Multi-Wavelength Studies of X-Ray Binary Systems

MULTI-WAVELENGTH STUDIES OF X-RAY BINARY SYSTEMS

THESIS Submitted for the degree of DOCTOR OF PHILOSOPHY (PHYSICS)

to KUMAUN UNIVERSITY, NAINITAL

by Ramanpreet Kaur

Aryabhatta Research Institute of observational sciencES (ARIES) Manora Peak, Nainital 263 129, India

April 2009 i

DECLARATION

I hereby declare that the work presented in this thesis is a result of the investigation carried out by me at the Aryabhatta Research Institute of observational sciencES (ARIES), Nainital, under the supervision of Prof. Ram Sagar (Aryabhatta Research Institute of observational sciencES, Nainital). This thesis has not been submitted before for the award of any degree, diploma, associateship or fellowship of any University or Institute.

Place : Nainital Date : (Ramanpreet Kaur) ii

CERTIFICATE FROM THE SUPERVISOR

This is to certify that

1. The synopsis of the thesis entitled “Multi-wavelength studies of X-ray binary systems” for the award of the degree of Doctor of Philosophy in Physics was ap- proved by the Kumaun University, Nainital.

2. This thesis embodies the work of Mrs Ramanpreet Kaur herself.

3. Mrs Ramanpreet Kaur worked under my supervision for this thesis as a Research Fellow at the Aryabhatta Research Institute of observational sciencES (ARIES), Nainital. She has put in more than 200 days of attendance at ARIES, Nainital during this period.

4. This thesis has not been submitted before for the award of any degree, diploma, associateship or fellowship of any University or Institute.

Ram Sagar Place: Nainital ARIES Date: Nainital - 263 129 iii

To, My family iv Acknowledgments

During the slow and often interrupted evolution of this thesis I have accumulated many debts, only a portion of which I have space to acknowledge here. I owe a sincere thanks to my thesis supervisor Prof. Ram Sagar for his expert guidance and constant support. I highly acknowledge the freedom he gave me to work on research problems of my interest and his encouragements to make new collaborations in India as well as abroad. I express my sincere thanks to Biwsajit for his expert guidance throughout my PhD. His enthusiasm towards research always inspired me to enjoy every bit of my research work and gave me the courage to finish things before time. I am thankful to him for his collaboration and looking forward to many years of collaborations with him in future too. I would like to acknowledge the debt I owe to my collaborator Brijesh Kumar for making me acquainted with the optical data reduction techniques, scientific discussions and help in the thesis draft. I deeply acknowledge his encouragements and motivations throghout my PhD. Part of my research work presented in this thesis was carried out during my visit to the University of Amsterdam. Inspite of being it an unofficial visit from my home institute, Ralph, Michiel and Rudy helped me in all possible ways to continue my thesis work. I am deeply indebted to all of them. I am particularly grateful to Rudy for giving me an opportunity to work with him during my visit and making me a part of the XRT group. I am also thankful to him and Michiel for arranging finances for me to attend COSPAR in Montreal. My many thanks to Ale and Diego for their help and the scientific discussions we had during my visit. A number of friends in Amsterdam who made my visit comfortable and memorable one include Anna, Dipankar, Rachel, Manu, PG, Paolo, Nanda and Nathalie. It was nice to share an office with Prof. Wim Hermsen in UvA and I am thankful to him for all the discussions we had during my visit. I am also thankful to Wim for his help getting my visa for Canada. I would like to acknowledge the necessary help provided by Eva, Minou and Lide during my visit. I owe my sincere thanks to Prof. Dipankar Bhattacharya who gave me the necessary enthusiasm and motivation to do research and to Sunita for her encouragements and constant support. I never felt home-sick at ARIES and for that I would like to thank my close friends Arti, Jessy, Neelam, Manash and Sanjeev. Thanks to all of you for all the scientific v and non-scientific discussions we had for all these years. I really enjoyed all the breakfast and dinner parties we had together in our hostel. I heartily acknowledge the comfort I felt sharing an office with Arti for all these years and also for her help regarding my stay in the new hostel during thesis writing. I feel proud to share a good friendship with the strongest ladies of our institute Jessy and Neelam and I am thankful to them for their timely help at various stages of my PhD. Many thanks to Manash for the scientific discussions and making the atmosphere always lively. The encouragements and support provided by Sanjeev is highly acknowledged. I am also thankful to Lijo and Vinod for their company and encouragements. A special thanks to Yugantika for her smiles which used to help me forget all my worries. It gives me an immense pleasure to thank my friends David and Camille for a long lasting friendship. I am also thankful to David for acquainting me with the optical observations and data reduction techniques. I am grateful to Nautiyal ji for making me feel at home during my PhD. Thanks to Sumana and Rupak for all the chicken parties and sharing a good friendship. Thanks to my batch-mates Himali, Amitava, Prashant, Saurav for their company and constant support. I would like to express my thanks to Surjit Das, Srivastava ji, Giancarlo, Prashant Hegde, Kumar and Samresh for their company. I am grateful to my seniors Kuntal, Saurabh, Dumka, Bhuwan, Jeewan, Shashi, Santosh, Biman, Tejbir for their guidance and encouragements. Thanks to my friends Es- hwar, Rajesh, Pankaj, Akash, Narendra, Bindu, Ram Kesh, Brajesh, Chhavi, Chrisphin, Haritma, Tapaswini, Hema, Devesh, Giridhar, Kishore for their help and making the atmosphere always lively. I would like to acknowledge the help I recieved from my friends Neha, Kalyan, Ravi, Rupak, Iswar and Pankaj in thesis draft and accompanying me to the University for the thesis submission related trips. I would like to express my thanks to Pandey Sir, Alok Gupta, Hum Chand, In- dranil, Ramakant, Snehlata and Sanwal Sir for the scientific discussions and motivations. I am thankful to Soman Mondal for introducing me to the infrared data reduction techniques. Thanks to Amitesh and Manish for their help in various academic matters. I would like to acknowledge the help from the technical staff of our telescope at ARIES during observations, library and the computer staff (Rajesh Kumar, Bhat Ji, Giri Ji, Pant ji, Nitin, Sanjit, Purushottam, Dheeraj and Navin ji) for their help throughout my PhD. Administrative staff (particularly Jhinghan Sir, Abhishek and Hansa madam) of ARIES is highly acknowledged for providing me necessary support during my research work. During my PhD I made many visits to Raman Research Institute. I am thankful to everybody on astro-floor of RRI for the scientific discussions and encouragements. vi

I am very grateful to my RRI friends Ruta, Mamta, Kshitij, Wasim, Yogesh, Poonam, Kanhaiya, Peeyush, Nishant, Chetana, Rabai, Csg, Maushmi, Tripta and Ravi Shankar for their help and company during my visits. Thanks to Vidya, RRI library and administrative staff for providing me necessary help during my visits. This thesis would not have been possible without a constant support from Atish. I am thankful to him for his confidence in me which always helped me to tackle every difficult problem in my scientific and personal life. His encouragements, motivations and patience always helped me moving on with my research work. It is not possible to give a long list of help I received from him during my PhD but I would definitely like to thank him for his life long friendship and love. Most of all, I would like to thank my parents for providing me with the opportunity to be where I am. Without them, none of this would even be possible. They have always given me the strength and wisdom to be sincere in my work, for setting high moral standards and supporting me through their hard work and for their unselfish love and affection. Many thanks to my brothers ‘Ginny & Binny’ for their confidence in me and standing by me in all difficult situations. I wish to express my sincere thanks to my in-laws for their constant encouragement and support to continue my carrier in research. And finally I am thankful to ‘Waheguru’ for making this volume a reality in my life.

We succeed only as we identify in life, or in war, or in anything else, a single overriding objective, and make all other considerations bend to that one objective. - Dwight D. Eisenhower

Ramanpreet Kaur 31 March 2009 vii

List of Publications

Publications in refereed journals

1. Quasi-periodic Oscillations in XTE J0111.2-7317: Highest Frequency among the HMXB Pulsars Ramanpreet Kaur, Biswajit Paul, Harsha Raichur, Ram Sagar ApJ, 660, 1409 (2007)

2. A study of the long term evolution of quasi periodic oscillations in the accretion powered X-ray pulsar 4U 1626-67 Ramanpreet Kaur, Biswajit Paul, Brijesh Kumar, Ram Sagar ApJ, 676, 1184 (2008)

3. Multiwavelength study of the transient X-ray binary IGR J01583+6713 Ramanpreet Kaur, Biswajit Paul, Brijesh Kumar, Ram Sagar MNRAS, 386, 2253 (2008)

4. Chandra and XMM-Newton observations of the low-luminosity X-ray pulsators SAX J1324.4-6200 and SAX J1452.8-5949 Ramanpreet Kaur, Rudy Wijnands, Alessandro Patruno, Vincenzo Testa, Gianluca Israel, Nathalie Degenaar, Biswajit Paul, Brijesh Kumar MNRAS, in press (2009)

5. Nature of ﬁve low-luminosity X-ray pulsars AX J1700.1-4157, AX J1740.1-2847, AX J1749.2-2725, AX J1820.5-1434 and AX J1832.3–0840 Ramanpreet Kaur, Rudy Wijnands, Biswajit Paul, Alessandro Patruno, Nathalie Degenaar in preparation.

Astronomers Telegram

1. Optical photometric observations of the Be/X-ray binary V0322+53 Ramanpreet Kaur, Brijesh Kumar, Biswajit Paul, Ram Sagar Astronomers telegram, 1807, 1 (2008) viii

Publications in conference proceedings

1. Quasi periodic oscillations in XTE J0111.2–7317, highest frequency among the HMXB pulsars Ramanpreet Kaur, Biswajit Paul, Harsha Raichur, Ram Sagar Bull. Astr. Soc. India, 51 (2007)

2. A study of the long term evolution of quasi periodic oscillations in the accretion powered X-ray pulsar 4U 1626–67 Ramanpreet Kaur, Biswajit Paul, Brijesh Kumar, Ram Sagar 37th COSPAR Scientiﬁc Assembly. held 13 - 20 July 2008, in Montreal, Canada, 1470 (2008)

3. Chandra and XMM-Newton observations of 7 enigmatic faint persistently slow pulsators Ramanpreet Kaur, Rudy Wijnands, Nathalie Degenaar, Alessandro Patruno 37th COSPAR Scientiﬁc Assembly. held 13 - 20 July 2008, in Montreal, Canada, 1471 (2008) ix

NOTATIONS AND ABBREVIATIONS

The most commonly used notations and abbreviations in the thesis are given below. If a symbol has been used in a diﬀerent connection than listed here, it has been explained at the appropriate place.

Notations

Aλ Galactic extinction value at wavelength λ

AV Galactic extinction value at visual wavelength eV electron volt E(B V) thevalueofGalacticreddening − M Solar mass ⊙ R Solar radius ⊙ mag magnitude (stellar) log logarithm

U, B, V, R, I, J, H, Ks standard broad-band photometric pass bands

NH neutral hydrogen column density e eccentricity of an orbit of a star

MN mass of a neutron star

MWD mass of a white dwarf

ME mass of an early-type star

RN radius of a neutron star

RWD radius of a white dwarf

RE radius of an early-type star m˙ mass accretion rate

Lacc accretion luminosity

Frad radiative force

σT Thomson cross section

LEDD Eddington luminosity µ magnetic moment B magnetic ﬁeld G Gauss

Pmag magnetic pressure

Porb orbital period

Ps spin period x

P˙ spin period derivative rA Alfven radius rco co-rotation radius racc accretion radius rm magnetosphere radius Γ powerlaw photon index keV kilo electron volt kHz kiloHertz kpc kiloparsec km kilometer T temperature k Boltzmann constant

νk Keplerian frequency

νs spin frequency

νorb orbital frequency

νQPO quasi periodic oscillations frequency vn neutron star velocity s,sec seconds d days yr year hr hour mins minutes ρ stellar wind density rms root mean square a separation between two stars

L1 - L5 Lagrangian points of a binary system J speciﬁc angular momentum

Rcirc circularization radius α angle between magnetic axis and rotation axis in a pulsar β angle between line of sight and rotation axis in a pulsar l Galactic longitude of a star b Galactic latitude of a star Z zenith distance

LX X-ray luminosity

FX X-ray ﬂux xi f(E) observed X-ray spectrum R(I,E) instrumentresponse

Abbreviations

ACIS Advanced CCD imaging spectrometer on Chandra ADC Accretion Disk Corona ADU Analog to Digital Unit ARF Ancilliary Response File ARIES AryabhattaResearchInstituteofobservationalSciencES ASCA Advanced Satellite for Cosmology and Astrophysics, a satellite ASM All Sky Monitor on RXTE BAT Burst Alert Telescope on Swift BATSE Burst and Transient Source Experiment on CGRO BFM Beat Frequency Model BL Boundary Layer CCD(s) Charge Coupled Device(s) CCF Current Calibration Files CD Color-color Diagram CDA Chandra Data Archive CGRO Compton Gamma-Ray Observatory, a satellite CRSF Cyclotron Resonance Scattering Feature CV Cataclysmic Variable star CXC Chandra X-ray observatory Science Center DAOPHOT Dominion Astrophysical Observatory Photometry (software) DEC. Declination DIT Detector Integration Time EPIC European Photon Imaging Camera ESA European Space Agency ESO European Southern Observatory EW Equivalent Width FFT Fast Fourier Transform FOV Field Of View FWHM Full Width at Half Maxima xii

GIS Gas Imaging Spectrometers on ASCA HCT Himalayan Chandra Telescope HEASARC High Energy Astrophysics Science Archive Research Center HFOSC Hanle Faint Object Spectrograph Camera HID Hardness Intensity Diagram HMXB High Mass X-ray Binary HRC High Resolution Camera IAO Indian Astronomical Observatory INTEGRAL International Gamma-ray Astrophysics Laboratory IP Intermediate Polar IRAF Image Reduction and Analysis Facilities KFM Keplerian Frequency Model LECS Low Energy Concentrator Spectrometer on BeppoSAX LMC Large Magellanic Cloud LMXB Low Mass X-ray Binary MECS Medium Energy Concentrator Spectrometer on BeppoSAX MIDAS Munich Image and Data Analysis System MJD Modified Julian Date MOS Metal-Oxide-Silicon instrument on XMM-Newton NIR Near Infra-Red NTT New Technology Telescope ODF Observation Data Files OM Optical Monitor on XMM-Newton PCA Proportional Counter Array on RXTE PCU Proportional Counter Unit on RXTE PDS Power Density Spectrum PPS Pipeline Processing Subsystem PSF(s) Point Spread Function(s) QE Quantum Efficiency QPO Quasi Periodic Oscillations R.A. Right Ascension RGS Reflection Grating Spectrometer on XMM-Newton RXTE Rossi X-Ray Timing Explorer, a satellite SAS Science Analysis Software SFXT Supergiant Fast X-ray Transients xiii

SIS Solid state Imaging Spectrometer on ASCA SMC Small Magellanic Cloud SOFI Son OF Infrared spectrometer and array camera ST Sampurnanand Telescope TOA Time of Arrivals UV Ultraviolet waveband UVOT Ultraviolet/Optical Telescope on Swift VLT Very Large Telescope XRT X-Ray Telescope on Swift XSA XMM-Newon Science Archive XSPEC An X-Ray Spectral Fitting Package 2MASS Two Micron All Sky Survey xiv Contents

1 Introduction 1 1.1 Historicaloverview ...... 2 1.2 X-raybinaries...... 3 1.2.1 ClassificationofX-raybinaries ...... 3 1.2.2 PersistentversustransientX-raybinaries ...... 6 1.3 Basicsofaccretion ...... 9 1.3.1 TheEddingtonluminosity ...... 9 1.3.2 Stellarwindaccretion ...... 10 1.3.3 Roche-lobeoverflow ...... 12 1.3.4 Accretion from Be stars ...... 13 1.3.5 Discformationandboundarylayer...... 15 1.3.6 Accretionontomagnetizedobjects...... 17 1.4 AccretionpoweredX-raypulsars ...... 20 1.4.1 Pulseperiods ...... 21 1.4.2 Pulseprofiles ...... 22 1.4.3 OrbitalModulations ...... 25 1.4.4 Quasiperiodicoscillations ...... 26 1.4.5 LongtermX-rayvariabilities...... 31 1.4.6 X-rayEnergySpectra...... 33 1.4.7 Cyclotronabsorptionlines ...... 33 1.5 MotivationandOrganizationofthethesis ...... 36

2 Observations and Data Analysis 39 2.1 Introduction...... 39

xv xvi CONTENTS

2.2 Opticalobservations ...... 40 2.2.1 Telescopes ...... 40 2.2.1.1 1.04-mSampurnanandTelescope ...... 40 2.2.1.2 2.01-mHimalayanChandranTelescope...... 40 2.2.2 CCDdetectors ...... 41 2.2.3 Thespectrograph ...... 44 2.2.4 Photometry ...... 44 2.2.5 Spectroscopy ...... 51 2.3 Near-infraredobservations ...... 54 2.3.1 Telescopes ...... 54 2.3.1.1 NewTechnologyTelescope(NTT) ...... 54 2.3.1.2 TwoMicronAllSkySurvey(2MASS) ...... 55 2.3.2 Photometry ...... 55 2.4 X-rayobservations ...... 56 2.4.1 Telescopes ...... 56 2.4.1.1 Chandra ...... 56 2.4.1.2 XMM-Newton ...... 57 2.4.1.3 RXTE ...... 58 2.4.1.4 Swift ...... 60 2.4.1.5 OtherX-rayobservatories ...... 60 2.4.2 Extractingdataproducts ...... 62 2.4.3 TimingAnalysis ...... 64 2.4.4 Spectroscopy ...... 65

I True nature of X-ray binaries 69

3 The nature of seven faint X-ray pulsators 71 3.1 Introduction...... 71 3.2 General properties of persistent Be/X-raypulsars ...... 73 3.3 OursampleofX-raypulsators ...... 74 3.4 X-rayobservationsanddatareduction ...... 75 3.4.1 XMM-Newton ...... 75 3.4.2 Chandra ...... 76 3.5 X-raytiminganalysis ...... 78 3.6 X-rayspectralanalysis ...... 80 CONTENTS xvii

3.7 Infraredobservationsanddataanalysis ...... 85 3.7.1 NewTechnologyTelescope ...... 85 3.7.2 2MASS...... 90 3.7.3 Identiﬁcation of near-infrared counterparts ...... 90 3.7.4 SpectraltypeoftheNIRcounterparts ...... 95 3.8 HMXBs, LMXBs and Intermediate Polars - a comparison ...... 99 3.9 Discussion...... 100 3.10 Conclusions...... 111

4 Transient X-ray binary IGR J01583+6713 113 4.1 Introduction...... 113 4.2 IGRJ01583+6713 ...... 114 4.3 OpticalObservations ...... 114 4.3.1 UBVRI Photometry...... 115 4.3.2 Spectroscopy ...... 119 4.4 X-rayObservations ...... 123 4.5 ResultsandDiscussions...... 127 4.5.1 PhotometricVariability...... 127 4.5.2 Stabilityof the Hα emissionline...... 129 4.5.3 X-rayvariabilityandSpectrum...... 129 4.5.4 SpectralClassiﬁcation ...... 131 4.6 Conclusions...... 134

II QPOs in X-ray binaries 135

5 DiscoveryofQPOsinXTEJ0111.2–7317 137 5.1 Introduction...... 137 5.2 XTEJ0111.2–7317 ...... 138 5.3 RXTE observations ...... 139 5.4 QuasiPeriodicOscillationinXTEJ0111.2–7317 ...... 141 5.5 Contamination by a nearby X-ray binary pulsar SMC X-1 ...... 144 5.6 Discussion...... 146 5.7 Conclusions...... 148 xviii CONTENTS

6 QPOs in the X-ray pulsar 4U 1626–67 149 6.1 Introduction...... 149 6.2 4U1626–67...... 150 6.3 Observationsanddataanalysis ...... 152 6.4 QPOevolutionin4U1626–67 ...... 156 6.5 Discussion...... 157 6.6 Conclusions...... 160

7 Summary and future prospects 161

A Magnetic Cataclymic Variables 167 Bibliography 175 List of Figures

1.1 Schematic diagram of a typical high mass X-ray binary (top) and low mass X-ray binary (bottom). The solid line represents Roche lobes of the stars in a binary system. In case of HMXB, the companion star does not fill its Roche lobe and the mass accretion process takes place through stellar wind. In case of LMXB, the companion star fills its Roche lobe and the accretion process takes place through Roche lobe overflow (from TaurisandvandenHeuvel2006)...... 4 1.2 Lightcurve of a transient X-ray binary pulsar V0332+53 observed with Rossi X-ray timing explorer - All Sky Monitor (ASM), clearly showing an X-ray outburst during November2004 - March 2005. The RXTE -ASM 1 count rate reached nearly 90 counts s− as compared to approximately 1 1 count s− duringitsquiescentstate...... 7 1.3 Left: compact star accreting from the stellar wind of an early-type companion Right: closer view of bow shock produced near the compact object, along with the accretion radius in stellar wind accretion. This figure istakenfromFranketal.(2002)...... 11 1.4 A cross-section in the equatorial plane of the critical equipotential surfaces in a binary. The symbol ( ) represents the center of mass of the × system. The thick curve crossing through L1 is the Roche lobe. The other

Lagrangian points of the binary system are marked as L2,L3,L4, and L5 (fromTaurisandvandenHeuvel2006)...... 14

xix xx LIST OF FIGURES

1.5 Left: Perspective view of an idealized accretion disk showing the fluid ‘spiraling’ in. Right: Accretion disk around a magnetized neutron star. The magnetic lines shown represent schematically the boundary of the magnetosphere. This Figure is taken from Frank et al. (2002)...... 16 1.6 A schematic diagram of an accretion powered X-ray pulsar. The spin axis and the magnetic axis are misaligned (thus: a pulsar). X-rays are emitted along the magnetic field axis as the pulsar accretes near its magnetic poles. We have also marked the region in accretion disk where quasi periodic oscillations (refer to Section 1.4.4) are assumed to originate. (Credit : http://lheawww.gsfc.nasa.gov.) ...... 20 1.7 Accreting pulsar models. Left: The pencil beam model. Right: The fan beam model. M and ω represent the magnetic axis and spin axis of an X-raypulsarrespectively...... 23 1.8 The folded 1 - 10 keV light curves of EXO 2030+375 showing the evolution of the pulse profiles during the decay of an outburst (from Parmar etal.1989)...... 24 1.9 The power density spectrum of an accretion-powered X-ray pulsar 4U 1626–67 showing detection of mHz quasi-periodic oscillations (from Chakrabartyetal.2001)...... 29 1.10 Schematic model of a Be-star X-ray binary system. The neutron star moves in an eccentric orbit around the Be-star which is not filling its Roche lobe. However, near the periastron passage the neutron star accretes circumstellar matter, ejected from the rotating Be-star, resulting in an X-ray burst lasting several days. This Figure is taken from Tauris and vandenHeuvel(2006)...... 32 1.11 The observed (histogram) and model spectrum (solid line) of the accretion-powered pulsar 4U 0115+63 showing evidence for cyclotron lines with as many as four harmonics (from Heindl et al. 1999)...... 34

