
A&A 311, L25–L28.
BIBLIOGRAPHY 151
Serlemitsos, P. J., E. A. Boldt, S. S. Holt, R. E. Rothschild and J. L. R. Saba (1975).

Spectral variability of Cygnus X-3. ApJ 201, L9–L13.
Shakura, N. I. and R. A. Sunyaev (1973). Black holes in binary systems. Observational
appearance.. A&A 24, 337–355.
Shapiro, S. L., A. P. Lightman and D. M. Eardley (1976). A two-temperature accretion
disk model for Cygnus X-1 - Structure and spectrum. ApJ 204, 187–199.
Singh, N. S., S. Naik, B. Paul, P. C. Agrawal, A. R. Rao and K. Y. Singh (2002). New
measurements od Orbital period change in Cygnus X-3. A&A 392, 161–167.
Sood, R. and D. Campbell-Wilson (1994). GX 339-4. In: International Astronomical
Union Circular. pp. 1–+.
Stark, M. J. and M. Saia (2003). Doppler Modulation of X-Ray Lines in Cygnus X-3.
ApJ 587, L101–L104.
Stern, B. E., J. Poutanen, R. Svensson, M. Sikora and M. C. Begelman (1995). On the
Geometry of the X-Ray–Emitting Region in Seyfert Galaxies. ApJ 449, L13+.
Stirling, A. M., R. E. Spencer, C. J. de la Force, M. A. Garrett, R. P. Fender and R. N.
Ogley (2001). A relativistic jet from Cygnus X-1 in the low/hard X-ray state. MNRAS
327, 1273–1278.
Sunyaev, R. A. and L. G. Titarchuk (1980). Comptonization of X-rays in plasma clouds
- Typical radiation spectra. A&A 86, 121–138.
Sunyaev, R. A. and L. G. Titarchuk (1985). Comptonization of low-frequency radia-
tion in accretion disks Angular distribution and polarization of hard radiation. A&A
143, 374–388.
Sunyaev, R. and M. Revnivtsev (2000). Fourier power spectra at high frequencies: a way
to distinguish a neutron star from a black hole. A&A 358, 617–623.
Tanaka, Y. (2000). Observation of Black Holes in X-ray Binaries. In: IAU Symp. 195:
Highly Energetic Physical Processes and Mechanisms for Emission from Astrophysi-
cal Plasmas. pp. 37–+.
Tanaka, Y. and W. H. G. Lewin (1995). X-ray Binaries, ed. W. H. G. Lewin, Van Paradijs,
and E. P. J. van den Heuvel. Cambridge Astrophysics Series. Cambridge University
Press.
Tingay, S. J., D. L. Jauncey, R. A. Preston, J. E. Reynolds, D. L. Meier, D. W. Murphy,
A. K. Tzioumis, D. J. McKay, M. J. Kesteven, J. E. J. Lovell, D. Campbell-Wilson,
S. P. Ellingsen, R. Gough, R. W. Hunstead, D. L. Jones, P. M. McCulloch, V. Migenes,
J. Quick, M. W. Sinclair and D. Smits (1995). Relativistic Motion in a Nearby Bright
X-Ray Source. Nature 374, 141–+.
Titarchuk, L. (1994). Generalized Comptonization models and application to the recent
high-energy observations. ApJ 434, 570–586.
Vadawale, S. V. (2003). Study of Superluminal X-ray Sources. Ph.D. Thesis.
152 BIBLIOGRAPHY
Vadawale, S. V., A. R. Rao, A. Nandi and S. K. Chakrabarti (2001a). Observational