2.1 The1.04-mSTatARIESand2.01-mHCTatIAO,Hanle...... 42

2.2 Response curve of Wright 2k 2k CCD used at ST, ARIES at -120◦C. × ThecurveindicatesthattheCCDpeaksnear6700Å...... 44 2.3 A chart summarizing the optical data reduction procedure...... 51 2.4 The Rossi X-ray Timing Explorer (RXTE)...... 59 LIST OF FIGURES xxi

3.1 Pulse profiles of our sample of X-ray pulsators obtained from the XMM- Newton observations. The name of source is written at the top of each pulse profile. No pulsations were detected in SAX J1452.8–5949, thus its pulseprofileisnotshown...... 81 3.2 The XMM-Newton EPIC-MOS and pn spectra of SAX J1324.4–6200 and SAX J1452.8–5949 fitted with an absorbed powerlaw model. The EPIC- pn, MOS1 and MOS2 data points are represented by blue, red and black colorrespectively...... 83 3.3 The XMM-Newton EPIC-MOS and pn spectra of AX J1700.1–4157 and AX J1740.1–2847 fitted with an absorbed powerlaw model. The EPIC- pn, MOS1 and MOS2 data points are represented by blue, red and black colorrespectively...... 84 3.4 The XMM-Newton EPIC-MOS and pn spectra of AX J1749.2–2725 and AX J1820.5–1434 fitted with an absorbed powerlaw model. The EPIC- pn, MOS1 and MOS2 data points are represented by blue, red and black colorrespectively...... 86 3.5 Top : The XMM-Newton EPIC-MOS and pn spectra of AX J1832.3-0840 fitted with an absorbed blackbody model. The EPIC-pn, MOS1 and MOS2 data points are represented by blue, red and black color respectively. Bot- tom : The same spectrum but in the energy range 5 - 8 keV to show the Feemissionlines...... 87

3.6 Left : TheNIR Ks waveband image of SAX J1324.4–6200, SAX J1452.8– 5949 and AX J1700.1–4157 taken with ESO-NTT. The X-ray positions of all these sources measured from their XMM-Newton observations are represented by a black circle and the positions measured from their Chandra observations are represented by a white circle. The radius of the circle represents an error on their position measurement. Right : The Chandra ACIS-I images of the same pulsators. The black circle represents an error circle on their position as measured using XMM-Newton observations. . . 91 xxii LIST OF FIGURES

3.7 Left : The ESO-NTT NIR Ks waveband image of AX J1740.1–2847, AX J1749.2–2725 and 2MASS K band image of AX J1820.5–1434. The color and radius of the the circles are as explained in the Figure 3.6 except AX J1820.5–1434 for which its XMM-Newton error circle on the position

is represented by a white circle in its Ks waveband image. Right : The Chandra ACIS-I images of the same pulsators. The black circle represents an error circle on their position as measured using XMM-Newton observations...... 92

3.8 Left : The NIR Ks waveband image of X-ray pulsator AX J1832.3–0840 taken from the 2MASS point source catalog. The XMM-Newton error circle (in black color) and the Chandra error circle (in white colour) on the position are also plotted. Right : The Chandra ACIS-I image of AX J1832.3–0840 along with the error circle on the position as measured using the XMM-Newton observations...... 93 3.9 The observed (ﬁlled circles) and extinction-free (open circles) spectral energy densities of NIR counterparts of our sample of X-ray pulsators. The solid line (red color) and the dashed-dotted line (blue color) represent blackbody model spectral energy density of the star at temperature, T = 20,000 K and T = 3,000 K respectively and are shifted to match the extinction-free spectral energy density of the star at NIR H band...... 98

4.1 Identiﬁcation chart of IGR J01583+6713 taken with the ST in B passband on December 13, 2006. The transient is marked as ‘T’ and comparison starsaremarkedas1,2,3,4,5and6...... 117

4.2 The diﬀerence in the measured (Bobs, Vobs, Robs and Iobs) and standard

(Bst, Vst, Rst and Ist) BVRI magnitudes of IGR J01583+6713 and Com- parisonstar1fromJD2453710toJD2453810...... 118 4.3 Flux calibrated optical spectrum of IGR J01583+6713 taken on October 15, 2006. Grism 7 and Grism 8 spectrum are combined together to show it overa wavelength range of 3800 - 9000 Å. Diﬀuse Interstellar bands are marked as "DIB", telluric absorption bands are marked with a ﬁlled triangle and the Sodium doublet is marked as "NaD". "P" represents Paschen lines. Hα and Hβ arealsomarked...... 121 LIST OF FIGURES xxiii

4.4 Continuum spectrum of IGR J01583+6713 in the wavelength range of 6100 - 7900 Å taken on October 15, 2006 and October 16, 2006. A few weakly identified features like FeII, HeI, SiII and OI are marked...... 122 4.5 Hα line profile variations of IGR J01583+6713 ...... 123 4.6 ASM lightcurve of IGR J01583+6713 from MJD 53660 to MJD 54160 including the outburst observed by the INTEGRAL and the Swift during MJD 53710 to MJD 53720, marked as "T" in the figure. Also marked are optical photometric observations (by filled triangle), optical spectroscopic observations (by up-arrow), the RXTE observations (by open circles with a dot inside) and the Swift observations(byopenstar)...... 125 4.7 X-ray spectrum of IGR J01583+6713. Top - the Swift observations made on December 13, 2005, middle - RXTE observations on December 14, 2005, bottom - the Swift observations made on April 05, 2006. The points with error bars are the measured data points and the histograms are the respective best fitted model spectrum consisting of absorbed powerlaw model components, convolved with the respective telescope/detector responses...... 128 4.8 Light curve of IGR J01583+6713 observed using the Swift observatory onDecember13,2005,foldedmodulo469.2s...... 130 4.9 Bottom - Flux Normalized optical spectrum of IGR J01583+6713, Top - Flux calibrated optical spectrum of HD 164284 from Andrillat et al. (1988), a comparison. P12 to P20 are the Hydrogen Paschen lines from P12 to P20. FeII, OI, CaII and NI spectral features are also marked. . . . 132

5.1 RXTE-ASM Lightcurve of XTE J0111.2–7317 and SMC X-1 over a period of 300 days are shown here along with rescaled RXTE-PCA lightcurve from observations made towards XTE J0111.2–7317. Two time ranges MJD 51165.56 to 51177.32 and MJD 51228.00 to 51228.09 marked with ‘A’ and ‘B’ correspond to the time when the two sources XTE J0111.2–7317 and SMC X-1 were bright respectively. The power density spectra for the segments ‘A’ and ‘B’ are shown in Figure5.2. . . . 140 xxiv LIST OF FIGURES

5.2 Power density spectra generated from the lightcurves obtained from RXTE-PCA observations made towards XTE J0111.2–7317 are shown here. The top and bottom spectra are for the time ranges ‘A’ and ‘B’ respectively shown in Figure 5.1. The top figure has been multiplied by a factorof500forthesakeofclarity...... 142 5.3 Power density spectrum of XTE J0111.2–7317 generated from the lightcurve over the entire energy band of the PCA. The line represents the best fitted model for the continuum and a Gaussian centered at the QPOfrequency...... 143 5.4 RXTE-ASM light curve of SMC X-1 folded with its orbital period along with RXTE-PCA light curve of XTE J0111.2–7317 during segment ‘A’ of Figure5.1...... 145 5.5 Relation between average count rate and pulsed count rate for SMC X-1 is shown here. The formal correlation coefficient is determined to be 0.97. 147

6.1 Power density spectrum generated from the lightcurve obtained from the EXOSAT observationmadeonAugust30,1983...... 155 6.2 All power density spectrum are generated from the lightcurves obtained from observations listed in Table 6.1. in chronological order. Different constant numbers were added to each plot for clarity. The year of observations is written along with each PDS. A vertical line is drawn at 49.77 mHz, QPO frequency of ASCA 1993 observations, to clearly see the de- creaseinQPOfrequencywithtime...... 158 6.3 QPO frequency evolution history of 4U 1626–67 from 1983 to 2004. The solid line is a linear fit to the data from 1993 to 2004. Error bars represent the 1σ confidenceintervals...... 159

A.1 The schematic of the magnetic cataclysmic variable having B 107 G ∼ (polar). (Credit : http://heasarc.gsfc.nasa.gov/images)...... 169 A.2 A schematic of an intermediate polar showing a white dwarf accreting matter from its companion star. The formation of accretion disk takes place in intermediate polars and the inner accretion disc is usually truncated by the magnetic ﬁeld of the white dwarf. (This ﬁgure is taken from http://heasarc.gsfc.nasa.gov/images)...... 172 List of Tables

1.1 SummaryofbasicpropertiesofHMXBsandLMXBs ...... 6 1.2 Quasi periodic oscillations in accretion powered X-ray pulsars ...... 27 1.3 Energies of Cyclotron lines in accretion-powered X-ray pulsars...... 35

2.1 ParametersofthetwoIndianopticaltelescopes...... 41 2.2 Characteristics of optical CCD Wright used with ST and SITe ST-002 usedwithHCT...... 42 2.3 Speciﬁcations of the HFOSC spectrograph at HCT, Hanle...... 43

2.4 Details of thebroadband filters used with ST, Nainital. λe f f istheeffective wavelength in the units of Angstrom (Å). The thickness of the glass in mm isindicatedinthebrackets...... 43 2.5 The IRAF and DAOPHOT II tasksusedforopticaldatareduction. . . . . 52 2.6 ParametersoftheNewTechnologyTelescope(NTT) ...... 53 2.7 Characteristics of CCD used in the NIR instrument SOFI ontheNTT. . . 54 2.8 Confidenceintervalfortheparameters1,2and3 ...... 67

3.1 Summary of basic parameters of our sample of X-ray pulsators...... 75 3.2 Logofthe XMM-Newton EPIC X-ray Observations of X-ray pulsators . . 76 3.3 Position of X-ray pulsators as determined from their XMM-Newton observations. The count rates of them in EPIC-MOS1, MOS2 and pn instru- mentsarealsolisted...... 77 3.4 Logofthe Chandra ACIS-IX-rayObservations ...... 77 3.5 Position of X-ray pulsators as determined using the Chandra observations, along with the total counts with which these were detected...... 78

xxv xxvi LIST OF TABLES

3.6 The pulse period (measured using the task efsearch in FTOOLS) and pulse fractional amplitude (%) measured using the task efold in FTOOLS. The spin period derivative (P)˙ is also estimated using the previous and presentpulseperiodmeasurements...... 82 3.7 Spectral parameters of X-ray pulsators as measured from their XMM- Newton EPICobservations...... 88 3.8 The center, width and equivalent width (EW) of Fe emission lines detected in the X-ray spectrum obtained from the XMM-Newton observations of X-ray pulsators in the units of keV. An upper limit on EW of Fe 6.4 keV line is given in case of non-detection of any Fe emission line in theX-rayspectrum...... 89 3.9 X-ray luminosity of X-ray pulsators at a distance of 1 kpc, 8 kpc and 20 kpc...... 90 3.10 Log of the near-infrared observations of ﬁve X-ray pulsators taken with 3.52-m ESO-NTT. NDIT represents the Number of single frames, having exposure times of DIT (Detector Integrator Time) seconds and are used to generate an output image having exposure time equal to one DIT. The number of output frames are represented by Nframes...... 93

3.11 The observed J, H and Ks magnitudes of the most likely NIR counterparts oftheX-raypulsators...... 94

3.12 Galactic extinction in V, J, H and Ks wavebands for our sample of X-ray pulsators...... 97 3.13 Dereddened NIR magnitudes of counterparts of X-ray pulsators in J, H

and Ks wavebands...... 97 3.14 Spin-periodhistoryofSAXJ1324.4-6200 ...... 102

4.1 Log of broadband optical photometric observations of the transient source andLandolt(1992)standardﬁelds ...... 116 4.2 BVRI magnitudesofcomparisonstars ...... 119 4.3 Log of optical spectroscopic observations of the X-ray transient IGR J01583+6713 ...... 120 4.4 Spectral Parameters for the Swift and the RXTE observations...... 126

6.1 LogofObservationsof4U1626–67 ...... 153

A.1 Classiﬁcationofcataclysmicvariables...... 171 Chapter 1

Introduction

"Every atom in your body was once inside an exploding star...... "

- Arthur Lawrence Kraus More than ninety percent of the bright Galactic X-ray sources in the sky are binary systems in which a compact object (neutron star/black hole) accretes matter from its binary companion. Since their discovery by Giacconi et al. (1962), they have oﬀered unique insights into the astrophysics of the end stages of stellar evolution and physics of matter in extreme astrophysical conditions. These systems have also oﬀered the measurement of masses of compact objects in them through multiwavelength observations, providing the strongest evidence for the existence of black holes in the Universe (Mc-

Clintock and Remillard, 1986). Even after ﬁve decades since the ﬁrst X-ray binary was discovered, the study of these systems is far from complete and is still among the most exciting topics in the modern astrophysics. In this thesis we report our studies of ten such X-ray binary systems using multiwavelength observations, most of which are neutron star X-ray binaries and a few are probable white dwarf binaries. In the next Section, we have given a historical overview of X-ray binaries.

1 2 1. Introduction

1.1 Historical overview

The first extrasolar X-ray source was discovered in Scorpius constellation during a rocket flight by Giacconi et al. (1962) and was named Sco X-1. Later on, the distance estimation of this source suggested that the X-ray output of Sco X-1 is about 60000 times the total luminosity of the Sun. In the following years, a few more celestial X-ray sources were discovered but the underlying X-ray emission mechanism remained a mystery for years. A possible explanation of X-ray emission in Sco X-1 and the other X-ray sources was suggested in late sixties through a model in which a highly compact star emits X-rays by accreting matter from its companion star (Salpeter , 1964). The companion star in the binary system was suggested to be an ordinary star. The first confirmation of this mechanism was made with the discovery of X-ray source Cen X–3 by Giacconi et al. (1971) using data from the Uhuru satellite, the first X-ray astronomy satellite launched in December 1970 (Fabian, 1975). A careful analysis of X-ray ligthcurves of Cen X-3 showed regular pulsations with a period of about 4.8 s. The analogy with radio pulsar suggested that the pulsations can arise from a ‘light-house’ effect close to the surface of a rotating neutron star which possesses a strong magnetic field ( 1012 G) misaligned ∼ with respect to the rotation axis. Moreover, it was found that the frequency of pulsations underwent small periodic changes. This suggested a Doppler effect due to the orbital motion of X-ray source in a binary system. The observations of periodic eclipses, in agreement with observations of the Doppler curve, showed that the orbital period of the system is 2.09 d. After removing this effect, it was noticed that the pulse period of Cen X-3 was increasing progressively, which suggested that the neutron star is spinning up with time. Later on, both spin-up and spin-down behaviours were seen in a number of X- ray pulsars (Nagase, 1989). The most feasible way of explaining the origin of this effect is by a torque exerted on the neutron star by accreting material. Later on, many other X-ray binaries with wide range of temporal and spectral characteristics were discovered and studied extensively. From the multiwavelength observations of a number of X-ray 1.2 X-ray binaries 3 binaries, it became clear that the compact object in most of the bright X-ray binaries is a black-hole or a neutron star. However, as the black-holes cannot produce pulsations due to the absence of a solid surface, the compact objects in X-ray binary pulsars are suggested to be neutron stars.

These systems are observed to have a wide variety of companion stars and are thus classiﬁed into various sub-classes on the basis of mass of the companion star. The most commonly used classiﬁcation scheme of X-ray binaries is described in detail in the following Section.

1.2 X-ray binaries

1.2.1 Classiﬁcation of X-ray binaries

The X-ray binaries are mainly divided into two categories, high mass X-ray binaries (HMXBs) and low-mass X-ray binaries (LMXBs), on the basis of the mass of the donor star which can vary from 1 M to 20 M . In HMXBs, a compact object (neutron ≤ ⊙ ≥ ⊙ star or a black hole) orbits a massive star ( 10 M ) which could be a supergiant or an ≥ ⊙ early-type star like Be star (Liu et al., 2000) and are accordingly called supergiant X-ray binaries and Be/X-ray binaries respectively. Most of the Be/X-ray binaries have relatively wide orbits with moderate eccentricities and their compact companions spend most of their time far away from the disk surrounding the Be stars (van den Heuvel and Rappaport 1987; Bhattacharya and van den Heuvel 1991; Apparao 1994). However, during the time of the neutron star’s periastron passage, these systems undergo X-ray outbursts from a low-velocity and high-density wind around Be stars. In the supergiant X-ray binaries, the compact object orbits a supergiant early-type star deep inside the highly supersonic wind and is either powered by pure stellar wind accretion (refer Section 1.3.2; Figure 1.1) or, in the case of the brighter systems, by Roche lobe overﬂow via an accretion disk (refer

Section 1.3.3) or both. The LMXBs consist of a compact object accreting matter from a companion 4 1. Introduction

Figure 1.1: Schematic diagram of a typical high mass X-ray binary (top) and low mass X-ray binary (bottom). The solid line represents Roche lobes of the stars in a binary system. In case of HMXB, the companion star does not fill its Roche lobe and the mass accretion process takes place through stellar wind. In case of LMXB, the companion star fills its Roche lobe and the accretion process takes place through Roche lobe overflow (from Tauris and van den Heuvel 2006). 1.2 X-ray binaries 5 star whose mass is generally 1M . These companions are cooler than the early-type ≤ ⊙ companions of HMXBs (Lang, 1992) and do not have strong winds so the orbit of these binaries are assumed to be compact enough for Roche lobe overflow to take place. A schematic diagram of accretion process through the Roche lobe overflow is shown in

Figure 1.1 (bottom). The optical emission of LMXBs is dominated by thermal emission from the accretion disk while that of HMXBs is by the companion star. Most of the LMXBs are Population II objects and their spatial distribution shows no preference for the Galactic plane, although they are mostly concentrated towards the Galactic center. On the other hand, most of the HMXBs are in the population I objects and are concentrated in the Galactic plane. It is found that neutron stars in LMXBs usually have weaker magnetic fields as compared to HMXBs, which is understood to be due to accretion-induced decay of the neutron star’s magnetic field and is still an active field of research (Bhattacharya and Srinivasan, 1995). The differences between HMXBs and LMXBs are summarized in Table 1.1. Currently there are 114 known HMXBs (including black-hole binaries) in our ∼ Galaxy (Liu et al., 2006) and an equal number of HMXBs are known in Magellanic Clouds (Small Magellanic Cloud & Large Magellanic Cloud; Liu et al. 2005) while 190 ∼ LMXBs (including black-hole binaries) are known altogether in our Galaxy and Magel- lanic Clouds. (Liu et al., 2007). Some of the common observational properties of these systems help us to classify the nature of newly discovered systems e.g., an unidentified system is classified as a HMXB if it shows : strong flaring and absorption variability on a timescale of minutes, transient outbursts, pulsations and/orhasa hard1- 10keVspectrum with a power law index of order 0 - 1 (White et al., 1995) and similarly, an unidentified system is classified as a LMXB if it shows : type-I X-ray bursts1, the 1 - 10 keV spectrum is soft with a characteristic temperature of 5 - 10 keV, the orbital period is less than about

1These are the ﬂux enhancements assumed to occur due to the thermonuclear explosions on the surface of the accreting neutron stars, triggered by unstable hydrogen or helium burning (Babushkina et al., 1975) 6 1. Introduction

Table 1.1: Summary of basic properties of HMXBs and LMXBs

Properties HMXBs LMXBs Donorstar O-B(M 10 M ) K-MorWD(M 1M ) ≥ ⊙ ≤ ⊙ Population I(107 yr) II(109 yr)

LX/Lopt 0.001-10 100-1000 optical spectrum stellar-like reprocessing Accretiondisk yes,small yes X-ray spectrum hard (kT 15keV) soft(kT 10 keV) ≥ ≤ Orbitalperiod 1-100d 10min-10d X-ray eclipses common rare Magnetic ﬁeld 1012 G 109 G ∼ ∼

12 hr (White et al., 1995). Irrespective of the various known classes of X-ray binaries (described above), they appear both as persistent and transient X-ray sources. The physical origin of their X-ray emission behaviours are described below.

1.2.2 Persistent versus transient X-ray binaries

An X-ray binary is classiﬁed as a persistent X-ray binary if it has been at detectable ﬂux levels at least for a few years. However, they are usually variable by up to several factors of tens and vary on a timescale of milliseconds to months. On the other hand, transient X-ray binaries are characterized by their long periods of inactivity, lasting from months to years, that is interrupted by short outbursts, during which their X-ray brightness increases by several orders of magnitude (Bradt et al. 2000; Figure 1.2). Most of the transient X-ray binaries become very faint or go below the detection limit of most of the X-ray telescopes during their quiescent phase. However, they cover an enormous dynamic range in the

33 38 1 luminosity (typically 10 to 10 erg s− ), which allows models for the emission region and the accretion processes to be tested over a wide range of mass accretion rates. The transient high mass X-ray binaries are especially important to study properties of the 1.2 X-ray binaries 7

V0332+53 80

(Counts per second) 20

0 100 200 300 400 500 600 700 (MJD − 53000)

Figure 1.2: Lightcurve of a transient X-ray binary pulsar V0332+53 observed with Rossi X-ray timing explorer - All Sky Monitor(ASM), clearly showing an X-ray outburst during 1 November 2004 - March 2005. The RXTE -ASM count rate reached nearly 90 counts s− 1 as compared to approximately 1 count s− during its quiescent state. companion stars (Psaltis, 2006). The transient episodes in the X-ray binaries are understood to occur due to instability in the accretion disk, or a mass ejection episode from the companion star like in

Be/X-ray binaries. Most of the LMXBs with black hole are observed to be transient while majority of the LMXBs with neutron star are persistent. Among the neutron star LMXBs, all millisecond accreting pulsars are transients ( nine sources). ∼ Among HMXBs, Be/X-ray binaries are mostly transient while the early discovered supergiant X-ray binaries are persistent. The study of a large sample of Be/X-ray binaries have revealed that the orbit of most of these systems are highly eccentric (e ∼ 0.3 - 0.5) and show periodic outbursts, usually associated with the periastron passage of compact object (mostly neutron star) through the highly dense regions around the Be star (e.g., A 0535+262; Motch et al. 1991). These systems also show aperiodic outbursts, 8 1. Introduction which arises due to sudden mass ejections from the Be star. A typical X-ray outburst observed in high mass X-ray binaries is shown in Figure 1.2. In supergiant X-ray binaries, the compact object accrete from a nearly constant strong radiative wind of the OB supergiant, in a nearly circular orbit, making them persistent in nature (Bildsten et al.,

1997). Over last 10 years, a group of Be/X-ray binaries have been discovered which are almost persistent in nature (Reig and Roche 1999, Pfahl et al. 2002, Kaur et al. 2008d). These sources seem to be diﬀerent from the transient Be/X-ray binaries as they are mostly

36 1 found at the X-ray luminosities below 10 erg s− while the X-ray luminosities of tran-

38 1 sient Be/X-ray binaries go up to 10 erg s− . The persistent nature of these sources is associated with their nearly circular orbit and long orbital period. However, the detail study of these sources has yet to be done and is taken up in this thesis (Chapter 3).

Another new class of X-ray binaries have come up very clearly in last few years called Supergiant Fast X-ray Transients (SFXTs). These X-ray binaries are transient in nature as they display fast X-ray outbursts which lasts from 3 to 8 hrs and have su- ∼ pergiant companions. These systems diﬀer from classical wind-fed supergiant X-ray bi-

36 1 naries, whose X-ray luminosity is variable but always detectable around 10 erg s− . ∼ Quiescent ﬂuxes of SFXTs have been near the sensitivity limit of focusing observato-

32 33 1 ries, with values or upper limits in the range of 10 to 10 erg s− (Negueruela et al., 2006). The exact physical mechanisms behind the transient episodes in supergiants X-ray binaries have not yet been understood (in’t Zand 2005; Prinja et al. 2005). The source of energy in the X-ray binary systems is understood to be accretion of matter onto a compact star (neutron star/black) from its binary companion and is discussed in the following Section for a neutron star accreting matter from a spherically symmetric distribution of gas. 1.3 Basics of accretion 9

1.3 Basics of accretion

We assume a spherically symmetric distribution of gas accreting onto a neutron star as depicted in Figure 1.3, wherein a parcel of gas with mass m accreting onto a neutron star, will have a gravitational potential energy

GM m E = n (1.1) Rn the Mn and Rn in the above equation are the mass and radius of the neutron star. This energy will be dissipated at the rate

GMnm˙ Lacc = (1.2) Rn wherem ˙ is the mass accretion rate and G is the gravitational constant. As can be seen in equation 1.2, the accretion luminosity, Lacc depends very strongly on the mass-radius ratio (Mn/Rn) of accreting star; thus the accretion luminosity in case of neutron star can

17 be very high. A neutron star of mass, Mn accreting with mass accretion rate,m ˙ = 10 g/s

9 (or 1.6 10− M /yr) can reach a luminosity × ⊙

37 1 1 L = 1.8 10 m˙ M R− erg s− (1.3) acc × 17 n 6

17 6 wherem ˙ 17 = m˙ /(10 g/s), R6 = Rn/(10 cm), Mn = Mn/(1.4 M ) ⊙

1.3.1 The Eddington luminosity

There is a limit on the highest accretion luminosity and is deﬁned as the Eddington luminosity. A balance between the gravitational pressure of the accreted matter and the radiative pressure of the neutron star sets an upper limit on accretion luminosity of a star. For luminosity above which the radiative pressure of the neutron star exceeds the gravitational force, it blows the material away and the accretion stops, is called Eddington luminosity (Longair, 1994). The radiative force on the ionized matter near the compact object is given by L σ F = T (1.4) rad 4πr2 c 10 1. Introduction

where σT is the Thomson cross section and c is the velocity of light. Equating the gravitational force of accreted matter by a neutron star with the radiative force deﬁned above, we get

4πmcG 38 Mn 1 LEDD = Mn = 1.5 10 erg s− (1.5) σT × M ! ⊙ The Eddington luminosity also sets a limit on the rate at which the neutron star can accrete matter, i.e.,m ˙ which can be calculated by setting the accretion luminosity (Eq. 1.2) equal to the Eddington luminosity (Eq. 1.5).

4πcmR R = n = 8 1 m˙ EDD 1.5 10− 6 M yr− (1.6) σT × 10 cm ⊙

The compact object in an X-ray binary can accrete matter from its binary companion through a number of mass transfer processes e.g., Roche lobe overﬂow, stellar wind or the mass ejections from the Be star and the same are discussed in the forthcoming Sections.

1.3.2 Stellar wind accretion

The majority of the observed X-ray pulsators have companions of early spectral type (O or B). Such systems can be very luminous and many of them are the ﬁrst Galactic X-ray sources discovered (e.g., Cen X-3, Cyg X-1 and Vela X-1). The stellar wind of the early-

6 5 1 type companion is both intense, with mass loss rates 10− - 10− M yr− , and highly ⊙ supersonic. The velocity of the stellar wind of an early-type star at a distance r from the center of it is deﬁned as

1/2 2GMe 3 1 vw(r) vesc(Re) = 10 km s− (1.7) ∼ Re ! ≈

Here Me and Re are the mass and radius of the early-type star, vw is the velocity of the stellar wind and vesc is the escape velocity of the stellar wind at the surface of the early-type star. For typical parameters, vw is generally a few thousand km/s which greatly exceeds the sound speed, c 10(T/104K) km/s. If the orbital velocity of neutron star about its s ∼ 1.3 Basics of accretion 11

Figure 1.3: Left: compact star accreting from the stellar wind of an early-type companion Right: closer view of bow shock produced near the compact object, along with the accretion radius in stellar wind accretion. This ﬁgure is taken from Frank et al. (2002).

1 companion is vn, the wind sweeps past the neutron star at an angle β tan− (vn/vw) to the line of centers with the speed 2 2 1/2 vrel (vn + vw) (1.8)

The wind particles which pass so close to the neutron starthat their kinetic energy is less than the gravitational potential energy will be captured and ultimately accreted by the neutron star (Frank et al., 2002). This capture will occur within a cylindrical region with axis along the relative wind direction (vrel) and radius

2GMn racc 2 . (1.9) ∼ vrel where Mn is the mass of the neutron star. Figure 1.3 (Left) shows schematic diagram of the wind accretion by a neutron star from an early-type star. A closer view of the compact object in a stellar wind accretion in shown in Figure 1.3 (Right).

Thus, the net amount of gas captured by the neutron star is given as

2 m˙ = πraccρvrel (1.10) where a is the orbital separation and ρ is the wind density near the accretion radius. 12 1. Introduction

The Kepler’s third law and the continuity equation assuming a spherically symmetric wind and steady mass loss are given as

G(M + M ) v2 = n e (Kepler’s law) (1.11) n a

2 M˙ e = 4πa ρvrel (Continuityequation) (1.12)

If we neglect the orbital motion of neutron star about its companion and substi- tute racc (from Eq. 1.9) and ρ (from Eq. 1.12) in Eq. 1.10, we get

m˙ (GM )2 G2 M2 = n n 2 2 2 2 2 4 (1.13) M˙ e a (vn + vw) ∼ a vw and substituting vw from Eq. 1.7, the above equation will become

m˙ 1 M 2 R 2 n e (1.14) ˙ Me 4 Me ! a For parameters typical of the X-ray binaries, Eq. 1.14 implies accretion rates

4 3 of order 10− - 10− of the mass-loss rate M˙ e. The mass-loss rates in supergiant X-ray

6 5 1 binaries are of the order of (10− - 10− M yr− ), thus the sources powered in this way ∼ ⊙ 37 1 are observable with X-ray luminosities 10 erg s− (Frank et al., 2002). ∼

1.3.3 Roche-lobe overﬂow

In case of the LMXBs, the mass accretion takes place through Roche lobe overflow as stellar wind is usually not strong in the low-mass stars and/or late-type stars in these systems. In the Roche lobe approximation, the gravitational field generated by two stars, moving in a circular orbit around their center of mass, is approximated by the two point masses. It is also assumed that the stars co-rotate with the binary system. Under these conditions, the potential (gravitational and centrifugal) and its corresponding equipotential surfaces can be derived. Close to each object, the potential is dominated by the gravitational potential of the star, thus the surfaces are almost spherical. As one moves away from the individual stars, two effects start dominating in the binary system. a) the tidal effect, which causes an elongation of equipotential surface in the direction of the companion and b) flattening 1.3 Basics of accretion 13 due to the centrifugal force. As a result, the equipotential surfaces are elongated in the way that their largest dimensions are along the line of centers (Frank et al., 2002). The most important equipotential surface, from the point of view of binary star evolution is a figure-of-eight cross section which passes through its inner Lagrangian point

L1. In three dimensions, this critical surface has a dumb-bell shape; the part surrounding each star is known as a Roche lobe, and is shown in Figure 1.4 for binary stars of masses 15 M anda7M . The importance of Roche geometry lies in the fact that the stars which ⊙ ⊙

ﬁll their Roche lobe starts transferring matter to the other star through L1.

1.3.4 Accretion from Be stars

Apart from the above two steady mass-loss processes, steady wind from early-type stars and Roche lobe overﬂow, there exists a third type of mass-loss process : the irregular mass ejections from a rapidly rotating B-emission stars. Such B-type stars show, at irregular time intervals, the outbursts of equatorial mass ejections which produce a rotating ring of matter around the star, giving rise to sudden appearance of hydrogen emission lines in the optical spectrum. The mass-ejections in Be star are intrinsic to it and takes place independent of whether it is in a binary system or not. The star can appear as a normal

B-star for many years and suddenly go through a B-emission phase, becoming a Be star for periods ranging from a few weeks to years (Slettebak and Snow, 1987), while others

3 1 can be Be stars almost permanently. These stars usually have a high velocity (10 km s− )

2 1 low-density polar region winds and a low-velocity (10 km s− ) high-density equatorial regions winds (Dachs et al. 1986; Waters 1986). If such a Be star is in a binary system with a neutron star as a companion, then the sudden mass-ejections from the Be star can result into bright X-ray outbursts, when a fraction of the mass ejected from a Be star gets captured by a neutron star. These systems also undergo large X-ray outbursts when the neutron star in them passes through the dense equatorial wind of the Be star. The diﬀerent outbursts seen in Be/X-ray binaries are described in Section 1.4.5. However, the accretion 14 1. Introduction

Figure 1.4: A cross-section in the equatorial plane of the critical equipotential surfaces in a binary. The symbol ( ) represents the center of mass of the system. The thick curve × crossing through L1 is the Roche lobe. The other Lagrangian points of the binary system are marked as L2,L3,L4,and L5 (from Tauris and van den Heuvel 2006). 1.3 Basics of accretion 15 mechanism in Be/X-ray binaries is not far from the case of stellar wind accretion (Section

1.3.2) and the same description can be assumed to work as a ﬁrst approximation. The relation in Eq. 1.9 can be used for the accretion radius. The incoming matter from the companion star through one of the above mentioned processes cannot directly fall onto the compact object due to its initial angular momentum and thus forms an accretion disk around the compact object. In the following Section, we have discussed the formation of accretion disk in X-ray binaries.

1.3.5 Disc formation and boundary layer

Accreting matter forms a disk when its specific angular momentum J is too large for it to hit the accreting object (e.g., neutron star of mass Mn) directly. This typically requires that the circularization radius J2 Rcirc = (1.15) GMn larger than the effective size of the accretor. In X-ray binaries, the accretor is a neutron star or a black-hole, and the condition always holds if accretion is via Roche lobe overflow (Frank et al., 2002). For typical orbital parameters, we obtain

M 1/3 P 2/3 R 3.5 109 e orb cm (1.16) circ ≥ × M ! 1 hr ⊙ which is much larger than the radius of the neutron star ( 106 cm). This outcome is less ∼ clear if wind accretion is at work, as J and Rcirc are much lower. In that case, the typical size of Rcirc is 1/3 2/3 6 Me Porb Rcirc 1.3 10 cm (1.17) ≈ × M ! 10 d ⊙

For an early type star of mass, Me = 10 M , this value comfortably exceeds the ⊙ radius of the neutron star. Usually it is not easy to decide whether the disk accretion or the stellar wind accretion take place in these systems. However, it is a generally accepted scenario that a small accretion disk can form occasionally both in prograde and retrograde direction, such that the net sign of J changes in an erratic manner (Frank et al., 2002). 16 1. Introduction

Figure 1.5: Left: Perspective view of an idealized accretion disk showing the ﬂuid ‘spiraling’ in. Right: Accretion disk around a magnetized neutron star. The magnetic lines shown represent schematically the boundary of the magnetosphere. This Figure is taken from Frank et al. (2002).

Since the orbit of lowest energy for a given angular momentum is a circle, matter will follow a sequence of circular orbits about the compact accretor (Figure 1.5). The agency for both energy and angular momentum loss in an accretion disk is called viscosity and is assumed to be a form of the magneto-rotational instability (Balbus and Hawley 1991). The required Kepler’s angular velocity cannot be maintained at the inner edge of the disk if it is to join smoothly to a non-magnetic accreting star spinning at below the break-up velocity. The region over which gas moving at Keplerian velocities in the disk is decelerated to match the star angular velocity is called boundary layer (BL). The picture of the boundary-layer accretion described above can only be relevant if the disk extends right down to the surface of the accreting star. Quite often this is not the case, like in HMXBs, the neutron star possesses magnetic ﬁelds of the order of 1012 G and truncate the accretion disk at a distance far away ( 500 - 1000 km) from the neutron ∼ star surface. The detailed accretion process onto magnetized objects is explained in the following Section. 1.3 Basics of accretion 17

1.3.6 Accretion onto magnetized objects

The presence of a magnetic field plays an increasingly important role in the dynamics of the gas flow as it approaches the surface of the compact star leading to the disruption of the accretion disk. Thus the term magnetosphere is usually defined as that volume within which the magnetic field strongly effects the dynamical properties of the in-falling flow, such as the trajectory, energy and angular momentum (Vasyliunas 1979, Lamb 1979, Henrichs 1983).

The characteristic distance where the magnetic field interrupts the accretion flow is called the Alfven´ radius, rA and can be estimated by balancing the ram pressure of an incoming matter to the magnetic pressure exerted by neutron star (Frank et al. 2002 and Carroll and Ostlie 1996) as B2 ρv2 = (1.18) 8π where ρ is the density of the in-falling matter. For a dipole-like magnetic field, the field strength B varies roughly as B µ/r3 at radial distance r from the star of radius R ; here ∼ n

µ is a magnetic moment of the star speciﬁed by the surface magnetic ﬁeld strength Bn at r

3 = Rn (µ = BnRn). Thus our magnetic pressure term becomes

µ2 P = (1.19) mag 8πr6

For simplicity, we assume that the neutron star is accreting from a spherically symmetric distribution of gas, in which case the velocity of the accreted material is given approximately by its free-fall velocity from rA;

2GM v = (1.20) r rA and for such a spherically symmetric gas, the mass accretion rate is given by

2 m˙ = 4πr ρvr (1.21)

where vr and ρ is the radial velocity and density of the accreted gas. 18 1. Introduction

Substituting the free fall velocity v,m ˙ and B from above equations and rearrang- ing the terms in 1.18, the expression of rA becomes

8 2/7 1/7 4/7 r = 5.1 10 m˙ − M− µ cm (1.22) A × 16 30

30 3 where µ30 is µ in units of 10 G cm . It is often convenient to replacem ˙ in the above equation in terms of the accretion luminosity (Eq. 1.2), which is more directly related to the observational quantities like X-ray ﬂux in these sources. Thus

8 1/7 2/7 2/7 4/7 r = 2.9 10 M R− L− µ cm (1.23) A × 6 37 30

33 1 37 1 where L33 = Lacc/10 erg s− and L37 = Lacc/10 erg s− in the above equations. Besides typical accretion length scales, racc and rA, there is a third characteristic length scale, namely the co-rotation radius which is deﬁned as the distance at which neutron star rotation velocity matches the Keplerian one. This happens where the centrifugal force just balances the local gravity, i.e.,

GM 1/3 = n = . 8 1/3 2/3 rco 2 1 5 10 Mn Ps cm (1.24) ωn ! × where Ps is the spin-period (rco is independent of the accretor size) and ωn is the neutron star spin frequency. The minimum requirement for a X-ray binary system to show a signiﬁcant magnetic behaviour is r R . It is thus possible to distinguish several likely physical regimes A ≥ n depending on the relative size of racc, rA, and rco as well as the magnetic ﬁeld strength (Stella et al., 1986). However, the most of the magnetic neutron star X-ray binaries are found in the following two regimes.

Accretor regime : r > r and r > r • acc A co A

In these conditions, the captured material flows from the accretion radius down to the magnetosphere radius, where it is stopped by a collision-less shock. The material then penetrates the magnetosphere till a point where it is forced to follow the magnetic field 1.3 Basics of accretion 19 lines towards the star surface (direct accretion). For high magnetic field (B 1010 G) mat- ≥ ter falls in the small region located near the polar cap(s) of the neutron star and produces the X-rays from that region. With the spin of the compact object, this region comes along the line-of-sight of Earth and goes off, hence we see pulsations with the spin period of the compact object (neutron star or white dwarf). If B 109 G, matter can get through ≤ the magnetosphere without being forced to follow the magnetic field lines and accretion occurs at all star latitudes.

Propeller regime : r > r > r • acc A co

The material which penetrates through the accretion radius is stopped at the magnetospheric boundary rA and cannot advance any further because of the drag exerted by the magnetic ﬁeld is super-Keplerian (rA > rco). Some or all the material might be ejected beyond the accretion radius via the propeller mechanism (centrifugal inhibition of accretion). If the wind material accumulates at or nearby the magnetosphere more rapidly than the ejection rate, a build up of material outside the magnetosphere may occur (Maraschi et al., 1983). In this thesis, we have mainly worked on high magnetic ﬁeld X-ray binaries.

The magnetic field of the compact object in these systems is of the order of 1012 G and is expected to truncate the accretion disk at a distance of 500 - 1000 km away from the neutron star surface. The matter present at the inner accretion disk is thus forced to travel along the magneticfield lines and finally hits the surface of the neutron star at the magnetic poles which emits in X-rays. Such a system is expected to give rise to pulsations when neutron star spins around its own axis. Most of the neutron star high mass X-ray binaries are observed to be X-rays pulsars and are also called accretion powered X-ray pulsars. A schematic diagram of an accretion powered X-ray pulsar is shown in Figure 1.6. In this

Section, we have described the observational properties of these pulsars which mainly include their timing and spectral properties. 20 1. Introduction

Figure 1.6: A schematic diagram of an accretion powered X-ray pulsar. The spin axis and the magnetic axis are misaligned (thus: a pulsar). X-rays are emitted along the magnetic ﬁeld axis as the pulsar accretes near its magnetic poles. We have also marked the region in accretion disk where quasi periodic oscillations (refer to Section 1.4.4) are assumed to originate. (Credit : http://lheawww.gsfc.nasa.gov.)

1.4 Accretion powered X-ray pulsars

At present, altogether more than 100 X-ray pulsars are known in our Galaxy and the Magellanic clouds. These systems show a variety of timing properties (periodic, quasi- periodic and aperiodic) and spectral properties which help us to understand the physics of these exotic systems. In this Section, we have explained the common properties of these systems, mainly their timing and spectral properties. 1.4 Accretion powered X-ray pulsars 21

1.4.1 Pulse periods

X-ray pulsars have been discovered with pulse periods ranging from 70 ms to 8000 s (Liu et al. 2006; Liu et al. 2007) and a change in the intrinsic pulse period has been seen in almost all of them. The most common pulse period behaviours seen in these systems are as follows :

a linear decrease of pulse period with time with erratic variations around the trend •

a steady increase in pulse period •

no long term trend, only a random walk in the period •

Spin-up In a number of neutron star X-ray binaries, the spin-period is found to be increasing steadily or spinning-up with time scale which vary from 100 yr to 100,000 yr. This spin-up is understood in terms of torques exerted by the matter accreting onto the neutron star and is given by

N = m˙ GMnrm (1.25) p

The rate of change of intrinsic spin period, P˙ is related to the X-ray luminosity, LX and the physical properties of the neutron star itself (Ghosh and Lamb, 1979) as follows

P˙ P L 6/7 = 5 X 1 3 10− f 37 1 yr− (1.26) P − × 1s 10 ergs− ! where the dimensionless function f is expected to be of order of unity for a neutron star.

Spin-down

Contrary to spin-up behaviour, a number of X-ray pulsars also show an increase in pulse period or spin-down behaviour which can be explained as follows. An equilibrium is reached when the neutron star magnetosphere co-rotates with the inner edge of the disk, i.e., when rm is equal to rco, and the spin-up torque diminishes. Close to co-rotation, 22 1. Introduction the ﬁeld lines in the transition zone are swept backwards, and a negative torque is exerted on the neutron star. At this stage, the neutron star will be spun-down, even though accretion continues (Nagase, 1989).

Period ﬂuctuations

37 1 The X-ray binary pulsars with OB companions and with L 10 erg s− show X ≤ no overall trend in the pulse period evolution during their lifetime. The large fluctuations in the pulse period are usually seen in these sources even on the short timescales of a few days. It is assumed that the presence of inhomogeneities in the wind causes the instabilities in the accretion flow, which results in the oscillatory motion of the pulse period. Also it is assumed that these flip-flops occur because the accretion disk briefly forms, disappears and forms again (White et al., 1995).

1.4.2 Pulse proﬁles

The pulse profiles of X-ray pulsars show a great variety from sinusoidal-like profiles to highly structured and assumed to be an indicator of the X-ray emission geometry in them. In many X-ray pulsars, the pulse profiles can be reproduced by simply assuming a beam pattern and varying the geometrical aspect of the rotation axis, the magnetic axis and the line of sight (e.g., Wang and Welter 1981). If the angles of the magnetic axis and the line of sight with respect to the rotation axis are α and β, respectively, then depending upon

π π whether α + β < 2 or > 2 or in other words, if one or both the magnetic poles are visible, they show single or double pulse proﬁle. If the beam has a maximum either along or perpendicular to the magnetic axis, the beam can be described as a pencil or a fan beam, respectively depending upon if the matter falls onto the magnetic pole in an accretion column or along the stellar surface (slab accretion geometry), shown in Figure 1.7. However, the preferred beaming direction from a hot spot on the neutron star surface primarily depends on whether a stand-oﬀ shock is present (e.g., Nagel 1981a, Nagel

37 1 1981b). For high X-ray luminosities (L 10 erg s− ) a radiative shock is expected to X ≥ 1.4 Accretion powered X-ray pulsars 23

Figure 1.7: Accreting pulsar models. Left: The pencil beam model. Right: The fan beam model. M and ω represent the magnetic axis and spin axis of an X-ray pulsar respectively. form and as a result of that, the photons will escape preferentially from the sides of the high density post-shock accretion column, giving rise to a fan beam pattern (Basko and Syunyaev 1976; Wang and Frank 1981). However, in some circumstances, a pencil beam may still emerge in some energy bands (Nagel 1981b). For lower X-ray luminosities (LX

37 1 10 erg s− ), the infalling matter might be decelerated in a collisionless shock above the ≤ polar cap possibly arising from plasma instabilities, or by coulomb and nuclear collisions at the neutron star surface (Basko and Sunyaev 1975 ; Kirk and Galloway 1981) and the emission region will be located in a thin layer on the neutron star surface. In such a case, the radiative transfer effects in the strong magnetic field will favor photons escaping in the direction of the field lines, therefore giving rise to a pencil-beam pattern (Meszaros et al., 1983).