evidence for mass ejection during soft X-ray dips in GRS 1915+105. A&A 370, L17–
L21.
Vadawale, S. V., A. R. Rao and S. K. Chakrabarti (2001b). Spectral differences between
the radio-loud and radio-quiet low-hard states of GRS 1915+105: Possible detection
of synchrotron radiation in X-rays. A&A 372, 793–802.
Vadawale, S. V., A. R. Rao and S. Naik (2002). X-ray and radio emission from GRS
1915+105: the disk-jet connection. In: New Views on Microquasars. pp. 344–+.
Vadawale, S. V., A. R. Rao, S. Naik, J. S. Yadav, C. H. Ishwara-Chandra, A. Pramesh
Rao and G. G. Pooley (2003). On the Origin of the Various Types of Radio Emission
in GRS 1915+105. ApJ 597, 1023–1035.
van der Klis, M. (1989). Quasi-periodic oscillations and noise in low-mass X-ray bina-
ries. ARA&A 27, 517–553.
van der Klis, M. (1994). A comparison of the power spectra of Z and atoll sources,
pulsars and black hole candidates. A&A 283, 469–474.
van der Klis, M. (1997). Quantifying Rapid Variability in Accreting Compact Objects.
In: Statistical Challenges in Modern Astronomy II. pp. 321–+.
van der Klis, M. (2000). Millisecond Oscillations in X-ray Binaries. ARA&A 38, 717–
760.
van der Klis, M. and F. A. Jansen (1985). Transient quasi-periodic oscillations in the
X-ray flux of Cygnus X-3. Nature 313, 768–771.
van der Klis, M. and J. M. Bonnet-Bidaud (1981). A change in light curve asymmetry
and the ephemeris of Cygnus X-3. A&A 95, L5–L7.
van der Klis, M. and J. M. Bonnet-Bidaud (1982). The cycle-to-cycle variability of
Cygnus X-3. A&AS 50, 129–140.
van der Klis, M. and J. M. Bonnet-Bidaud (1989). The X-ray ephemeris of Cygnus X-3.
A&A 214, 203–208.
van Kerkwijk, M. H. (1993). Spectroscopic and Photometric Variability of CYGNUS-
X-3. A&A 276, L9+.
van Kerkwijk, M. H., P. A. Charles, T. R. Geballe, D. L. King, G. K. Miley, L. A.
Molnar, E. P. J. van den Heuvel, M. van der Klis and J. van Paradijs (1992). Infrared
helium emission lines from Cygnus X-3 suggesting a Wolf-Rayet star companion.
Nature 355, 703–705.
van Kerkwijk, M. H., T. R. Geballe, D. L. King, M. van der Klis and J. van Paradijs
(1996). The Wolf-Rayet counterpart of Cygnus X-3.. A&A 314, 521–540.
Wagner, R. M., S. G. Starrfield, R. M. Hjellming, S. B. Howell and T. J. Kreidl (1994).
ROSAT observations of the black hole candidate V404 Cygni in quiescence. ApJ
429, L25–L28.
BIBLIOGRAPHY 153
Waltman, E. B., F. D. Ghigo, K. J. Johnston, R. S. Foster, R. L. Fiedler and J. H. Spencer

(1995). The Evolution of Outbursts in Cygnus X-3 at 2.25 and 8.3 GHz. AJ 110, 290–
+.
Waltman, E. B., R. S. Foster, G. G. Pooley, R. P. Fender and F. D. Ghigo (1996).
Quenched Radio Emission in Cygnus X-3. AJ 112, 2690–+.
Wardziński, G., A. A. Zdziarski, M. Gierliński, J. Eric Grove, K. Jahoda and W. Neil
Johnson (2002). X-ray and 0 -ray spectra and variability of the black hole candidate
GX 339-4. MNRAS 337, 829–839.
Watanabe, H., S. Kitamoto, S. Miyamoto, R. L. Fielder, E. B. Waltman, K. J. Johnston
and F. D. Ghigo (1994). Correlation between X-ray intensity and radio outbursts of
Cygnus X-3. ApJ 433, 350–356.
Watarai, K. and S. Mineshige (2001). Slim Disk: Viscosity Prescriptions and Observa-
tional Implications. PASJ 53, 915–921.
Webster, B. L. and P. Murdin (1972). Cygnus X-1 – a Spectroscopic binary with a heavy
companion?. Nature 235, 37–38.
White, N. E. and S. S. Holt (1982). Accretion disk coronae. ApJ 257, 318–337.
White, T. R., A. P. Lightman and A. A. Zdiziarski (1988). Compton reflection of gamma
rays by cold electrons. ApJ 331, 939–948.
Zdziarski, A. A. (2000). Radiative Processes and Geometry of Spectral States of Black-
hole Binaries. In: IAU Symp. 195: Highly Energetic Physical Processes and Mecha-
nisms for Emission from Astrophysical Plasmas. pp. 153–+.
Zdziarski, A. A., J. E. Grove, J. Poutanen, A. R. Rao and S. V. Vadawale (2001). OSSE
and RXTE Observations of GRS 1915+105: Evidence for Nonthermal Comptoniza-
tion. ApJ 554, L45–L48.
Zdziarski, A. A., J. Poutanen, W. S. Paciesas and L. Wen (2002). Understanding the
Long-Term Spectral Variability of Cygnus X-1 with Burst and Transient Source Ex-
periment and All-Sky Monitor Observations. ApJ 578, 357–373.
Zycki, P. T., C. Done and D. A. Smith (1998). Evolution of the Accretion Flow in Nova
MUSCAE 1991. ApJ 496, L25+.
Zycki, P. T., C. Done and D. A. Smith (1999). X-ray spectral evolution of GS 2023+338
(V404 Cyg) during decline after outburst. MNRAS 305, 231–240.