The simple geometrical model explained above can reproduce the observed variety of pulse profiles (Wang and Welter, 1981). In some cases, the offset of the magnetic axis and/or two polar cap emission regions with different sizes are required to give an asymmetric pulse profile (Leahy, 1991). A change of profile can also occur if the configuration of the X-ray emitting region changes. In some cases, the pulse profile evolution is 24 1. Introduction

Figure 1.8: The folded 1 - 10 keV light curves of EXO 2030+375 showing the evolution of the pulse profiles during the decay of an outburst (from Parmar et al. 1989). 1.4 Accretion powered X-ray pulsars 25 caused by dominant beam changing from a fan beam to a pencil beam configuration with a decrease in luminosity. In particular, the transient X-ray pulsar EXO 2030+375 (42 s pulse period) offered to study the pulse profile of an individual source (without uncertainties related to the different distance and magnetic field strength of different sources) for

36 38 1 luminosities ranging from 10 to 10 erg s− (Parmar et al. 1989). A drastic change in ∼ the shape of the pulses were observed in this system with the decrease in luminosity. In particular, the broad peak at the pulse phase 0.6 - 0.9 became gradually less pronounced

36 1 as the luminosity decreased and it virtually disappeared for luminosities of 10 erg s− ∼ (Figure 1.8). On the contrary, the feature centered around a phase 0.4 increased as the ∼ 36 1 luminosity decreased during the outburst and became dominant around 10 erg s− . ∼

1.4.3 Orbital Modulations

The orbital periods of HMXBs have been measured in a range from4.8hrto262dandfor LMXBs range from 0.19hrto 48 d (Liu et al. 2006; Liuet al. 2007) and can be determined from the observations of one or more of the following features : eclipses, smooth periodic modulations, periodically recurring X-ray absorption dips, periodically recurring transient X-ray outbursts, pulsar arrival time variations, radial-velocity variations and pulse-orbital beat period (White et al., 1995). More than half of the orbital period measurements of X-ray binaries have come from their optical observations and the reason behind it is as follows. An X-ray modulation is seen only when the system is viewed close to the orbital plane, so as to view the eclipses by the companion star and absorption by the material in the accretion ﬂow while an optical modulation is also seen from the rotating face of an X-ray heated or tidally distorted companion star. Due to this reason, the optical modulations are detected over a greater range of inclination angles (White et al., 1995).

Among HMXBs, the supergiant systems typically are eclipsing and show extreme intensity and absorption variability on all timescales. The shorter orbital period 26 1. Introduction supergiant systems have circular orbits, whereas the longer period systems are eccentric

(White et al., 1995). The Be/X-ray binaries are often bright pulsating transient sources (Maraschi et al., 1976) and the orbital period in most of them are measured from Doppler variations in the pulse period and usually have long orbital periods ranging from a few tens of days to a few hundred days. These systems rarely show eclipses.

1.4.4 Quasi periodic oscillations

Quasi periodic oscillations (QPOs) in X-ray binaries are usually associated to the motion of inhomogeneously distributed matter in the inner accretion disk, and thus assumed to provide useful information about the interaction between accretion disks and the compact object. These oscillations have been discovered in a number of X-ray binaries and usually lies at the kHz frequencies in LMXBs and between a few mHz to a few Hz in HMXBs. The QPOs associated with neutron stars of similar magnetic fields are observed to share common properties and thus we have divided the further discussion on QPOs into two parts : QPOs in low magnetic field X-ray binaries and QPOs in high magnetic field X-ray binaries.

QPOs in low magnetic field X-ray binaries (B 109 G) ≤ Low-magnetic field neutron star X-ray binaries form a clear sub-division into two classes : Z sources and Atoll sources, on the basis of their position in color-color diagram (CD; hard color vs soft color) and hardness-intensity diagram (HID; hardness vs intensity). Z sources are the most luminous sources and they accrete at an appreciable fraction of the Eddington critical rate (perhaps 0.5 - 1 LEDD). These sources trace out roughly Z-shaped tracks in the CD/HID. Atoll sources, many of which are X-ray burst sources, cover a much wider range in luminosities (from perhaps 0.001 LEDD all the way up to the range of the Z sources), trace out a well-defined curved banana branch in the CD/HIDs along which the sources move back and forth with no hysteresis on timescales of hours to a day or so. There also exists a third class of LMXBs called weak LMXBs 1.4 Accretion powered X-ray pulsars 27

Table 1.2: Quasi periodic oscillations in accretion powered X-ray pulsars

Source Spin frequency (mHz) QPO frequency (mHz) Reference 4U 1907+09 2.27 55 a XTE J1858+034 4.5 111 b A 0535+26 9.71 27-72 c EXO 2030+375 24 187-213 d LMCX-4 74 0.65-1.35,2-20 e 4U1626–67 130 1,48 f CenX-3 207 35 g V 0332+53 229 51 h 4U 0115+63 277 2,62 i HerX-1 807.9 8,12,43 j SMCX-1 1410 60? k GRO1744–28 2140 40000 l XTEJ0111.2–7317 32 1270 m

Reference : (a). in ’t Zand et al. 1998 (b). Paul and Rao 1998 (c). Finger et al. 1996 (d). Angelini et al. 1989 (e). Moon and Eikenberry 2001a (f). Shinoda et al. 1990, Chakrabarty 1998, Chakrabarty et al. 2001, Kaur et al. 2008b, Orlandini et al. 1998 (g). Takeshima et al. 1991, Raichur and Paul 2008 (h). Takeshima et al. 1994 (i). Heindl et al. 1999 (j). Boroson et al. 2000, Moon and Eikenberry 2001b (k). Wojdowski et al. 1998 (l). Zhang et al. 1996 (m). Kaur et al. 2007b

(LEDD < 0.01), which comprise the overlapping groups of faint burst sources, millisecond pulsars and low-luminosity transients (van der Klis 2006b, van der Klis 2006a). QPOs at kilo Hertz (kHz) frequency have been seen in nearly all Z sources, Atoll sources as well as weak LMXBs, including the millisecond pulsars. It has been observed that generally two QPO peaks (twin peaks ) occur in the power spectrum and move up and down in frequency together in the 200 - 1200 Hz range in correlation with source state. The higher frequency of these two peaks is called the upper kHz QPO frequency,

νu, and the other lower kHz QPO frequency, νl. The several hundred hertz peak separation between the peak centroids ∆ν = νu - νl is typically within 20% of the neutron star spin 28 1. Introduction frequency, or half that depending on source, and usually decreases by a few tens of hertz when both peaks move up by hundreds of hertz (van der Klis, 2006b). Weak sidebands in the kHz domain have also been reported in some kHz QPO sources. A number of models have been suggested to explain the generation of QPOs in LMXBs which involve orbital motion of inhomogeneities in the disk at one of the kHz QPO frequencies, like beat frequency model (Alpar and Shaham 1985, Shibazaki and Lamb 1987), relativistic precession models (Stella and Vietri, 1998), relativistic resonance models (Kluzniak and Abramowicz, 2001). However, none of the models seem to be explaining all the properties of QPOs in LMXBs.

QPOs in high magnetic field X-ray binaries (B 1011 G) ≥ QPOs have been discovered in almost a dozen high magnetic field X-ray binaries in the frequency range of a few mHz to a few Hz (Psaltis, 2006), listed in Table 1.2. The QPOs are known to occur sporadically only in a few percent of the X-ray observations in most of the high magnetic field X-ray pulsars. For example, QPOs are detected only in 15% of the out-of-eclipse observations of Cen X-3 (Raichur and Paul, 2008). Our independent investigation of RXTE-PCA lightcurves of several persistent sources show that the QPOs are quite rare. Exceptions to this are the transient sources, such as 3A

0535+262 (Finger et al., 1996) and XTE J1858+034 (Paul and Rao, 1998), which showed QPOs during most of the observations made during their outbursts. The most widely used models for the QPO generation mechanism are : Keplerian frequency model (KFM) and beat frequency model (BFM). Both the models assume the presence of inhomogeneities in the inner accretion disk. However, the main diﬀerences between the two models are brieﬂy discussed below.

Keplerian frequency model (KFM) According to this model, QPOs arise from modulation of the X-rays by inhomogeneities in the inner accretion disk at the Keplerian frequency (van der Klis, 1987). Thus, 1.4 Accretion powered X-ray pulsars 29

Figure 1.9: The power density spectrum of an accretion-powered X-ray pulsar 4U 1626– 67 showing detection of mHz quasi-periodic oscillations (from Chakrabarty et al. 2001). the frequency of QPOs is given by the frequency of the orbiting material as follows:

νQPO = νk (1.27)

where νQPO is the frequency of QPOs and νk is the Keplerian frequency of matter in the inner accretion disk. This model is applicable only if the νQPO >νs, because otherwise the system will be in a propeller regime (Section 1.3.6) and hence would not allow any inhomogeneity to sit on the accretion disk.

Beat frequency model (BFM) According to this model, the QPOs occur at the beat frequency between the orbital frequency of matter in the accretion disk at the Alfven´ radius and the stellar spin frequency (Alpar and Shaham 1985, Lamb et al. 1985). In this model, the frequency of

QPOs, νQPO is given as ν = ν ν (1.28) QPO k − s 30 1. Introduction

In other words, it is the frequency at which a particle orbiting in the disk periodically overtakes a given point on the spinning star. This model is applicable in both the situations when (νQPO >νs)or(νQPO <νs), as in both the cases, the system will be in accretor regime (Section 1.3.6). Figure 1.9 shows the detection of QPOs in a high magnetic ﬁeld X-ray pulsar 4U 1626–67 during its RXTE observations in April 1998. With the assumption that the QPOs arise at the inner accretion disk of the compact object, we can deﬁne the radius of the QPOs production area, rQPO using the Kepler’s third law as GM 1/3 r = n (1.29) QPO 2 2 4π νk ! where G is the gravitational constant, Mn is the mass of the neutron star. The radius of the inner accretion disk (rm) of the compact object having magnetic moment µ is given as

8 2/7 4/7 r = 3 10 L− µ , (1.30) m × 37 30

37 1 where L37 is the X-ray luminosity in units of 10 erg s− and µ30 is the magnetic moment in units of 1030 cm3 G.

If the QPOs are as per KFM (νk = νQPO), then we expect

ν L3/7 or ν L3/7 (1.31) K ∝ 37 QPO ∝ 37

Hence, a systematic change of the QPO frequency is predicted with the mass-accretion rate or with the ﬂux for KFM (Kaur et al., 2008b). This change in QPO frequency with the X-ray ﬂux of the system has a potential to study the evolution of accretion disk and the existing relations between the mass accretion rate and the luminosity of the system. Till now, QPOs have been detected only in about dozen X-ray pulsars and the QPOs frequency evolution could be studied in detail in a very few systems e.g., A 0535+262 (Finger et al.,

1996) EXO 2030+375 (Angelini et al., 1989) and 4U 1626–67 (Kaur et al., 2008b). 1.4 Accretion powered X-ray pulsars 31

1.4.5 Long term X-ray variabilities

It has been observed that the neutron star X-ray binaries undergo ﬂux variations on all timescales from millisecond to a few years, periodic and aperiodic, depending upon the physical mechanism at work. In the previous sections, we have already talked about short-term variabilities - pulse period, orbital period and quasi-periodic oscillations. In this section, we discuss the various long-term variabilities in neutron star X-ray binaries.

Type-I X-ray outburst The periodic X-ray outbursts seen in Be/X-rays binaries are called Type-I X-ray outbursts. Most of the Be/X-ray binaries have eccentric orbits (e 0.3 - 0.5; Bildsten ∼ et al. 1997) and the Type-I outbursts in them are understood to arise when the neutron star in them crosses the circumstellar disk of the Be star during its periastron passage, as shown in Figure 1.10. During the periastron passage, the increase in mass accretion rate

36 37 1 give rise to increase in the X-ray luminosity, which goes up to 10 - 10 erg s− . The ∼ duration of these outbursts varies from a few days to a few weeks.

Type-II X-ray outburst The aperiodic X-ray outbursts seen in Be/X-ray binaries, which show no clear orbital dependence are called Type-II X-ray outbursts. The origin of these outbursts have not yet been fully understood. However, it is assumed that a sudden mass-ejections from the Be star cause an increase in the X-ray ﬂux in these systems which lasts from a few days to a few months. The X-ray luminosity during these outbursts can reach 1037 erg ≥ 1 s− , thus sometimes these outbursts are also called ‘giant outbursts’ (Negueruela et al. 1999, Kaur et al. 2008a, Kaur et al. 2008c).

Super-orbital period Super-orbital (long) periods are known to exist in more than 20 HMXBs and

LMXBs, and their values range from 24 to 600 d (Ogilvie and Dubus, 2001). The most stable super-orbital periods have been found in Her X-1, LMC X-4, 2S 0114+650 and 32 1. Introduction

%HVWDU

H[SHOOHGPDWHULDO

QHXWURQVWDU

;UD\IOX[

3RUE

WLPH

Figure 1.10: Schematic model of a Be-star X-ray binary system. The neutron star moves in an eccentric orbit around the Be-star which is not ﬁlling its Roche lobe. However, near the periastron passage the neutron star accretes circumstellar matter, ejected from the rotating Be-star, resulting in an X-ray burst lasting several days. This Figure is taken from Tauris and van den Heuvel (2006). 1.4 Accretion powered X-ray pulsars 33

SS 433. The mechanism behind these periods is generally not well understood, though the precession of a radiation-warped accretion disk modulating the X-ray ﬂux from the compact object is the favored model (Clarkson et al., 2003).

Long term steady ﬂux variations In a few X-ray binaries like 4U 1626–67 (Kaur et al., 2008b) and 4U 1822–37

(Jain et al., 2009), the X-ray ﬂux is seen to be steadily decreasing on a timescale of years. Till now, the exact mechanism behind this variability is not known. However the depletion of accretion disk is proposed as one of the causes for the decrease in the X-ray luminosity in these sources (Kaur et al., 2008b).

1.4.6 X-ray Energy Spectra

The X-ray emission (or the X-ray energy spectra) of X-ray binaries is observed to be primarily dependent on the current accretion flow structures and the nature and magnetic field properties of the compact star. The energy spectrum of high magnetic field neutron star X-ray binaries (B 1012 G) are usually observed a flat hard power-law function with ∼ an energy index, Γ = 0.0 - 1.0 up to a high energy cut-off at 10 - 20 keV, above which the spectrum decays more steeply (White et al., 1996), and is generally interpreted as a result of inverse Compton scattering of low energy disk photons by electrons. Several high magnetic field X-ray pulsars have also shown a soft excess below 1 keV which is assumed to represent a thermal emission from hot material in the magnetosphere. The X-ray energy spectrum of these systems is sometimes also characterized with weak Fe emission lines (mostly with neutral Fe 6.4 keV and rarely with He-like Fe 6.7 keV and

H-like Fe 6.9 keV; Ebisawa et al. 1996).

1.4.7 Cyclotron absorption lines

Cyclotron resonance scattering features (CRSFs) or "cyclotron lines" are broad absorption line-like spectral features resulting from photons being resonantly scattered by electrons 34 1. Introduction

11.8 103 4U 0115+63 10-1 24.1 34.5 102 47.0 10-2

66.5 101 10-3 -s-keV 2

100 10-4 counts/s-keV

10-1 10-5 photons/cm

10-2 phase D 10-6

4 2 0

Sigma -2 -4 3 10 30 60 100 Energy (keV)

Figure 1.11: The observed (histogram) and model spectrum (solid line) of the accretion- powered pulsar 4U 0115+63 showing evidence for cyclotron lines with as many as four harmonics (from Heindl et al. 1999). out of the observers line of sight. The electron energies are quantized into Landau levels, so the energy of the cyclotron line is a direct measure of the magnetic ﬁeld strength in the scattering region. The non-relativistic cyclotron energy is given by the formula

~eB B = = . En n 11 6 12 (1.32) mc 10 G where B is the magnetic field strength and n=2,3....is the number of the harmonic. We refer to n = 1 as a fundamental harmonic, n = 2 is the first harmonic and so on. These lines have been observed in about a dozen X-ray binaries and the harmonics of the cyclotron line has also been observed in a few systems ( 4). Table 1.3 lists the X-ray binaries ∼ in which the cyclotron feature has been detected and Figure 1.1 shows the harmonics of cyclotron absorption feature detected in 4U 0115+63. It has been observed that the widths of the resonance features are correlated with the energies of the continuum cut-offs and are also proportional to their central energies and to the inferred scattering depths (Coburn et al., 2002). The central energies of these features provide us the direct measurements 1.4 Accretion powered X-ray pulsars 35

Table 1.3: Energies of Cyclotron lines in accretion-powered X-ray pulsars

Source Spin Period (s) Energy (keV) Reference 4U 0115+63 3.61 12 1 4U 1907+09 438 18 2 4U1538-52 530 20 3 VelaX-1 283 25 4 V 0332+53 4.37 27 5 CepX-4 66.2 28 6 CenX-3 4.82 28.5 7 4U 0352+309 835 29 8 XTE J1946+274 15.8 36 9 MX0656–072 160.7 36 10 4U1626–67 7.66 37 11 GX301–2 681 37 12 HerX–1 1.24 41 13 A 0535+26 105 50or110 14

References : (1). Wheaton et al. 1979 (2). Makishima and Mihara 1992 (3). Clark et al. 1990 (4). Kendziorra et al. 1992 (5). Makishima et al. 1990 (6). Mihara et al. 1991 (7). Santangelo et al. 1998, Heindl and Chakrabarty 1999 (8). Coburn et al. 2001 (9). Heindl et al. 2001 (10). Heindl et al. 2003 (11). Heindl and Chakrabarty 1999, Orlandini et al. 1998 (12). Makishima and Mihara 1992 (13). Truemper et al. 1978 (14). Kendziorra et al. 1994 of the surface magnetic ﬁelds of accreting pulsars, which can be used to constrain the models of disk-magnetosphere interaction. There exists another class of X-ray binary pulsators called ‘magnetic cataclysmic varaibles’ which are comprised of a white dwarf and a late-type star. These systems diﬀer from the X-ray pulsars (discussed above) in both timing and spectral behaviour and have been discussed in detail in Appendix A. 36 1. Introduction

1.5 Motivation and Organization of the thesis

We have broadly understood the properties of X-ray binaries, however there are many is- sues related to them which are not yet resolved. The origin of quasi-periodic oscillations is one such issue which has not yet been resolved even after their discovery in a number of X-ray binary systems (van der Klis, 2006b). It was assumed that the Be/X-ray pulsars are mostly transient in nature while the supergiant X-ray binaries are persistent till the discovery of persistent Be/X-ray binaries (Kaur et al., 2008d) and transient supergiant X- ray binaries (Negueruela et al., 2006). The discovery of these new classes have opened up the possibilities of examining these systems from diﬀerent perspectives and need to be studied in detail. Many new faint X-ray sources (persistent as well as transient) have been discovered in recent years, and require multi-waveband (optical/infrared, UV etc.) observations to understand them. Also even in the well studied class of transient Be/X- ray binaries, we have not yet known the reasons behind the origin of irregular outbursts in them and their relation to the generation of QPOs. It is usually assumed that the accretion disk forms during the X-ray outbursts in these systems, however it has not yet been strongly proved by observations.

In this thesis, we have tried to look at both transient and persistent X-ray binaries through two diﬀerent aspects namely : QPOs and multiwavelength observations. As QPOs are assumed to arise due to inhomogeneities at the inner accretion disk radius, their study give us information about the processes near the neutron star and their frequency evolution with time and luminosity give us information about the system as a whole. QPOs particularly in HMXBs have a potential to probe the accretion disk properties in detail with the existing QPOs models which require the detection of pulsations and the measurement of magnetic ﬁelds, which are still lacking in LMXBs. With the help of QPOs, we could study the accretion disk properties in one transient X-ray binary pulsar XTE J0111.2–7317 and in another long term variable X-ray binary pulsar 4U 1626–67. With the help of multiwavelength observations, a transient X-ray binary IGR 1.5 MotivationandOrganizationofthethesis 37

J01583+6713 is studied during its X-ray outburst to investigate the accretion disk properties and a group of seven faint and persistent X-ray binaries are studied to ﬁnd their true nature. This thesis contains seven chapters. In Chapter 2, we have described various ground based and space based telescopes used and diﬀerent data reduction techniques adopted to derive the physical parameters of the X-ray binary systems. The rest of the thesis is divided into 2 parts - Part I (Chapters 3 & 4) - ‘True nature of X-ray binaries’ and Part II (Chapters 5 & 6) - ‘QPOs in X-ray binaries’

Chapter 3 deals with the multiwavelength observations (X-ray and near infrared) to identify and study a newly discovered class of X-ray pulsars which are expected to be persistent Be/X-ray pulsars in nature. In Chapter 4, we have described the detailed study of a newly discovered transient X-ray binary IGR J01583+6713 with the help of optical and X-ray observations. Chapter 5 deals with the detection of QPOs in a transient X-ray binary pulsar XTE J0111.2–7317 during its X-ray outburst in November 1999 and in Chapter 6, we have described a long term QPOs frequency evolution in 4U 1626–67 to study the recession of accretion disk in this source. The summary of the thesis and future prospects are given in the Chapter 7. 38 1. Introduction Chapter 2

Observations and Data Analysis

2.1 Introduction

In this thesis, we have studied the X-ray binaries using observations obtained from various ground based and space based observatories. In the ﬁrst part of the thesis, we have used multiwavelength (X-ray, optical and near-infrared) observations to study transient and persistent high mass X-ray binaries. The optical observations for this work were obtained using two Indian telescopes namely the 1.04-m Sampurnanand Telescope (ST) at Aryabhatta Research Institute of Observational Sciences (ARIES), Nainital and the Himalayan Chandra Telescope (HCT) at Indian Astronomical Observatory (IAO), Hanle.

The X-ray observations for this work was obtained from the X-ray observatories Chandra, XMM-Newton and Swift for the position measurement of our sources with a few arcsecond accuracy and X-ray timing and spectral analysis. We borrowed the near-infrared (NIR) observations for persistent X-ray binaries from ESO - New Technology Telescope (NTT) archive and Two Micron All Sky Survey (2MASS) point source catalog. In the second part of the thesis, we have mainly used archived Rossi X-ray Tim- ing Explorer (RXTE) data to investigate the quasi periodic oscillations (QPOs) in the accreting X-ray binary pulsars and the archived data from the EXOSAT, Advanced Satellite

39 40 2. Observations and Data Analysis for Cosmology and Astrophysics (ASCA), BeppoSAX, Swift and XMM-Newton observatories to study the temporal evolution of QPOs. The present Chapter deals with the telescopes used for the multi-wavelength observations and the data reduction techniques followed to analyze these observations and is organized as follows : Section 2.2 describes the optical telescopes used for the present work and also discusses the optical photometry and spectroscopic data reduction techniques. In Section 2.3, we have described the NIR telescopes used and data reduction procedures followed for NIR waveband photometry and in the last Section 2.4, we have described the X-ray observatories used and discussed the X-ray timing and spectral analysis techniques.

2.2 Optical observations

2.2.1 Telescopes

The optical photometric follow-up observations of transient X-ray binaries were carried out using the 1.04-m ST at ARIES and optical spectroscopic observations were carried out using the 2.01-m HCT at IAO, Hanle. The basic parameters of ST and HCT are given in Table 2.1

2.2.1.1 1.04-m Sampurnanand Telescope

The 1.04-m Sampurnand Telescope (ST) is a two-pier equatorial English mounting telescope at ARIES (Longitude: 79◦27′24′′ E, Latitude: 29◦21′42′′ N, Altitude 1951 m), ∼ equipped with a Ritchey-Chretien optics. For the present work, we have used CCD chip of size 2048 2048 pixels available at the Cassegrain focus of ST (refer to Table 2.2). ×

2.2.1.2 2.01-m Himalayan Chandran Telescope

The 2.01-m aperture optical-infrared Himalayan Chandran Telescope (HCT) is installed at IAO, Hanle (Longitude: 78◦57′51′′ E, Latitude: 32◦46′46′′ N, Altitude 4500 m). ∼ 2.2 Optical observations 41

Table 2.1: Parameters of the two Indian optical telescopes.

Telescope Sampurnanand Himalayan Chandran Telescope Telescope Place ARIES IAO Nainital,India Hanle,India

Longitude 79◦27′24′′ E 78◦57′51′′ E

Latitude 29◦21′42′′ N 32◦46′46′′ N Altitude 1950m 4500m System Ritchey-Chretien Ritchey-Chretien Focal length Cassegrain : f/13 Cassegrain : f/9

Platescale 15′′. 5/mm 11′′. 5/mm

The telescope is remotely operated from Centre for Research and Education in Science & Technology (CREST) Hosakote, via a dedicated satellite link. It is provided with a

Richey-Chretien optics with an altitude over azimuth mount. We have used the Himalayan Faint Object Spectrograph (HFOSC) available on HCT for the optical spectroscopic observations (refer to Table 2.3).

2.2.2 CCD detectors

Detectors are used at the focus of the telescope to collect light/photons from the astronomical objects. In this thesis, we have used CCDs as a detector with optical, NIR and a few X-ray telescopes (e.g., Chandra, XMM-Newton and Swift). These are considered to be one of the best detectors for astronomical purposes as they have linear response with photon energy, high quantum eﬃciency ( 80% at λ 6900 Å), low intrinsic noise ( a ∼ ∼ ∼ few electrons) and a high dynamic range (105), especially at optical and NIR wavebands. A typical CCD camera used with an optical/NIR telescope consists of a two dimensional array of photon detectors in a layer of semi-conducting material silicon (Si).

Each element of a CCD is called a pixel, typically a few microns in size. CCDs operate on the principle of photoelectric eﬀect i.e., the incident photons on a pixel are absorbed 42 2. Observations and Data Analysis

Figure 2.1: The 1.04-m ST at ARIES and 2.01-m HCT at IAO, Hanle.

Table 2.2: Characteristics of optical CCD Wright used with ST and SITe ST-002 used with HCT.

CCD Chip Size Pixel Gain Readout Dark Operating FOV size noise current temperature 1 1 1 (pixels) (µm) (e− /ADU) (e− /pix) (e− /pix/s) Wright 2048 2048 24 24 10 5.3 < 10 5 -120 C 13 13 × × − ◦ ′ × ′ SITe ST-002 2048 4096 15 15 1.22 4.8 < 10 4 -100 C 10 10 × × − ◦ ′ × ′ and liberate electrons from the silicon chip which move to the conduction band. Each individual pixel is capable of storing the produced electrons which can be read out from the CCD array to a computer to produce a digital image. Usually CCDs are exposed for a few seconds (integration time) to get suﬃcient electrons build up in pixels to produce a digital image of celestial objects and at the end of integration period, the accumulated charge is transferred to an on-chip ampliﬁer pixel-by-pixel (Howell, 2006). The information from the rows of pixels move down to a single parallel row (the serial register) which is read out sequentially by Analog-to-Digital (A/D) converter where it is measured and then recorded. The measuring device is emptied and once again the process is repeated.

This process continues until all the pixels have been measured (read out). 2.2 Optical observations 43

Table 2.3: Speciﬁcations of the HFOSC spectrograph at HCT, Hanle.

Parameter Values Wavelength range 350-900 nm Detector 2048 2048 CCD with 15 µm2 size pixels × Collimator focal length 252 mm Camera focal length 147 mm Spatial resolution 0.30 arcsecond/pixel Spectral resolution 0.16 nm to 5 nm for 1 slit ∼ ′′

Table 2.4: Details of the broadband ﬁlters used with ST, Nainital. λe f f is the eﬀective wavelength in the units of Angstrom (Å). The thickness of the glass in mm is indicated in the brackets.

Filter λe f f Band Width Glass combination of ﬁlter (Å) (Å) U 3650 680 UG1(2) + CuSO4 Solution(5) B 4400 980 GG385(2) + BG 18(1) + BG 12(1) + KG 3(2) V 5500 890 GG495(2) + BG 12(2) + KG 3(2) R 6500 2200 OG570(2) + KG 3(3) I 8000 2400 RG9(3) + WG 305(2)

CCDs are usually cooled down to temperatures where the thermal noise becomes negligible as compared to the signal received from the astronomical objects. CCDs used in our optical observations were therefore, cooled to about -120◦ C in a liquid nitrogen ∼ dewar to minimize the effect of thermal noise. The characteristics of the CCDs used for our optical observations are given in Table 2.2. The response curve of 2k 2k CCD on × ST is shown in Figure 2.2. For astronomical purposes, filters are used with CCDs to restrict the wavelength range of the incoming light. For optical photometric observations, we have used broadband filters Johnson UBV and Cousins RI with ST, Nainital. The details of these filters are given in Table 2.4. 44 2. Observations and Data Analysis

80.00

70.00

60.00

50.00

40.00

30.00

20.00

10.00

3000 4000 5000 6000 7000 8000 9000 10000

Figure 2.2: Response curve of Wright 2k 2k CCD used at ST, ARIES at -120◦C. The × curve indicates that the CCD peaks near 6700 Å.

2.2.3 The spectrograph

The Hanle Faint Object Spectrograph Camera (HFOSC) is an optical imager cum spectrograph mounted at the 2.01-m HCT at IAO, Hanle. The instrument allows a low and a medium resolution grism spectroscopy by inserting dispersive elements between the collimator and camera. The speciﬁcations of the HFOSC are listed in Table 2.3.

2.2.4 Photometry

We obtained the optical photometric follow-up observations of our sources using the ST, Nainital and optical spectroscopic observations using HCT, Hanle. Landolt standard stars (Landolt, 1992) were observed at different air-masses to determine the nightly extinction coefficient parameters and calibration purposes. The spectrum of an object star was calibrated with respect to the spectrum of a spectroscopic standard star taken on the same night. A number of bias frames were taken intermittently during the observing 2.2 Optical observations 45 runs. Dawn and/or dusk twilight flat frames were observed to derive pixel non-uniformity across the CCD chip. Image processing of CCD data The ground based telescopes suffer from atmospheric extinction and seeing variations from one night to another. The seeing conditions of a night can severely affect the resolution and image quality of the instrument. The quality of CCD images is also degraded by the telescope optics (geometrical distortion), CCD effects (dark current, hot pixel) and the electronics associated with the CCDs (thermal noise). Along with the sources of interest, other unwanted cosmic signals also get detected on the CCD. These are usually cosmic rays incident from all direction in the sky. All the effects described above can be removed from the raw CCD frames using IRAF1, MIDAS2 and DAOPHOT3 software. In this thesis, we have used IRAF software to clean the object frames and

DAOPHOT software to determine the magnitude of the target stars. For the present work, the main tasks used in both the software IRAF and DAOPHOT to analyze the optical observations are listed in Table 2.5. We have divided the CCD data reduction process into three steps : pre- processing, processing and post-processing.

Pre-processing The pre-processing of CCD images of object frames mainly include the bias- subtraction, flat-fielding and removal of cosmic ray events from the images. The raw frames i.e., the images of the sky as obtained by the CCD are processed using a set of bias frames (i.e., a zero second exposure) and a set of flat frames (i.e., a few seconds exposure frames) obtained using the same filter as that of the raw images.

1Image Reduction and Analysis Facility (IRAF) is distributed by the National Optical Astronomy Ob- servatories, Arizona which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. 2Munich Image and Data Analysis System (MIDAS) designed and developed by the European Southern Observatory ESO in Munich, Germany 3Dominion Astrophysical Observatory Photometry (DAOPHOT) 46 2. Observations and Data Analysis

The bias frames are zero second exposure frames, taken with the shutter of the

CCD closed, consists of only CCD readout. The bias frames thus contains the pixel to pixel information about the noise level due to the CCD electronics. To remove this noise offset from all the raw frames, a number of bias frames are median combined to obtain a master bias which can be subtracted from all the raw images. Generally a set of more than 3 - 4 bias frames are used for statistically significant sampling of the electronics noise and the variation. Even the resultant bias-subtracted frames do not have uniform response across the CCD because each pixel of the CCD acts as an independent detector and may have different response from rest of the pixels. To measure response of the individual pixels, the CCD is illuminated uniformly by exposing it to a patch of the sky which is not very bright but uniformly illuminated. Under such exposure, the images of the sky are obtained under different filters. They are called the flat-frames. The bias subtracted flat frames of a given filter are combined to get average flat frames for different filters. The bias subtracted object images in a filter are then divided using the corresponding average flat frames. The bias-subtracted and flat-fielded images of astronomical objects thus obtained may contain a few bright specks which are distinct from the stars in their intensity profile. Unlike the stars which have Gaussian-like profiles, the bright specks have sharply rising and falling profiles. These bright specks are cosmic ray events and are removed from the flat-fielded frame.

Processing

Processing of CCD images of object frames include listing the position of stars in the CCD frame according to the pixel number and measuring the intensity in terms of instrumental magnitude. We used DAOPHOT II software package to process the CCD frames obtained for our target stars.

Object Detection • 2.2 Optical observations 47

Unlike extended objects like galaxies, the stellar brightness distribution is supposed to have a Gaussian or a similar proﬁle which peaks at the center and falls oﬀ rapidly on either sides. This property is used by DAOPHOT to identify and list the stars in a given CCD frame in terms of their pixel based X and Y position on the CCD.

Photometry •

Photometry of stars implies assigning quantitative measure to the brightness of the detected stars. There are two popular procedures which are routinely followed for do- ing the photometry in DAOPHOT : a) Aperture photometry b) Proﬁle ﬁtting Photometry

1. Aperture Photometry

This technique is used to ﬁnd the brightness of a star by summing up all the photons in a circular geometric region around the star called aperture and subtracting the sky photons, estimated from a similar circular aperture on the same CCD frame in a star- free region. This technique is best used for isolated stars in the ﬁeld as there is less chance of contamination of a given aperture by nearby stars.

2. Profile-fitting photometry This technique assumes or estimates functional form of a point spread function (PSF) of a few observed bright stars and for the rest scales it according to the peak brightness of individual stars. This technique is mainly used for crowded fields and is based on the assumption that under similar observing conditions (atmospheric and technologi- cal) all the stars would have identical brightness profile and would differ from one another only by scaling the peak intensity. The PSF is approximated by a number of mathematical functions, the most common among them are :

r2 Gaussian G(r, a) exp − (2.1) ∝ 2a2 ! 1 Modified Lorenzian L(r, a, b) b (2.2) ∝ + r2 1 a2 48 2. Observations and Data Analysis

1 Moffat M(r, a, b) b (2.3) ∝ + r2 1 a2 Where r is the distance from the center and a & b are the parameters to be fitted (Stetson, 1987). These type of functional forms can be used to define the PSF for a star within an image, assuming that they provide a good representation of the data itself and the method is called analytical. Sometimes a combination of one or more of these functions is used to obtain a better fit. However, we see that for CCD frames having good image quality, the χ2 value in fitting Gaussian analytical function is better than that obtained from other two functions. This analytical method has the advantage of integrating numerically over the entire stellar image but the disadvantage being that the function used to fit the profile can only be approximated. The other PSF method is known as empirical where it is possible to simply store the observed profile of several bright stars as data array. The empirical method has the advantage that it uses the observed profile, but the drawback in this case is that near the central region of the stellar image, the brightness varies too rapidly between the adjacent pixels to be interpolated correctly. Better results can be obtained by combining these two methods.

Post-processing The magnitudes thus estimated are called the instrumental magnitudes. The instrumental magnitude of a star can significantly differ from one night to another depending on atmospheric conditions. Sometimes, the measurement of instrumental magnitude of a star differs by observing it with a different telescope. To be able to compare the results from different telescopes and observed under different atmospheric conditions, astronomers have come up with a set of standard stars. Any field observed with any telescope can be calibrated with respect to these standard stars and hence can be compared to each other. The steps involved in the post-processing are listed below: (i) Synthetic-aperture photometry is performed for several selected bright, isolated stars in each CCD frame using a series of concentric apertures of increasing sizes. 2.2 Optical observations 49

In profile fitting photometry, the instrumental magnitude of a star is identical using the height of a model PSF for that frame scaled to the intensity values recorded inside the star’s image. This magnitude is restricted to the chosen aperture and has to be corrected for the excess counts in the wings of the stellar profile. The correction from profile fitting magnitudes to aperture magnitudes are carried out by the process of determining aperture growth curve. This correction is known as aperture correction and is applied to the PSF magnitudes to obtain the aperture instrumental magnitudes. (ii) To determine the standard magnitudes and colours, we have to cross identify the stars of different frame and filters. It is usually done by considering 30 brightest stars ∼ between the two frames and providing the coordinate transformation equations between the various frames with respect to a reference frame. These transformation equations are used to cross-match all the stars in all the object frames with respect to the reference frame. If, after transformation to the coordinate system of the master frame, a star lies within the specified distance of a star in the master list, it is provisionally identified with that star; if it lies near no star in the master list it is added to the list as a possible new detection.

(iii) Now, one has to transform the instrumental magnitudes to the standard magnitudes. For this purpose, the transformation equations are as follows :

u = U + a + a (U B) + a X (2.4) CCD o 1 − 2 b = B + b + b (B V) + b X (2.5) CCD o 1 − 2 v = V + c + c (B V) + c X (2.6) CCD o 1 − 2 r = R + d + d (R I) + d X (2.7) CCD o 1 − 2 i = I + e + e (V I) + e X (2.8) CCD o 1 − 2 where a0, b0, c0, d0 and e0 are the zero points; a1, b1, c1, d1 and e1 are the color coeﬃ- cients; a2, b2, c2, d2 and e2 are the earth’s atmospheric coeﬃcients and X is the air mass. 50 2. Observations and Data Analysis

U, B, V, R and I are the standard magnitudes and uCCD, bCCD, vCCD, rCCD and iCCD are the corresponding aperture instrumental magnitudes. The second order colour correction terms are ignored as they are generally small in comparison to other errors present in the photometric data reduction.

We know that the light of celestial objects in optical wavebands is reduced because of absorption and diﬀusion by the medium when propagating through the Earth’s atmosphere. This reduction of incident ﬂux is called atmospheric extinction. It depends on the zenith distance (Z) of the object, wavelength, atmospheric condition during the observation and altitude of the observatory. Atmospheric extinction plays an integral part in photometric data reduction. For this, the air mass (X) is calculated using the following equation (Hardie and Ballard, 1962).

X = secZ 0.0018167(secZ 1) 0.002875(secZ 1)2 0.0008083(secZ 1)3 (2.9) − − − − − −

1 where secZ = (sin(φ)sin(δ) + cos(φ)cos(δ)cos(h))− , φ, δ and h being the observers latitude, the declination and the instantaneous hour angle of the star respectively. The air mass defines the thickness of the atmosphere crossed by the light rays. By definition, air mass is 1 at the zenith and increases with the zenith distance. The atmospheric extinction coefficients for each filter are determined from the plot of aperture magnitudes observed versus air mass for the comparison stars on each night using the method of least square fit. Using the transformation equations, we determine the values of transformation coefficients which along with the standard magnitudes from Landolt (1992) are used to generate secondary standards in the object frame. We determine a single consistent photometric zero point using all the frames from the local secondary stars. In this way, a list of stars and the corresponding positions, standard magnitudes and colours which are present in the CCD frame of interest can be obtained.

We have used IRAF tasks to perform pre-processing on optical data while the tasks in DAOPHOT II are used for processing and post-processing. The complete proce- 2.2 Optical observations 51

Figure 2.3: A chart summarizing the optical data reduction procedure dure to get the standard magnitude of a target star is given in the form of a chart in Figure 2.3. The IRAF and DAOPHOT II tasks used for the same are given in Table 2.5.

2.2.5 Spectroscopy

Optical spectra of stars are obtained using HFOSC of HCT, Hanle in a two-dimensional array. The long slit spectroscopic data consists of bias frames, flat frames, calibration frames and object frames. These frames are acquired for each observing night. The object frames contains the spectral image of the star focused at a slit. This image is made to disperse on the CCD along a particular direction known as dispersion axis, which is parallel to the slit width direction when the CCD axis is properly aligned with the slit axis. In order to avoid any degradation of resolution as well as the sky background contribution, the slit widths are usually kept equal to the full width at half maxima of the star’s seeing limited image. The pixels along the dispersion axis are measured in wavelength units, while the pixels along the spatial axis in photon counts. In order to convert these pixels and counts in wavelength and flux respectively, calibration frames are required. Arc frames and standard star frames are required for wavelength and flux 52 2. Observations and Data Analysis

Table 2.5: The IRAF and DAOPHOT II tasks used for optical data reduction.

Task Performs IRAF ZEROCOMBINE median combines a number of bias frames to obtain a master bias FLATCOMBINE combines a set of flat frames taken with a given filter to get a average flat frame COSMICRAYS used to remove the cosmic ray events from the flat-fielded frame DAOPHOT II FIND used to find and list stars in the CCD frame in terms of X and Y position on the CCD. This task performs a pixel by pixel search for brightness enhancements and then fits a Gaussian profile to such enhancements and hence lists all the stars present in the CCD frame PHOT measures the brightness or magnitude of all the stars present in the field by extracting the counts within a given aperture and subtracting the sky background close to the given star. This routine allows the user to specify a series of increasing concentric apertures and an annulus to evaluate the star free region. ALLSTAR measures the magnitude of a star by PSF photometry DAOGROW performs the aperture correction on given magnitudes using aperture growth curve (Stetson, 1990) DAOMATCH it considers brightest stars between the two frames (usually 30) and ∼ provides the coordinate transformation equations between the various frames with respect to a reference frame. DAOMASTER this programme accepts the approximate transformation equation provided by DAOMATCH, reads in all the stars listed for that field and cross-matches all the stars by spatial proximity. COLLECT & determines the values of transformation coefficients CCDSTD CCDAVE generates the secondary standards in the object frame with the help of standard magnitudes from Landolt (1992) FINAL it makes the final transformation for all the stars in the object frame 2.2 Optical observations 53

Table 2.6: Parameters of the New Technology Telescope (NTT)

Place La Silla Observatory, Chile

Longitude 70◦43′54 W

Latitude -29◦15′18 S Altitude 2375m System Ritchey-Chretien Diameter 3.58m,thinmeniscus(24 cm) Focal length Cassegrain: f/2.2

Platescale 5′′. 4/mm calibration respectively.

Extraction of optical spectrum Spectroscopic data reduction is done using the software package IRAF. The pre- processing steps like bias subtraction, flat fielding, dark correction and cosmic ray removal are similar to those discussed in the case of photometric data reduction. The one dimensional spectra are extracted using APALL task, which is based on the extraction algorithm by Horne (1986). This task eliminates the sky noise, delivers maximum possible signal-to-noise ratio and takes care of the effects of moderate geometric distortion.

The one dimensional spectrum is in the form of intensity versus pixel number. In order to calibrate the pixel number in terms of wavelength, it is necessary to take spectra of a laboratory standard source (e.g., Fe-Ar and Fe-Ne arc lamp). For such a standard source, wavelength of diﬀerent spectral features such as emission is known.

The wavelength calibration is done using IDENTIFY, HEDIT and DISPCOR tasks in IRAF. Usually a high order polynomial is employed to fit the identified pixels against wavelength. Flux calibration is performed using the tasks STANDARD, SENSFUNC (used to fit a polynomial to the observed magnitude as a function of wavelength after applying extinction correction in the standard and programme star spectra) and calibrated. 54 2. Observations and Data Analysis

Table 2.7: Characteristics of CCD used in the NIR instrument SOFI on the NTT. Array format Hawaii HgCdTe 1024 1024 × PixelSize 18.5 µm Quantum eﬃciency 65% Gain 5.4 e/ADU ∼ Read out noise 2.1 ADU ∼ Non-Linearity < 1.5% over 0 to 10000 ADU

Dark Current < 0.1 e−/s

2.3 Near-infrared observations

2.3.1 Telescopes

For the present work, we have used NIR data obtained from ESO - NTT archive and from 2MASS point source catalog (Cutri et al., 2003).

2.3.1.1 New Technology Telescope (NTT)

The ESO-NTT is an Alt-Azimuth, 3.58-m Richey-Chretien telescope located at La Silla Observatory, Chile which pioneered the use of active optics. This is claimed to be the ﬁrst telescope using full Active Optics and a great care is taken for a good ventilation of the telescope, and to avoid heat sources in and around the telescope. The basic parameters of NTT are listed in Table 2.6. In this thesis, we have used NIR imaging data from SOFI (Son of ISAAC; a large ﬁeld infra-red spectrograph and imaging camera) instrument on

NTT (refer Table 2.7). The largest FOV of SOFI is 4′.92, and it covers a wavelength range from 0.9 - 2.5 µm with J(1.25 µm, 0.29 µm), H(1.65 µm, 0.30 µm) and Ks(2.16 µm, 0.30 µm) ﬁlters, where the values in the bracket are the central wavelength and width of a given ﬁlter. 2.3 Near-infrared observations 55

2.3.1.2 Two Micron All Sky Survey (2MASS)

The digital Two Micron All Sky Survey (2MASS) was carried out with two identical highly-automated 1.3-m aperture, open-tube, equatorial fork-mount telescopes. One was the northern telescope located at 2306 m elevation in Arizona (Longitude: 110◦52′41′′ W,

Latitude: 31◦40′51′′ N) and the second was the southern telescope at 2171 m elevation in

Chile (Longitude: 70◦48′18′′ W, Latitude: 30◦10′04′′ S). Both telescopes have Cassegrain focus mount for the infrared cameras. The detailed description of the 2MASS camera optical design is given in Milligan et al. (1996). In the 2MASS survey three band-passes are used namely J(1.25 µm, 0.25 µm), H(1.65 µm, 0.30 µm) and K(2.17 µm, 0.31 µm).

Each pixel on the CCD camera mounted on the 2MASS telescopes subtends 2′′. 0 on the sky. The gain of the 2MASS electronics is 8 electrons per ADU count and read out ∼ noise is 40 electrons. The 2MASS data base provides photometry in the NIR J, H and K wavebands to a limiting magnitude of 15.8, 15.1 and 14.3 respectively, with S/N greater than 10.