A Thesis On Microquasars

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Thesis On Microquasars

Uploaded by

Copyright:

Available Formats

Hard X-ray and Gamma Ray

Properties of Cosmic Sources

Ph. D. (Science) Degree

Under the Guidance of

Tata Institute of Fundamental Research

First and foremost, I wholeheartedly acknowledge the institutional structure, geograph-

host to me in Boston during my participation in the international meeting on “X-ray Tim-

2 X-ray detectors and techniques of instrumentation; Radio astronomy 26

2.3.3 Spectral analysis . . . . . . . . . . . . . . . . . . . . . . . . . 42

3 Cygnus X-3: spectral studies 49

4 Cygnus X-3: temporal studies 73

5 Disc-jet connection in microquasars: low (hard) states 94

6 Two Component Accretion Flow model 117

7 Summary and conclusions 131

An X-ray is a quantum of electromagnetic radiation with an energy, to an order of magni-

emission is presented to provide a coherent picture of the systems.

Chapter-wise organization of the thesis:-

An X-ray is a quantum of electromagnetic radiation with an energy, to an order of magni-

1.1 High energy physical processes and phenomena

body radiation, bremsstrahlung, Compton scattering. Soon, however, it was discovered,

Thermal bremsstrahlung. Free-free emission occurs with the interaction (accelera-

Synchrotron radiation. The process of radiation due to acceleration of charged par-

Comptonization. The scattering of an energetic photon with a stationary electron is

only possible in environment with extremely high temperature or with extremely

1.1.1 Accretion disc as a source of high energy emission

It should be noted that this is derived assuming spherically symmetric geometry. It

1.2 Black hole sources: stellar mass and supermassive

sources. These were subsequently identified to be accreting supermassive black holes

1.3 Microquasars: general properties and behaviour

X-ray temporal characteristics:-

X-ray spectral characteristics:-

1.3.1 Transient X-ray blackhole binaries

1.3.2 Persistent sources: canonical states of X-ray emission

1.4 Accretion in X-ray binaries

eter. Little did they realize the forthcoming impact of this Õ

commonly used modification of the Shakura-Sunyaev : (SS) prescription: is to assume a

ûoüþýaû , with maximal temperature possibly as high as û X

log τ (vertical optical depth through the disk)

1.4.1 Unification of hydrodynamic solutions of accretion flow

at a fixed radius of the accretion disc, are as follows:-

higher energies than electrons, where radiative processes include bremsstrahlung,

1.4.2 Hard X-ray emission models

1.4.3 Geometrical structure of the accretion system

Fig. 1.8: The geometry for the sombrero

1.5 Outflows from microquasars

1.6 Aim of this thesis

X-ray detectors and techniques of

2.1 X-ray detectors

Semi-conductors (solid state detectors). The principle of operation of semi-conductor

the efficiency of creating electron-hole pairs in a semi-conductor detector is about ten

2.2 Modern X-ray observatories

Proportional Counter Array (PCA). Developed at GSFC, it consists of : a system of

2.2.2 Compton Gamma Ray Observatory (CGRO)

Oriented Scintillation Spectrometer Experiment (OSSE) The Oriented Scintilla-

2.2.3 Other notable X-ray missions

2.3 X-ray astronomical data analyses and techniques

2. Analysis of these standard products to understand the physical mechanism and

2.3.1 RXTE data analysis

2.3.2 Timing analysis

timing analysis in this thesis.

2.3.3 Spectral analysis

A spectrum is the distribution of flux as a function of energy. The understanding of

Calibration issues. The response matrix of a given instrument, in principle, depends

from this source to be k ,(