2.3.2 Photometry

Unlike the optical observations, a typical exposure time of a desired image taken with a ground based NIR CCD photometer contains background contributions from telescope and atmospheric radiation. Thus, a proper background subtraction is required in NIR observations before applying cleaning procedure to the raw data. The sky background in

NIR also varies continuously, because of changes in temperature or OH airglow emission, which varies on the timescale of a few minutes. Because of this reason, the exposures in NIR wavebands are usually short. In NIR observations, a number a dithered exposures are made followed by a median averaging. Usually, more than one set of dithered images is obtained to increase S/N ratio of the object. Along with the object frames, a number of bias and ﬂat frames are obtained to clean the images from CCD eﬀects. We used the standard routines in IRAF to reduce the NIR observations. 56 2. Observations and Data Analysis

Image processing of CCD data

It is usually impossibleto obtain a sky frame withoutstars to subtract background from the object frame. Thus, an empty sky frame is generated by median averaging a number of object frames taken with random shifts, but with the same exposure time and filter, etc. The stellar images are eliminated by this process if the fields are reasonably sparsely populated. The empty sky frame generated this way, can be used as a sky frame and can be subtracted from individual object frames. This process is called sky subtraction. Once the NIR object frames are corrected for sky background, the images can be cleaned using flat frames and bias frames as described for optical observations in Section 2.2.4. The magnitude of stars can be obtained using the standard routines in IRAF as described for optical observations (Section 2.2.4).

2.4 X-ray observations

The Earth’s atmosphere does not allow X-rays from stellar sources to penetrate through it. Thus X-ray observatories are flown above the earth’s atmosphere with different instrument characteristics and scientific objectives to observe various X-ray sources. For the present work, we have used the following X-ray telescopes to obtain observations of our sources.

2.4.1 Telescopes

2.4.1.1 Chandra

NASA’s Chandra X-ray observatory (Weisskopf et al., 2002), which was launched and deployed by Space Shuttle Columbia on July 23, 1999, is the most sophisticated X-ray observatory built to date. The Wolter-type high precession mirror assembly used in Chan- dra provides measurement of position of the X-ray sources with an accuracy of 0′′. 5, which is the highest position accuracy provided by any X-ray observatory so far. This is also the ﬁrst X-ray observatory to provide the X-ray spectral resolution (E/∆E) of 1000 to detect ∼ the various X-ray spectral lines in the spectrum of celestial sources. It spends 85% of its 2.4 X-ray observations 57 orbit above the belts of charged particles that surround the Earth, thus can be used for uninterrupted observations as long as 55 hours.

Science Instruments The Chandra X-Ray Observatory combines the mirrors with four instruments to capture and probe the X-rays from astronomical sources : Advanced CCD Imaging Spec- trometer (ACIS), High Resolution Camera (HRC) and the high resolution spectrometers - HETGS and LETGS. For the present work, we have used only ACIS onboard Chandra and is described below :

Advanced CCD Imaging Spectrometer (ACIS) : The Chandra Advanced CCD Imaging

Spectrometer (ACIS) is an array of CCDs and can be used both in imaging and spectral mode. The incoming X-rays in the Chandra observatory are focused by the mirrors to a tiny spot on the focal plane and the ACIS onboard Chandra is well matched to capture the sharp images formed by these mirrors and provides information about the incoming

X-rays such as their number, position, energy and time of arrival. ACIS when operated in the imaging mode provides the position of the X-ray sources with a sub-arc second accuracy ( 0′′. 5). ∼

2.4.1.2 XMM-Newton

The European Space Agency’s (ESA) X-ray Multi-Mirror Mission (XMM-Newton) was launched on December 10, 1999. It carried three high throughput X-ray telescopes with an unprecedented effective area, and an optical monitor, the first flown on an X-ray observatory to observe X-ray sources simultaneously in the X-ray and the optical wavebands. Each of the three X-ray telescopes on board XMM-Newton (Jansen et al., 2001) consists of 58 Wolter I grazing-incidence mirrors which are nested in a coaxial and cofocal configuration.

Science instruments 58 2. Observations and Data Analysis

The instruments onboard XMM-Newton observatory are : The European Photon

Imaging Camera (EPIC), The Reﬂection Grating Spectrometer (RGS) and Optical Mon- itor (OM). For the present work, we have used only EPIC camera (Strüder et al. 2001, Turner et al. 2001) onboard XMM-Newton satellite, the details of which are given below :

The European Photon Imaging Camera (EPIC) : It is a set of three X-ray CCD cameras onboard XMM-Newton observatory mainly used for X-ray imaging over the telescope’s

ﬁeld of view (FOV) of 30′, moderate resolution spectroscopy (E/∆E 20-50), and X-ray ∼ photometry. It operates in the energy range 0.15 to 15 keV and registers the energy and arrival time of each photon falling on it. Two of the three EPIC cameras are Metal Oxide

Semi-conductor (MOS) CCD arrays (referred to as the MOS cameras) and are installed behind the X-ray telescopes that are equipped with the gratings of the Reflection Grating Spectrometers (RGS). The gratings divert about half of the telescope incident flux towards the RGS detectors such that (taking structural obscuration into account) about 44% of the original incoming flux reaches the MOS cameras. The third X-ray telescope has an unobstructed beam; the EPIC instrument at the focus of this telescope uses pn CCDs and is referred to as the pn camera.

2.4.1.3 RXTE

The Rossi X-ray Timing Explorer (RXTE), named after astronomer Bruno Rossi, was launched on December 30, 1995 into low-earth orbit (about 600 km and 23◦ inclination). The RXTE is highly maneuverable (6 deg/minute), so that it can be made to point to a chosen source rapidly. This ﬂexibility allows the instrument to respond to short-lived or new phenomena as they are discovered. RXTE probes the physics of cosmic X-ray sources by making sensitive measurements of their variability over time scales ranging from milliseconds to years.

Science instruments 2.4 X-ray observations 59

Figure 2.4: The Rossi X-ray Timing Explorer (RXTE)

The three instruments onboard RXTE are : Proportional Counter Array (PCA), High Energy X-ray Timing Experiment (HEXTE) and the All Sky Monitor (ASM). An artist’s impression of RXTE is shown in Figure 2.4 along with its onboard instruments. For the present work, we have used data obtained with PCA and ASM of RXTE. The details of these two instruments are given below :

Proportional Counter Array (PCA) It is an array of ﬁve units of proportional counter detectors with a large collecting area of

6250 cm2 (Jahoda et al., 1996). PCA contains xenon gas and is sensitive to X-rays with energies from 2 - 60 keV. This instrument can provide the observations of celestial X-ray sources with a maximum time resolution of 1 microsecond and spectral resolution of < 18% at 6 keV.

The All Sky Monitor (ASM) The ASM of RXTE (Levine et al., 1996) consists of three wide-angle shadow cameras equipped with proportional counters with a total collecting area of 90 cm2. It operates in the energy range 2 - 12 keV and scans 80% of the skyeach orbit(in every90 min) witha ∼ spatial resolution of 3′ 15′. ASM plays an important role in identifying state transitions × 60 2. Observations and Data Analysis and outbursts from transient sources, allowing us to trigger follow-up observations with other instruments within a few hours. The instrument also permits us to monitor the long- term intensity and behavior of the brightest X-ray sources.

2.4.1.4 Swift

Swift is a part of NASA’s medium explorer (MIDEX) program and was launched into a low-Earth orbit on November 20, 2004. Swift is a multiwavelength observatory and makes observations of celestial sources in the X-ray and the optical/ultra-violet wavebands simultaneously.

Science Instruments Three instruments onboard Swift are Burst Alert Telescope (BAT), X-ray Tele- scope (XRT) and UV/Optical Telescope (UVOT). However, we have used only XRT for the present work.

X-ray Telescope (XRT) : The XRT (Burrows et al., 2005) is a sensitive, ﬂexible, autonomous X-ray CCD imaging spectrometer onboard Swift, used to measure the position, spectrum, and brightness of transient X-ray sources over a wide dynamic range covering more than 7 orders of magnitude in ﬂux. It works in the energy range 0.3 - 10 keV and provides the position of the X-ray sources with 5′′ accuracy. ∼

2.4.1.5 Other X-ray observatories

BeppoSAX BeppoSAX was launched on April 30, 1996 and operated for 6 years (April 30, 1996 - April 30, 2002). It was the first X-ray mission with scientific payloads 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. BeppoSAX contained five science instruments : 2.4 X-ray observations 61

Low Energy Concentrator Spectrometer (LECS) •

Medium Energy Concentrator Spectrometer (MECS) •

High Pressure Gas Scintillation Proportional Counter (HPGSPC) •

Phoswich Detector System (PDS) •

Wide Field Camera (WFC) •

LECS and MECS had imaging capability, whereas the high-energy narrow ﬁeld instruments were non-imaging. For the present work, we have used data obtained using LECS and MECS. The MECS of BeppoSAX contained three identical gas scintillation proportional counters operating in the 1.3 to 10 keV range. LECS was almost identical to the MECS units, except that it had a thinner window that allows photons with lower energies down to 0.1 keV to pass through.

ASCA

ASCA satellite was launched on February 20, 1993. ASCA operated successfully till July 15, 2000 and re-entered on March 2, 2001 after 7 and half years of scientific observations. ASCA was the first satellite to use CCD detectors for X-ray astronomy and worked in energy band 0.4 - 10 keV. It was the first X-ray mission to combine imaging capability with broad pass band, good spectral resolution, and a large effective area. ASCA carried four large-area X-ray telescopes. At the focus of two of the telescopes were Gas Imaging Spectrometers (GIS), while Solid-state Imaging Spectrometers (SIS) were at the foci of the other two telescopes. The X-ray Telescopes (XRTs) utilized multiple-nested, thin-foil, conical mirrors and could provide a spatial resolution of 3′. ∼ The two SIS detectors (0.4 - 10.0 keV) were identical CCD cameras with a FOV of 22′ × 22′, and were the first X-ray CCD spectrometers in orbit. The GIS (0.8 - 10.0 keV) was a gas imaging scintillation proportional counters with a circular FOV of radius 25′, having area 4 times as compared to SIS. 62 2. Observations and Data Analysis

EXOSAT

The EXOSAT satellite was operational from May 1983 until April 1986 and in that time made observations of a wide variety of objects in the X-ray wavebands. The payload consisted of three instruments that produced spectra, images and light curves in various energy bands. Two Low-Energy Imaging Telescopes (LEITs), each with a channel-multiplier array at their focus, were sensitive over the energy range 0.05 - 2 keV, and third payload the Medium-Energy (ME) experiment consisted of an array of proportional counters with a total area of 1600 cm2 that gave moderate spectral resolution in the

1 - 50 keV band.

2.4.2 Extracting data products

The observations made by diﬀerent X-ray observatories are archived at their science centers and can be retrieved from there. Archived data can also be retrieved from NASA’s High energy Astrophysics Science Archive Research Center4 (HEASARC). The observations taken with each X-ray observatory has to be processed using its data reduction tools and procedure to bring it to a standard format which can be further reduced using X-ray astronomical data analysis package FTOOLS. In this section, we summarize the procedures adopted for various X-ray observatories to extract standard products from their raw data.

Chandra The observations from pointed Chandra observations are archived at the Chan- dra Data Archive5 (CDA) which is a part of the Chandra X-ray Observatory Science

Center (CXC) operated for NASA by the Smithsonian Astrophysical Observatory. Event ﬁles of observations obtained using Chandra can be processed using the software CIAO 3.46 provided by the Chandra X-ray center.

4http://heasarc.gsfc.nasa.gov/ 5http://cxc.harvard.edu/cda/ 6http://cxc.harvard.edu/ciao/ 2.4 X-ray observations 63

XMM-Newton

The XMM-Newton data for both pointed and archived observations can be retrieved from the XMM-Newton Science Archive7 (XSA). The full set of XMM-Newton observation data ﬁles, corresponding to a given XMM-Newton observation, is basically composed of :

the observation data files (ODF) which include the raw event science files from • EPIC, RGS and OM instruments, the instruments house-keeping files, the radiation monitor files and the spacecraft files.

All the data products generated by the Pipeline Processing Subsystem (PPS) at the • XMM-Newton Survey Science Center (SSC).

The observations taken with XMM-Newton X-ray observatory can be processed using the software package Scientiﬁc Analysis Software (SAS8). The tasks EMPROC and

EPPROC in SAS generate the event ﬁles of a particular observations for MOS and pn instruments respectively. These event ﬁles can be further processed using the task XMM- SELECT in SAS to extract images, lightcurves and spectrum of a source.

RXTE

For the present work, we used RXTE archived data from HEASARC RXTE interface. For the present work, we have used binned modes ‘standard 1’ and ‘standard 2’ of RXTE observations, having time resolution of 0.125 s and 16 s respectively. The lightcurve and the spectrum of a source can be extracted from these modes using the task

SAEXTRCT in FTOOLS. A model dependent background spectrum of a source can be generated using a task PCABACKEST in FTOOLS.

Swift

7http://xmm.esac.esa.int/xsa/ 8http://xmm.esac.esa.int/sas/ 64 2. Observations and Data Analysis

Swift observations are archived at University of Leicester data baseandat theASI

Science Data Center (ASDC) in Italy. Data can be obtained from any of the centers or from HEASARC Swift interface. We have used the archived Swift XRT data for the present study. For reduction of the XRT data, we used XRTPIPELINE software in HEASARC.

BeppoSAX, ASCA and EXOSAT

Data observed by past X-ray observatories (BeppoSAX, ASCA and EXOSAT) can also be retrieved from HEASARC archive and their event ﬁles can be processed using task XSELECT in FTOOLS to produce lightcurves and X-ray spectrum.

2.4.3 Timing Analysis

1 The lightcurve (time versus counts or counts s− ) of a source can be extracted from a circular region of a few arc second to a few arc minute radius centered on the source of interest in an X-ray image using the task XSELECT in FTOOLS. Similarly, the background lightcurve can be extracted from the source free region on the same CCD. We used a task

LCMATH to subtract the background lightcurve from the observed source lightcurve (ac- tually a sum of the source and the background lightcurve). The task LCMATH allows to weight the individual light curves according to the area of the corresponding region used before subtracting. The light curves thus obtained contains only source counts and can be plotted as well as re-binned using the task LCURVE. The lightcurve of the source thus obtained can be checked for any periodic (pulse period, orbital period etc.) and aperiodic (X-ray outbursts, X-ray bursts etc.) variabilities present in the lightcurve. Sometimes, it is not easy to pick up a periodic signal from the lightcurve by visual inspection, thus a technique called ‘EFSEARCH’ in FTOOLS is used for the same. EFSEARCH searches for periodicities in a lightcurve by folding data at diﬀerent trial periods over a period range. At the correct trial period, the actual pulse proﬁle is produced which deviates maximally from a constant and results in a high value of chi square while at the incorrect trial period values, the pulses get smeared out resulting into a constant and 2.4 X-ray observations 65 a low value of chi square. EFSEARCH is best suited for the periodic signals, however for the detection of quasi periodic signals, this method is not suitable. We used ‘power density spectrum (PDS)’ to search for quasi periodic signals in the lightcurve of our X-ray binaries. The power density spectrum can be computed with the help of a Fast Fourier Transform (FFT) algorithm and as an output gives the power corresponding to each frequency present in the lightcurve. The power density spectrum,

P f of a lightcurve Ct is given by

(C f C∗f + C f C∗ f ) P = − − (2.10) f 2 where C f is the Fourier transform of the lightcurve Ct. Since Ct is real, the above equation can be written as

P f = C f C∗f (2.11)

Along with the sharp peaks corresponding to the periodic signals and their harmonics, PDS also gives a small and broad peak for quasi periodic signals, if present in the lightcurve. The task ‘POWSPEC’ in FTOOLS can be used to produce PDS for a given lightcurve. This task also allows the data to be re-bin and divide into intervals. The PDS from several intervals can be averaged in a frame, and the results can also be re-binned.

2.4.4 Spectroscopy

The X-ray spectrum obtained by spectrometers onboard X-ray observatories record photon counts (C) within speciﬁc instrument channels, (I). The observed spectrum is related to the actual spectrum of the source, f (E)by:

∞ C(I) = f (E)R(I, E)dE (2.12) Z0 where R(I, E) is the instrumental response and is proportional to the probability that an incoming photon of energy E will be detected in channel I. It is not possible to determine the actual spectrum of the source, f (E), by inverting the above equation but such inver- sions tends to be non-unique and unstable to small changes in C(I). The usual alternative 66 2. Observations and Data Analysis is to choose a model spectrum, f (E), that can be described in terms of a few parameters and fit to the data obtained by the spectrometer. For each f (E), a predicted count spectrum Cp(I) is calculated and compared to the observed data, C(I). Then a fit ‘statistic’ is computed from the comparison and used to judge whether the model spectrum fits the data obtained by the spectrometer. The model parameters then are varied to find the parameter values that give the most desirable fit statistic. These values are referred to as the best-fit parameters. The model spectrum, fb(E), made up of the best-fit parameters is considered to be the best-fit model. The most common fit statistic in use for determining the ‘best-fit’ model is χ2, defined as follows: (C(I) C (I))2 χ2 = − p 2 (2.13) X (σ(I)) where σ(I) is the (generally unknown) error for channel I (e.g., if C(I) are counts then σ(I) is usually estimated by √C(I) ; Wheaton et al. 1995). As a general rule, one wants the ‘reduced χ2’ (= χ2/ν) to be approximately equal to one (χ2 ν). For a given best-fit ∼ parameters, the range of values within which one can be confident that the true value of the parameter lies is called confidence interval. The confidence interval for a given parameter is computed by varying the parameter value until the χ2 increases by a particular amount above the minimum, or best-fit value. The amount that the χ2 is allowed to increase (also referred to as the critical ∆χ2) depends on the confidence level one requires, and on the number of parameters whose confidence space is being calculated. The critical χ2 for common cases are given in Table 2.8. We have used X-ray spectral analysis package ‘XSPEC9’ for spectral analysis (Arnaud, 1996). This software performs the spectral fitting as follows: C(I): The Observed Spectrum

To obtain each observed spectrum, XSPEC uses two files: the data (spectrum) file and the background file. It first reads the total photon counts detected by the instrument in

9http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/ 2.4 X-ray observations 67

Table 2.8: Conﬁdence interval for the parameters 1, 2 and 3

Conﬁdence Parameters 1 2 3 0.68 1.00 2.30 3.50 0.90 2.71 4.61 6.25 0.99 6.63 9.21 11.30

a given channel for both the data and background ﬁles. After that it uses the background ﬁle to derive the set of background subtracted spectra C(I) in units of counts per second. The background-subtracted count rate is given by, for each spectrum:

D(I) b B(I) C(I) = D(I) (2.14) aD(I)tD − bB(I) × aB(I)tB where D(I) and B(I) are the counts in the data and background files: tD and tB are the exposure times in the data and the background files; bD(I) and bB(I), aD(I) and aB(I) are the background and area scaling values from the spectrum and background respectively, which together refer to the background flux of the same area as the observations as necessary. After this step, we get a observed spectrum of the object.

R(I, E): The Instrumental Response

The actual spectrum of a source taken with an instrument is a convolution of the source spectra and the instrument response (described above). As such, the response is a continuous function of E. This continuous function is converted to a discrete function by the creator of a response matrix that deﬁnes the energy ranges, EJ such that :

E J R(I, E)dE EJ 1 RD(I, J) = R − (2.15) EJ EJ 1 − −

XSPEC reads both the energy ranges, EJ, and the response matrix RD(I, J) from a response ﬁle in a compressed format that only stores non-zero elements. XSPEC also includes an option to use an Auxilliary Response File (ARF), which contains an array 68 2. Observations and Data Analysis

AD(J) that is multiplied into RD(I, J) as follows :

R (I, J) R (I, J) A (J) (2.16) D → D × D This array is designed to represent efficiency of the detector with the response file representing a normalized Redistribution Matrix Function, or RMF. Conventionally, the response is in units of cm2. To improve the statistics, the final X-ray spectrum obtained, can be binned such that at least 20 counts are collected in each energy bin using a task GRPPHA in XSPEC.

For the present work, we have mostly used power-law, black-body or bremsstrahlung models as described below :

Powerlaw model : Powerlaw model is a simple photon power law deﬁned as :

α A(E) = KE− (2.17) where α is a photon index of power law (dimensionless) and the normalization, K is

1 2 1 photons keV− cm− s− at 1 keV.

Blackbody model : A blackbody spectrum is deﬁned as : K 8.0525E2dE A(E) = × (2.18) (kT)4[exp(E/kT) 1] − 2 where kT is the temperature in keV and the normalization, K = L39/D10, where L39 is the

39 1 source luminosity in units of 10 erg s− ,D10 is the distance of the source in units of 10 kpc.

Thermal bremsstrahlung model : A thermal bremsstrahlung spectrum in XSPEC based on the Kellogg et al. (1975) polynomial ﬁts to the Karzas and Latter (1961) numerical values. A routine from Kurucz is used in at low temperature end. In this model, the He abundance is assumed to be 8.5 % of H by number and the free parameters are : the

15 = 3.02 10− plasma temperature in keV and the normalization 4π×D2 nenIdV, where D is the R 3 distance to the source (cm) and ne, nI are the electron and ion densities (cm− ) of the source. Part I

True nature of X-ray binaries

Chapter 3

The nature of seven faint X-ray pulsators

3.1 Introduction

Majority of the known high-mass X-ray binaries are Be/X-ray binaries (Liu et al. 2005; Liu et al. 2006). In Be/X-ray binaries, the compact object is almost always a neutron star and is typically in a wide and eccentric orbit (e 0.3 - 0.5; Bildsten et al. 1997), and ∼ spends little time in close proximity to the dense circumstellar disk surrounding the Be companion (Coe 2000; Negueruela 2004). No black-hole and Be star system has been found yet (Zhang et al., 2004). Most of these systems are transient in nature (mostly in X-ray waveband), and hence are called transient Be/X-ray binaries. The X-ray outbursts in them are associated with the passage of neutron star through the Be-star disk (Type-I outburst) and the sudden mass ejections from the Be star (Type II outburst; refer Chapter 1).

During the last 10 years, a new class of Be/X-ray binaries have been discovered and they are called "persistent Be/X-ray binaries". The persistent nature of these binaries is expected to be due to the circular orbit (e 0) of the neutron star around the Be star ≈ Research presented in this Chapter is partly published in Kaur et al. (2009)

71 72 3.ThenatureofsevenfaintX-raypulsators which has been conﬁrmed till now only in one member of this class, X-Persei (e = 0.12;

Levine et al. 2000). Most of these systems have low X-ray luminosities (L 1036 erg X ∼ 1 s− ), long spin periods ( 150 s) and long orbital periods ( 30 d). The long orbital periods ≥ ≥ of these systems suggest that the tidal circularization could not yet have occurred after the supernova explosion. Therefore, the low eccentricities of these systems must be primor- dial and the supernova explosion could not have been accompanied by a signiﬁcant kick to the neutron star (Pfahl et al., 2002). The exact mechanisms behind the low eccentricities are not yet understood. A possible explanation is given by Podsiadlowski et al., (2004) in which it is suggested that a core-collapse in an electron-capture supernovae leads to a prompt explosion rather than a very slow and delayed neutrino-driven explosion and it can lead to the formation of neutron stars with a low-velocity kicks. If true, then these systems form a very promising new window onto the formation mechanisms of the high mass X-ray binaries. Till now, persistent nature of only four Be/X-ray binaries have been conﬁrmed in our Galaxy (4U 0352+309, RX J0146.9+6121, RX J0440.9+4431 and RX J1037.5-564; Haberl et al. 1998; Reig and Roche 1999) and a number of X-ray binaries are expected to be the members of this class. However, in most of these expected Be/X-ray binaries, the nature of the companion star is not known, mainly due to the following reasons :

The large error circle on their position measurement provided by the X-ray obser- • vatories with which they were discovered. Most of these sources are in the Galactic

plane towards the Galactic center. Thus a large number of optical/infrared sources fall in a few arc-second error circle on their position, making diﬃcult to ﬁnd the exact optical/infrared counterpart.

The lack of additional observations in all wavelengths after their discovery reports. •

In this Chapter, we have investigated seven expected persistent Be/X-ray pulsars to conﬁrm the Be nature of their companion stars as follows : 3.2 General properties of persistent Be/X-ray pulsars 73

1. We determined the position of these X-ray objects in two steps: (a). X-ray pulsa-

tions and spectral measurements are made from their XMM-Newton observations to conﬁrm the identity of these systems and to determine the position with an accuracy of a few arcseconds (b). Chandra observations are analysed for position

determination with 0′′. 6 accuracy.

2. With an accurate knowledge of the position of the X-ray stars we have used new/archival NIR observations to identify their counterparts. In most of the se-

lected sources, the New Technology Telescope (NTT) imaging observations are used to identify their near infrared (NIR) counterparts.

3. Finally, an estimate of the nature of the companion star is made using the multiband

photometry in NIR and independently using a distance measurement method.

4. The X-ray timing and spectral measurement from the XMM-Newton observations further helped us to constrain the nature of these systems.

This Chapter is organized as follows : In Section 3.2, we have discussed the general properties of persistent Be/X-ray binaries. A brief introduction on the selected sample of sources is given in Section 3.3. The Chandra and XMM-Newton observations of pulsators are described in Section 3.4. The X-ray timing and spectral analysis of these pulsators are described in Section 3.5 and Section 3.6 respectively and the NIR observations are described in Section 3.7. We have compared the observational properties of HMXBs, LMXBs and Intermediate Polars (IPs) in Section 3.8 before the discussion of individual sources in detail (Section 3.9). The conclusions of this Chapter are given in Section 3.10.

3.2 General properties of persistent Be/X-ray pulsars

A classiﬁcation of persistent Be/X-ray pulsars has been suggested by Reig and Roche (1999) on the basis of the common properties seen among the known persistent Be/X-ray 74 3.ThenatureofsevenfaintX-raypulsators pulsars and is as follows :

1. Persistent Be/X-ray systems are slow pulsators (P > 200s) while most transients Be/X-ray systems spin faster.

2. The X-ray luminosity of the persistent Be/X-ray sources is always below 1035 erg

1 s− with generally small X-ray variability compared to the strong outbursts of the transient Be/X-ray systems.

3. The X-ray spectra of the persistent Be/X-ray sources diﬀer from the transient ones: lower cut-oﬀ energy ( 4 keV versus 10 - 20 keV) and no orweak ironline emission ∼ compared to their transient counterparts.

3.3 Our sample of X-ray pulsators

We selected a sample of seven X-ray pulsators (Table 3.1) from literature which are expected to be persistent X-ray binary pulsars with Be companions. This prediction is based on the X-ray observations (spectral and timing) during which they were discovered. We selected these X-ray pulsators from literature with pulse period, P 150 s. Anomalous s ≥ X-ray pulsars1, Soft Gamma Ray repeaters2 and low mass X-ray binary pulsars are most likely to have pulse period of a few seconds to few tens of seconds (except very few

LMXBs like GX 1+4, Ps = 140 s). Thus our sample of pulsators are unlikely to be the member of any of these classes. Among the seven selected X-ray pulsators (Table 3.1), SAX J1324.4–6200 was discovered serendipitously in the BeppoSAX observations of a LMXB XB 1323–619 while rest all six X-ray pulsators (SAX J1452.8–5949, AX J1700.1–4157, AX J1740.1–

1The Anomalous X-ray Pulsars (AXPs) are a small class of pulsars, widely believed to be magnetars (young, isolated, highly magnetized neutron stars) with spin periods in the 6 - 12 s range 2Soft Gamma Ray repeaters are also interpreted as magnetars, showing recurrent short ( 0.1 s) bursts ∼ of high-energy radiation in the tens to hundred keV energy range. Their X-ray lightcurves usually show ∼ 10 1 periodicities of the order of 10 s and period derivatives of the order of 10− s s− . 3.4 X-rayobservationsanddatareduction 75

Table 3.1: Summary of basic parameters of our sample of X-ray pulsators

Star l b Spin period Absorbed ﬂux Spectral Refer- 1 2 (in seconds) (erg s− cm− ) parameters ences

(2 - 10 keV) NH Γ SAX J1324.4–6200 306.79 0.61 170.84 0.04 0.5 - 0.8 7 1.0 1 ± SAX J1452.8–5949 317.65 -0.46 437.4 1.4 0.08 2 1.4 2 ± AX J1700.1–4157 344.04 0.24 714 0.3 0.3 - 0.6 6 0.7 3 ± AX J1740.1–2847 359.49 1.09 729 14 0.4 3 0.7 4 ± AX J1749.2–2725 1.70 0.11 220.38 0.20 0.2 - 3.0 9 0.9 5 ± AX J1820.5–1434 16.47 0.07 152.26 0.01 1 - 4 10 0.9 6 ± AX J1832.3–0840 23.04 0.26 1549.1 0.4 0.9 1 0.9 7 ± REFERENCES. – (1) Angelini et al. 1998 (2) Oosterbroek et al. 1999 (3) Torii et al. 1999 (4) Sakano et al. 2000 (5) Torii et al. 1998 (6) Kinugasa et al. 1998 (7) Sugizaki et al. 2000

2847, AX J1749.2–2725, AX J1820.5–1434 and AX J1832.3–0840) were discovered during the ‘Galactic plane/Galactic center’ survey project made using the BeppoSAX and the

ASCA X-ray observatories to search for faint X-ray sources in the Galactic plane (Sakano et al., 2000). Table 3.1 summarizes the spectral parameters of these pulsators along with the pulse period with which they were discovered.

3.4 X-ray observations and data reduction

Our X-ray observations of this sample of seven slow pulsators were carried out using the European Photon Imaging Camera (EPIC) aboard the XMM-Newton satellite and advanced CCD Imaging Spectrometer (ACIS) aboard the Chandra satellite.

3.4.1 XMM-Newton

The X-ray pulsators (Table 3.1) were observed for 7 -32 ks each using the XMM-Newton satellite. The observation details are summarized in Table 3.2. During our observations, both the EPIC-MOS and EPIC-pn cameras (Turner et al. 2001; Strüder et al. 2001) were operated in the Full Frame mode and with the medium ﬁlter. The ﬁeld of view (FOV) of 76 3.ThenatureofsevenfaintX-raypulsators

Table 3.2: Log of the XMM-Newton EPIC X-ray Observations of X-ray pulsators

Object Date ObsId TotalObservation Span (ks) SAXJ1324.4–6200 11Jan2008 0511010201 19.3 SAXJ1452.8–5949 07Feb2008 0511010501 6.9 AXJ1700.1–4157 17Feb2008 0511010601 7.9 AXJ1740.1–2847 27Feb2008 0511010701 9.3 AXJ1749.2–2725 04Mar2008 0511010301 8.9 AXJ1820.5–1434 30Sep2007 0511010101 11.2 AXJ1832.3–0840 16Oct2007 0511010801 31.3

XMM-Newton-EPIC camera is 30′. The EPIC observations data were processed using the XMM-Newton Science Analysis System (SAS version 7.1.0)1. Investigation of the full-field count-rate of X-ray pulsators revealed no high- count rate background particle flaring in all of them except in SAX J1324.4–6200, in which high and variable background count rate was detected for 3 ks and is removed ∼ for further analysis. We used the task edetect_chain to find the exact position of X-ray sources in the combined images of EPIC-MOS and EPIC-pn instruments. The positions of the X-ray sources determined from their XMM-Newton observations around the previous detections of the X-ray pulsators are listed in Table 3.3 along with the error circles. The error circle on the position of X-ray source is adopted as a quadratic sum of the bore sight error of the XMM-Newton telescope2 and the statistical error given by the task edetect_chain. We have also listed the count rate of each X-ray pulsator detected in EPIC-

MOS (MOS1 + MOS2) and EPIC-pn instruments in Table 3.3.

3.4.2 Chandra

X-ray pulsators were observed for 1 ks each using the Chandra satellite. We summa- ∼ rized the details of the Chandra observations of the X-ray pulsators in Table 3.4. The

1See http://xmm2.esac.esa.int/sas/ 2See http://xmm2.esac.esa.int/docs/documents/CAL-TN-0018.pdf 3.4 X-rayobservationsanddatareduction 77

Table 3.3: Position of X-ray pulsators as determined from their XMM-Newton observations. The count rates of them in EPIC-MOS1, MOS2 and pn instruments are also listed.

Star Count rate R.A. DEC. error circle

MOS1 MOS2 pn hh:mm:ss ◦ ′ ′′ ′′ SAXJ1324.4–6200 0.093 0.092 0.30 13:24:26.64 62:01:18.48 2.0 − SAXJ1452.8–5949 0.018 0.014 0.055 14:52:52.80 59:49:08.04 2.0 − AXJ1700.1–4157 0.13 0.12 0.39 17:00:04.32 41:58:04.44 2.0 − AXJ1740.1–2847 0.10 0.10 0.30 17:40:09.12 28:47:26.16 2.0 − AXJ1749.2–2725 0.033 0.015 0.078 17:49:12.24 27:25:37.56 2.0 − AXJ1820.5–1434 0.044 0.044 0.13 18:20:30.00 14:34:23.16 2.0 − AXJ1832.3–0840 0.26 0.24 0.79 18:32:19.44 08:40:30.47 2.0 −

Table 3.4: Log of the Chandra ACIS-I X-ray Observations

Object Date ObsId TotalObservation Span (ks) SAXJ1324.4–6200 28Nov2007 9012 1.13 SAXJ1452.8–5949 30Dec2007 9014 0.95 AXJ1700.1–4157 30Jun2008 9015 1.13 AXJ1740.1–2847 06Sep2008 9016 1.17 AXJ1749.2–2725 27Apr2008 9013 1.18 AXJ1820.5–1434 03Sep2008 9011 1.15 AXJ1832.3–0840 07Aug2008 9017 1.16 78 3.ThenatureofsevenfaintX-raypulsators

Table 3.5: Position of X-ray pulsators as determined using the Chandra observations, along with the total counts with which these were detected.

Star Total counts R.A. DEC. error circle

hh:mm:ss ◦ ′ ′′ ′′ SAXJ1324.4–6200 87 13:24:26.70 62:01:19.49 0.65 − SAXJ1452.8–5949 4 14:52:52.70 59:49:08.07 0.95 − AXJ1700.1–4157 63 17:00:04.35 41:58:05.46 0.64 − AXJ1740.1–2847 28 17:40:09.12 28:47:26.02 0.64 − AXJ1749.2–2725 52 17:49:12.41 27:25:38.21 0.64 − AXJ1820.5–1434 33 18:20:30.09 14:34:23.52 0.64 − AXJ1832.3–0840 154 18:32:19.30 08:40:30.44 0.60 − observations were made with the ACIS-I CCDs operating in the FAINT mode for all the sources. The FOV provided by Chandra ACIS-I CCDs is 17′. We processed the event 2 ﬁles using the standard software packages CIAO 4.03 and CALDB 3.4.24. The task wavdetect was used to ﬁnd the exact position of X-ray sources in the image. We listed the position of the X-ray pulsators in Table 3.5 as determined from their Chandra observations. The errors on the position of X-ray pulsators is obtained as a quadratic sum of the bore sight error of the Chandra telescope5, 1-σ wavdetect errors and a contribution that depends on the number of detected counts (van den Berg et al. 2004; especially important for SAX J1452.8–5949) and is also listed in Table 3.5 against each source. Table 3.5 also summarizes the total number of X-ray photons with which X-ray pulsators were detected in their Chandra-ACIS observations.

3.5 X-ray timing analysis

The XMM-Newton EPIC-MOS and pn observations were used for the timing analysis of X-ray pulsators (Table 3.2). The X-ray events were extracted in a circular region of radius

3Chandra Interactive Analysis of Observations (CIAO), http://cxc.harvard.edu/ciao/. 4Chandra Calibration Database (CALDB), http://cxc.harvard.edu/caldb/ 5See http://cxc.harvard.edu/cal/ASPECT/celmon. 3.5 X-ray timing analysis 79

30′′ centered on the position of the target in both EPIC-MOS and EPIC-pn images for almost all X-ray pulsators except for SAX J1324.4–6200 for which a circular region of radius 15′′ was used to extract X-ray source events from EPIC-pn image. We were forced to take smaller circular region in EPIC-pn image of SAX J1324.4–6200 as it was located close to a CCD gap. The background X-ray events were extracted with a similar circular region on the same CCD in a source-free region for both EPIC-MOS and pn. The event times were transformed to barycentric times using the Chandra position of the source and the JPL-DE405 ephemeris with the task barycor in SAS software. The events were then re-binned with a time resolution of 0.1 s and the background was subtracted to generate the background corrected light curves. We added the background-subtracted EPIC-MOS and pn files to generate the final lightcurves. The background corrected EPIC-MOS and pn count rate for all X-ray pulsators are listed in Table 3.3. We searched for the spin-period of all X-ray pulsators in their respective combined EPIC-MOS and pn observations using powspec and refined it further using efsearch in FTOOLS7 software package. Using this technique, we could find the pulse period for all X-ray pulsators except for SAX J1452.8–5949, which did not show any significant peak in the power density spectrum. The pulse period detected for all X- ray pulsators using efsearch is listed in Table 3.6. To further refine their pulse periods, we folded the light curves in chunks of smaller time intervals (in most of the cases 3000 ≈ s each), taking the above period as the best available spin period. The pulse profiles were then cross correlated with a template profile obtained by folding the entire light curve. The cross correlation returned the time of arrivals (TOAs) of each pulse profile with the fiducial point for the measurement fixed to be the peak of the template profile. The technique applied to measure the TOAs and their statistical uncertainties closely resemble the radio pulsar technique (e.g., Taylor 1992). The pulsations were then phase connected with a linear fit to the TOAs to obtain a more precise pulsar spin period. We could apply

7http://heasarc.gsfc.nasa.gov/docs/software/ftools/ftools_menu.html 80 3.ThenatureofsevenfaintX-raypulsators this technique only for SAX J1324.4–6200, AX J1820.5–1434 and AX J1832.3–0840 and their measured spin periods are 172.86 0.02 s (at MJD 54460), 152.28 0.03 s(at ± ± MJD 54373) and 1560 7 s (at MJD 54389) respectively. The errors calculated here ± are with 68% conﬁdence level. The detection of a spin period derivative using a phase coherent technique is prevented given the short baseline of the observation. We ﬁtted all the previous and our new measured values of the spin periods with a linear relation

Ps(t) = P0 + P˙ st, where P˙ s is the spin period derivative of a pulsar and P0 is the spin period of it at the time t = 0. This technique returned us the spin-period derivatives for most of our X-ray pulsators and are listed in Table 3.6. We folded the entire combined XMM-Newton lightcurve of individual sources in one single profile to increase the S/N ratio and measure the harmonic content. Figure 3.1 shows the pulse profile of X-ray pulsators from their XMM-Newton lightcurve in the energy range of 0.2 - 12 keV. The harmonic content of pulsations can be measured as pulsed fractional amplitude defined as (Fmax - Fmin)/(Fmax + Fmin) where Fmax and Fmin are the maximum and minimum flux ( or counts) in a pulse profile and have been measured for all the sources from their XMM-Newton observations (Table 3.6).

We did not detect any pulsations in the combined EPIC-MOS and pn lightcurve of SAX J1452.8–5949 using powspec in FTOOLS software with an upper limit on the pulse fractional amplitude of 18% (at 95% conﬁdence level). To further investigate the presence of pulsations, we divided the EPIC light curve into soft (0.2 - 4 keV) and hard (4

- 12 keV) energy bands with approximately equal count rate. No significant pulse profiles were detected with an upper limit on the pulse fractional amplitude of 25% and 22% (at 95% confidence level) in soft and hard energy bands respectively.

3.6 X-ray spectral analysis

The XMM-Newton EPIC-MOS and pn observations were used for spectral analysis of our selected sample of X-ray pulsators. We used the same extraction region for EPIC- 3.6 X-ray spectral analysis 81

2 SAX J1324.4−6200 AX J1700.1−4157 1.5 1.5

1 1

0.5 0.5

0 0.5 1 1.5 2 0 0.5 1 1.5 2

AX J1740.1−2847 2 AX J1749.2−2725 2

1.5 1.5

1 1

0.5 0.5

0 0 0 0.5 1 1.5 2 0 0.5 1 1.5

AX J1820.5−1434 2 AX J1832.3−0840 2

1.5 1.5

1 1

0.5 0.5

0 0 0 0.5 1 1.5 2 0 0.5 1 1.5 2

Figure 3.1: Pulse profiles of our sample of X-ray pulsators obtained from the XMM- Newton observations. The name of source is written at the top of each pulse profile. No pulsations were detected in SAX J1452.8–5949, thus its pulse profile is not shown. 82 3.ThenatureofsevenfaintX-raypulsators

Table 3.6: The pulse period (measured using the task efsearch in FTOOLS) and pulse fractional amplitude (%) measured using the task efold in FTOOLS. The spin period derivative (P)˙ is also estimated using the previous and present pulse period measurements.

Object Pulseperiod Pulsefractional P˙ 9 1 (seconds) amplitude(%) ( 10− ss− ) × +0.22 SAX J1324.4–6200 172.57 0.22 52 6.34 0.08 − ± SAXJ1452.8–5949 - < 18% - +13.1 AXJ1700.1–4157 723.7 1.7 52 < 53 − +5.6 AXJ1740.1–2847 730.5 1.0 83 - − +1.3 AXJ1749.2–2725 218.1 1.9 79 5.6 3.9 − ± +0.37 AX J1820.5–1434 153.20 0.17 69 2.84 0.82 − ± +2.3 AXJ1832.3–0840 1552.3 0.8 67 10.1 4.9 − ±

MOS and pn data as were used for the timing analysis reported in Section 3.5. SAS tool xmmselect was used to extract both source and background spectra and XSPEC (version 12.0.0) was used for the spectral analysis. The resulting spectra were re-binned to have minimum of 20 counts per bin.

We fitted powerlaw, blackbody, bremsstrahlung and thin-plasma models to the EPIC-MOS and pn X-ray spectra for our selected sample of X-ray pulsators. In most of the cases, both power-law and blackbody models gave equally good fit. Figures 3.2 to 3.5 show the powerlaw model fit to X-ray pulsators where the red and black lines correspond to the MOS1 and MOS2 data respectively and blue line indicate the pn observations. For almost all X-ray pulsators (6 out of 7), bremsstrahlung model gave an unac- ceptable fit with a temperature, kT > 199 keV except for AX J1832.3–0840, for which the bremsstrahlung model with a partial absorption component well fitted the X-ray spectrum with a reduced χ2 of 1.1 for 1083 degrees of freedom. The parameters obtained from this

+10.64 ﬁt are as follows : bremsstrahlung temperature, kT = 29.40 6.9 , interstellar absorption, − +0.05 2 +0.90 2 NH = 0.81 0.05 cm− , partial absorption, NH1 = 6.19 0.76 cm− , partial covering fraction of − − +0.02 0.68 0.02. − The values of the measured ﬁt parameters for all the pulsators are given in Table 3.6 X-ray spectral analysis 83

0.1 SAX J1324.4−6200 −1 keV −1 0.01

10−3 Normalized counts s

χ 0

−5 0.51 2 5 10 Energy (keV)

0.02 SAX J1452.8−5949 −1 0.01 keV −1 5×10−3

2×10−3

10−3 Normalized counts s 1 0 χ −1 −2 2 5 10 Energy (keV)

Figure 3.2: The XMM-Newton EPIC-MOS and pn spectra of SAX J1324.4–6200 and SAX J1452.8–5949 ﬁtted with an absorbed powerlaw model. The EPIC-pn, MOS1 and MOS2 data points are represented by blue, red and black color respectively. 84 3.ThenatureofsevenfaintX-raypulsators

AX J1700.1−4157

−1 0.1 keV −1 0.01

10−3 Normalized counts s 10−4 5

χ 0

−5

1 2 5 10 Energy (keV)

0.1 AX J1740.1−2847 −1 keV −1 0.01

10−3 Normalized counts s 5

χ 0

−5 1 2 5 10 Energy (keV)

Figure 3.3: The XMM-Newton EPIC-MOS and pn spectra of AX J1700.1–4157 and AX J1740.1–2847 ﬁtted with an absorbed powerlaw model. The EPIC-pn, MOS1 and MOS2 data points are represented by blue, red and black color respectively. 3.7 Infraredobservationsanddataanalysis 85

3.7. The observed ﬂux for the ﬁtted models in the 2 - 10 keV energy band is also given in the same table. We have marked the detection of Fe emission line(s) as ‘y’ in Table 3.7 and the details of detected Fe line(s) are given in Table 3.8. In case, no Fe emission line is detected in the X-ray spectrum of an X-ray pulsator, the upper limit on Fe 6.4 keV line

flux is given in Table 3.8, calculated by fixing the line-center at 6.4 keV in their spectrum and fitting a Gaussian to the line. Using the observed X-ray flux, we have estimated the X-ray luminosity of the these sources at a distance of 1 kpc, 8 kpc and 20 kpc (Table 3.9).

3.7 Infrared observations and data analysis

3.7.1 New Technology Telescope

The NIR observations of our sample of X-ray pulsators (except AX J1820.5–1434 and AX J1832.3–0840) were obtained from ESO-NTT archive. These observations were made in

NIR J, H, Ks ﬁlters with a NIR imager and spectrograph SOFI and are summarized in Table 3.10. More details about the SOFI are given in Chapter 2. The instrument was set up in large imaging mode with a pixel scale of 0′′. 29 and a FOV of 5′ 5′ for SAX × J1324.4–6200, AX J1700.1–4157 and AX J1740.1–2847 and in small imaging mode with a pixel scale of 0′′. 14 and a FOV of 2′.5 2′.5 for SAX J1452.8–5949 and AX J1749.2– × 2725 (both in double ‘correlated’ readout mode). The observations of these sources were obtained with good seeing conditions, ranging from 0′′. 6 to0′′. 7 except for AX J1740.1–

2847 which was observed with 1′′. 0 seeing. Images were acquired with the auto-jitter sequence in which a number of single frames (NDIT) having exposure times of DIT (De- tector Integrator Time) seconds were acquired at different positions and then co-averaged to generate an output image. We followed the following steps to reduce the data : first, a master ‘sky’ image is obtained by median stacking all the frames of a source for each filter. This ‘sky’ image is then subtracted from all the images to generate sky-subtracted frames. This step also subtracts bias and dark current contributions. The images are then flat-fielded, aligned 86 3.ThenatureofsevenfaintX-raypulsators

−1 AX J1749.2−2725

keV 0.01 −1

10−3 Normalized counts s

10−4 4 2

χ 0 −2

2 5 10 Energy (keV)

AX J1820.5−1434 −1

keV 0.01 −1

10−3 Normalized counts s

χ 0

−5 1 2 5 10 Energy (keV)

Figure 3.4: The XMM-Newton EPIC-MOS and pn spectra of AX J1749.2–2725 and AX J1820.5–1434 ﬁtted with an absorbed powerlaw model. The EPIC-pn, MOS1 and MOS2 data points are represented by blue, red and black color respectively. 3.7 Infraredobservationsanddataanalysis 87

−1 AX J1832.3−0840 0.1 keV −1

0.01

Normalized counts s 10−3

10 5

χ 0 −5 −10 0.51 2 5 10 Energy (keV)

0.2

−1 AX J1832.3−0840

keV 0.1 −1

0.05

0.02

Normalized counts s 0.01

5×10−3 5

χ 0

−5 6 7 Energy (keV)

Figure 3.5: Top : The XMM-Newton EPIC-MOS and pn spectra of AX J1832.3-0840 fitted with an absorbed blackbody model. The EPIC-pn, MOS1 and MOS2 data points are represented by blue, red and black color respectively. Bottom : The same spectrum but in the energy range 5 - 8 keV to show the Fe emission lines. 88 3.ThenatureofsevenfaintX-raypulsators 03 03 03 03 05 05 02 02 ...... 0.53 0.51 0.08 0.10 0 0 0 0 0 0 0 0 − + − + − + − + 1085 1085 ± ± ± ± / / 0.82 1.90 0.99 0.23 16 43 15 17 97 82 88 06 ...... 0.67 7.04 1.19 6.65 0.12 6.51 0.13 6.43 0 0 0 0 0 2 0 1 115 1.1 115 1.5 / − + − + − + − + / ± ± ± ± 1.41 1.92 8.42 4.64 EPIC observations. 51 12 . . 49 58 34 42 34 95 ...... 3 4 3.02 2.54 1.69 1.60 0.33 1.93 0.17 1.55 55 1.0 55 0.94 0 0 0 0 2 2 / / − + − + − + − + ± ± ± ± 1.50 2.07 7.55 13.23 respectively and are measured in the energy XMM-Newton X uF 16 22 20 23 30 44 13 18 ...... 1.25 2.15 1.10 1.11 0.37 1.49 0.37 1.09 0 0 0 0 0 0 0 0 .1–2847 AX J1749.2–2725 AX J1820.5–1434 AX J1832.3-0840 111 1.11 111 1.17 / / − + − + − + − + ± ± ± ± and X 0.67 2.27 1.10 F ed from their 11 13 12 12 12 16 ...... 1.18 3.26 1.16 3.10 0.19 3.09 0.19 3.06 n line are marked as ‘y’ against Fe line in this Table. 0 0 0 0 0 0 245 0.97 245 0.93 − + − + − + / / ± ± ± ± 0.064 0.24 resented as < 0.56 1.93 0.52 35 36 31 31 67 77 ...... 0.29 3.75 0.25 3.68 0.08 3.27 0.08 3.27 0 0 0 0 0 0 19 1.1 19 1.0 / / − + − + − + ± ± ± ± 0.55 < 0.82 2.04 1.22 15 16 13 15 41 65 48 75 ...... 1.4 0.58 1.3 0.54 0.3 0.50 0.2 0.49 0 0 0 0 0 0 0 0 383 0.6 383 0.6 / − + − + − − + + / ± ± ± ± 1.0 0.95 1.01 6.0 4.5 5.0 4.4 SAX J1324.4–6200 SAX J1452.8–4959 AX J1700.1–4157 AX J1740 2 − 2 − cm cm 1 − 1 − ) 5.81 ) 3.07 2 2 − − ) erg s ) erg s Table 3.7: Spectral parameters of X-ray pulsators as measur 12 (cm 12 (cm − 12 − 12 − 22 − 10 22 10 10 10 10 × 10 ( × × × × ( × (keV) 2.24 X X /ν /ν H H X X 2 ν 2 ν F F Felineχ - - y y - - y uF Felineχ - - y y - - y uF Γ kT Power law N Black Body N Note : The observed and unabsorbed fluxes of these sources rep band 2 - 10 keV. The sources in which we detected the Fe emissio 3.7 Infraredobservationsanddataanalysis 89 of XMM- he 0.03 fixed 0.19 0.02 fixed 0.15 0.02 fixed 0.19 ± ± ± 0.09 0.61 - - - 0.10 0.79 - - - ± ± 0.08 0.17 0.07 0.20 0.02 fixed 0.14 6.96 0.02 fixed 0.15 6.96 0.02 fixed 0.14 6.96 ± ± ± ± ± limit on EW of Fe 6.4 keV line is given in case of non-detection mission lines detected in the X-ray spectrum obtained from t 0.17 1.77 ------0.20 1.04 ------± ± Fe 6.40 keV Fe 6.67 keV Fe 6.97 keV 0.14 0.61 0.03 fixed 0.06 6.67 0.12 0.44 0.03 fixed 0.06 6.68 0.02 fixed 0.06 6.67 ± ± ± ± ± Center width EW Center width EW Center width EW powerlaw fixedpowerlaw fixed fixedpowerlaw 0.12 fixedpowerlaw - - 0.44 6.40 powerlaw - - - fixedpowerlaw - - - 6.69 fixed fixedpowerlaw - - 0.09 6.39 fixed - - - 0.09 ------blackbody 6.50 blackbody fixedblackbody fixed fixed 0.09 fixed - 0.35blackbody - - fixedblackbody - - fixed fixed - - 0.08 fixed - - - 0.06 ------blackbody - - - 6.71 blackbody 6.39 bremsstrahlung 6.39 SAX J1324.4–6200 SAX J1452.8–4959 AX J1700.1–4157 AX J1740.1–2847 AX J1749.2–2725 AX J1820.5–1434 AX J1832.3-0840 observations of X-ray pulsators in the units of keV. An upper Newton any Fe emission line in the X-ray spectrum. Table 3.8: The center, width and equivalent width (EW) of Fe e 90 3.ThenatureofsevenfaintX-raypulsators

Table 3.9: X-ray luminosity of X-ray pulsators at a distance of 1 kpc, 8 kpc and 20 kpc.

Object LX at1kpc LX at8kpc LX at 20 kpc 32 1 34 1 35 1 ( 10 erg s− ) ( 10 erg s− ) ( 10 erg s− ) × × × SAXJ1324.4–6200 5.38 3.44 2.15 SAXJ1452.8–5949 0.65 0.41 0.26 AXJ1700.1–4157 4.40 2.82 1.76 AXJ1740.1–2847 3.71 2.37 1.48 AXJ1749.2–2725 1.33 0.85 0.53 AXJ1820.5–1434 1.91 1.22 0.77 AXJ1832.3–0840 7.95 5.09 3.18 and then average stacked to obtain the ﬁnal images. We used the daophot package within IRAF software for the photometry of the ﬁnal frames(refer to Chapter 2). We used the 2MASS catalog for the calibration and astrometry of these frames. The positions obtained in this way have (absolute) uncertainties of about 0′′. 2, and relative uncertainties far below a fraction of a pixel.

3.7.2 2MASS

X-ray pulsators (AX J1820.5–1434 and AX J1832.3–0840), for which NIR data was not present in the ESO-NTT archive, we obtained the NIR observations from the Two Micron

All Sky Survey (2MASS) in J (1.25 µm), H (1.65 µm) and Ks (2.17 µm) wavebands. The details about 2MASS survey are given in Chapter 2.

3.7.3 Identiﬁcation of near-infrared counterparts

We searched the NIR counterparts of X-ray pulsators in their NTT and 2MASS observations using the position measured from their Chandra observations. The images of X-ray pulsators in NIR Ks waveband, along with the X-ray images obtained using the Chandra telescope are shown in Figure 3.6 to Figure 3.8. The black circle in these Figures represents the XMM-Newton error circle while the white circle represents the Chandra error 3.7 Infraredobservationsanddataanalysis 91

5" SAX J1324.4-6200

C1 C2

N N

E E

5" SAX J1452.8-5949

N N

E E

5" AX J1700.1-4157

N N

E E

Figure 3.6: Left : The NIR Ks waveband image of SAX J1324.4–6200, SAX J1452.8– 5949 and AX J1700.1–4157 taken with ESO-NTT. The X-ray positions of all these sources measured from their XMM-Newton observations are represented by a black circle and the positions measured from their Chandra observations are represented by a white circle. The radius of the circle represents an error on their position measurement. Right : The Chandra ACIS-I images of the same pulsators. The black circle represents an error circle on their position as measured using XMM-Newton observations. 92 3.ThenatureofsevenfaintX-raypulsators

5" AX J1740.1-2847

N N

E E

5" AX J1749.2-2725

N N

E E

AX J1820.5-1434 5"

N N

E E

Figure 3.7: Left : The ESO-NTT NIR Ks waveband image of AX J1740.1–2847, AX J1749.2–2725 and 2MASS K band image of AX J1820.5–1434. The color and radius of the the circles are as explained in the Figure 3.6 except AX J1820.5–1434 for which its

XMM-Newton error circle on the position is represented by a white circle in its Ks waveband image. Right : The Chandra ACIS-I images of the same pulsators. The black circle represents an error circle on their position as measured using XMM-Newton observations. 3.7 Infraredobservationsanddataanalysis 93

Table 3.10: Log of the near-infrared observations of ﬁve X-ray pulsators taken with 3.52- m ESO-NTT. NDIT represents the Number of single frames, having exposure times of DIT (Detector Integrator Time) seconds and are used to generate an output image having exposure time equal to one DIT. The numberof output frames are represented by Nframes.

Source Date JHKs FWHM (UT) DIT NDIT Nframes DIT NDIT Nframes DIT NDIT Nframes (arcsec) SAXJ1324.4–6200 06 Feb 2001 3 5 8 3 5 8 3 5 12 0.6 SAXJ1452.8–5949 08 Feb 2001 3 5 10 3 5 12 3 5 16 0.6 AX J1700.1–4157 20 Mar 2001 3 5 4 3 5 4 3 5 6 0.6 AX J1740.1–2847 18 Jun 2002 3 20 5 3 20 5 3 20 5 1.0 AX J1749.2–2725 20 Mar 2001 3 5 4 3 5 4 3 5 6 0.7

5" AX J1832.3-0840

N N

E E

Figure 3.8: Left : The NIR Ks waveband image of X-ray pulsator AX J1832.3–0840 taken from the 2MASS point source catalog. The XMM-Newton error circle (in black color) and the Chandra error circle (in white colour) on the position are also plotted. Right : The Chandra ACIS-I image of AX J1832.3–0840 along with the error circle on the position as measured using the XMM-Newton observations. 94 3.ThenatureofsevenfaintX-raypulsators ulsators. 0.02 0.34 0.04 0.10 0.13 0.12 0.11 s ± ± ± ± ± ± ± 0.07 15.15 0.21 16.06 0.07 11.75 0.07 15.57 0.05 17.38 0.12 17.93 0.09 14.97 ± ± ± ± ± ± ± 0.23 16.22 0.15 13.25 0.03 15.76 0.04 17.04 0.12 17.75 0.11 16.61 ± ± ± ± ± ± JHK magnitude magnitude magnitude ◦ ′ ′′ 14:34:22.9 15.41 08:40:30.53 17.11 27:25:38.25 not detected 16.89 28:47:25.68 16.16 41:58:05.52 17.07 59:49:07.92 18.59 62:01:19.59 19.57 − − − − − − − magnitudes of the most likely NIR counterparts of the X-ray p s K and hh:mm:ss H , J AX J1832.3–0840 18:32:19.386 AX J1820.5–1434 18:20:30.10 AX J1749.2–2725 17:49:12.408 AX J1740.1–2847 17:40:09.14 AX J1700.1–4157 17:00:04.35 SAX J1452.8–5949 14:52:52.72 StarSAX J1324.4–6200 13:24:26.71 R.A. DEC. Table 3.11: The observed 3.7 Infraredobservationsanddataanalysis 95 circle except in AX J1820.5–1434 where the XMM-Newton error circle is also represented by a white colour circle. For all the sources, we have clearly identiﬁed the NIR counterparts, shown in Figure 3.6 to Figure 3.8. The J, H and Ks wavebands magnitudes of the most likely NIR counterparts are listed in Table 3.11. Here we do not deny that the NIR counterpart identiﬁed for the two sources (AX J1820.5–1434 and AX J1832.3–0840) in the 2MASS observations could be a combination of a few nearby stars.

3.7.4 Spectral type of the NIR counterparts

Using the observed hydrogen column density (NH) of the X-ray pulsators from the X-ray spectral ﬁt (mostly using powerlaw model; Table 3.7), the extinction in J, H and Ks wavebands is calculated (Predehl and Schmitt 1995; Fitzpatrick 1999) and is given in Table 3.12. Here, we assume that there is no local contribution of the absorbing matter from the

X-ray source and the companion star would experience the sameNH as observed by the X-ray source (Here please note that this is not necessarily true and in that case our conclusions about the possible nature of the companion star could be signiﬁcantly aﬀected).

However, NH measured this way can be used as an upper limit on extinction towards the

X-ray pulsators. The NH measured towards AX J1749.2-2725 and AX J1820.5-1434 using the powerlaw model is very high as compared to the Galactic NH in that direction (Dickey and Lockman, 1990). Thus, in both these sources we expect that the material local to the X-ray source is contributing signiﬁcantly towards the total NH, hence the extinction for them is calculated by considering the Galactic NH (Dickey and Lockman,

1990). For a comparison, we have given NH measured by both absorbed powerlaw model

ﬁt to the X-ray spectrum and the Galactic NH (Dickey and Lockman, 1990) in Table 3.12. We adopted two diﬀerent methods to make a rough estimation of the spectral type of the NIR counterparts and are described below. From here onwards, we would like to call these methods as ‘NIR method 1’ and ‘NIR method 2’.

NIR method 1 96 3.ThenatureofsevenfaintX-raypulsators

This method is based on the distance measurement of a given star from its extinction free NIR fluxes using the blackbody model. For the given dereddend magnitudes of a star in J, H and Ks wavebands (Table 3.13), we calculated the distance of different types of stars (main-sequence stars, supergiants and giants; Cox 2000) using the blackbody model. For the given magnitudes, a given spectral type of a star is rejected if the blackbody model gives a unrealistic distance for it. Hence, we broadly identified the NIR counterparts as early-type stars, supergiants, giants or late-type stars and is also explained with the help of an example as follows. If the dereddened magnitudeof a starin ’J’ waveband is 16.26 mag, then for this star to be an early-type star of spectral type B2, it should lie at a distance of 47 kpc while for it to be a late-type star of spectral type K5, it should ∼ lie at a distance of 1.5 kpc. It is very unlikely that a given star is present beyond 20 kpc (size of our Galaxy), thus the companion star of this source is most likely a low mass star.

NIR method 2

In this method, we have used the NIR multiband ﬂuxes at their corresponding wavelengths for a given source and a blackbody curve of an early-type star and a late-type star is compared to the observed dereddened ﬂuxes (Table 3.13). The observed multiband dereddend magnitudes of a given star following the blackbody curve of temperature, T =

3000 K is classified as a late-type star while the one followingthe T = 20,000 K is classified as an early-type star. The observed multiband NIR fluxes, along with the blackbody curve at T = 3,000 K (blue) and T = 20,000 K (red) for all pulsators are shown in Figure 3.9, where the observed fluxes are represented by filled circles and the extinction corrected

ﬂuxes are represented by open circles. During our observations, AX J1749.2–2725 was detected only in H and Ks wavebands, thus we did not include its multiband NIR ﬂuxes plot in Figure 3.9. The results obtained using these two methods for individual sources are discussed in Section 3.9. 3.7 Infraredobservationsanddataanalysis 97

Table 3.12: Galactic extinction in V, J, H and Ks wavebands for our sample of X-ray pulsators

g Object NH NH VJHKs 22 2 22 2 ( 10 cm− ) ( 10 cm− ) (mag) (mag) (mag) (mag) × × SAXJ1324.4–6200 5.81 1.39 32.54 9.01 5.56 3.77 SAXJ1452.8–5949 1.22 1.85 6.83 1.89 1.17 0.79 AXJ1700.1–4157 0.52 1.75 2.91 0.81 0.50 0.34 AXJ1740.1–2847 1.10 0.92 6.16 1.71 1.05 0.71

AX J1749.2–2725∗ 13.23 1.52 8.51 2.36 1.46 0.99

AX J1820.5–1434∗ 8.42 1.56 8.74 2.42 1.49 1.01 AXJ1832.3–0840 0.99 1.83 5.54 1.54 0.95 0.64

NH - neutral hydrogen column density measured using powerlaw model ﬁt to the X-ray spectrum, g NH - neutral hydrogen column density in the direction of the given source as measured using radio HI emission (Dickey and Lockman, 1990) - the extinction reported for this source in this Table is calculated using Galactic N ∗ H

Table 3.13: Dereddened NIR magnitudes of counterparts of X-ray pulsators in J, H and

Ks wavebands.

Star JHKs magnitude magnitude magnitude SAX J1324.4–6200 10.56 0.11 11.05 0.09 11.20 0.11 ± ± ± SAX J1452.8–5949 16.70 0.12 16.58 0.12 17.14 0.12 ± ± ± AX J1700.1–4157 16.26 0.04 16.54 0.05 17.04 0.13 ± ± ± AX J1740.1–2847 14.45 0.03 14.71 0.07 14.86 0.10 ± ± ± AXJ1749.2–2725 notdetected 15.43 0.07 14.16 0.02 ± ± AX J1820.5–1434 12.99 0.15 11.76 0.07 10.74 0.04 ± ± ± AX J1832.3–0840 15.57 0.23 15.27 0.21 15.42 0.34 ± ± ± 98 3.ThenatureofsevenfaintX-raypulsators

SAX J1324.4−6200 −8 −10 SAX J1452.8−5949 T = 20000K −2

−2 T = 20000K cm cm −1

−10 −1 −12 T = 3000K ) erg s λ ) erg s T = 6000K λ F F λ −12

λ −14 log ( log ( −14 −16 JHK JHK 0.51 2 5 0.51 2 5 λ (µm) λ (µm) −8 −8 AX J1700.1−4157 AX J1740.1−2847 −2 −10 −2 −10 T = 20000K cm T = 20000K cm −1 −1

−12 −12 ) erg s ) erg s λ λ T = 3000K F F λ T = 3000K λ log ( −14 log ( −14

JHK JHK −16 −16 0.51 2 5 0.51 2 5 λ (µm) λ (µm) −8 AX J1820.5−1434 −8 AX J1832.3−0840 T = 20000K −2 −2 −10 T = 20000K cm cm −10 −1 −1

−12

T = 3000K ) erg s ) erg s λ λ F

F −12 λ λ T = 3000K

log ( −14 log ( −14 JHK JHK −16 0.51 2 5 0.51 2 5 λ (µm) λ (µm)

Figure 3.9: The observed (ﬁlled circles) and extinction-free (open circles) spectral energy densities of NIR counterparts of our sample of X-ray pulsators. The solid line (red color) and the dashed-dotted line (blue color) represent blackbody model spectral energy density of the star at temperature, T = 20,000 K and T = 3,000 K respectively and are shifted to match the extinction-free spectral energy density of the star at NIR H band. 3.8 HMXBs,LMXBsandIntermediatePolars-acomparison 99

3.8 HMXBs, LMXBs and Intermediate Polars - a comparison

Our sample of X-ray pulsators is selected on the basis of their low luminosity and long pulse periods. Three classes of X-ray pulsators exist at low luminosities (1032 - 1038 erg

1 s− ) and long pulse periods (> 150 s): high mass X-ray binary pulsar, low mass X-ray binary pulsar and Intermediate Polar (IP; refer to Appendix A) and our selected pulsators could be members of one of these classes. In the present work, we have classiﬁed X-ray pulsators as a member of one of the above mentioned classes by matching the observed properties with the common properties seen in these three classes of X-ray pulsators. Before we discuss the nature of individual sources, a brief discussion on the common properties of these three classes of X-ray pulsators are given here.

High mass X-ray binary pulsars These pulsars have been observed with pulse periods ranging from few seconds to over a thousand seconds (Bildsten et al., 1997). The pulse proﬁles of these sources show a variety of shapes from a complete sinusoidal to a double peak (refer to Chapter 1). These sources usually display hard X-ray spectrum well ﬁtted with a powerlaw index, Γ 1.0 or with a blackbody model of temperature, kT 2 keV. Sometimes, their X-ray ∼ ∼ spectrum also show a strong neutral Fe emission line at 6.4 keV and very rarely ionized Fe emission lines at 6.7 and 6.9 keV (Ebisawa et al., 1996). The pulse period derivative,

7 11 1 P˙ in these sources have been measured from 10− to 10− s s− . The identiﬁcation of high mass companion star in these sources directly conﬁrm the high mass X-ray binary nature of these pulsars. Most of these sources are present beyond 1 kpc from Earth in the Galactic plane, mostly concentrated in the Galactic arms.

Low mass X-ray binary pulsars

The number of X-ray pulsars in LMXBs is very small ( 7 in our Galaxy). Most ∼ of these LMXB pulsars have been observed with a spin periods in the 5 - 9 s range except 100 3.ThenatureofsevenfaintX-raypulsators a very few LMXB pulsars like GX 1+4, which has got a pulse period of 140 s. Both sinusoidal single peak and double peak pulse proﬁles are seen in these sources. This class of pulsars usually display soft X-ray spectrum well ﬁtted with a powerlaw model of photon index, Γ 2 with no Fe emission lines. The pulse period derivatives of this class ∼ 8 11 of pulsars show a wide range e.g., in GX 1+4, P˙ is seen to be varied from 10− to 10−

1 ss− (Ferrigno et al., 2007). These sources are uniformly distributed in the Galaxy at a distance of 2 kpc to 8 kpc. ∼ Intermediate Polars Intermediate Polars (IPs) have been observed with typical pulse periods in the range of a few hundred seconds to a few thousand seconds (Kuulkers et al., 2006). The pulse profiles of most of the IPs with pulse period < 700 s are double-peaked while the slow spinning IPs show a single peak pulse profiles (Norton 2001). The observed X-ray spectrum of mostof theIPs are well fitted with a thin plasmamodel with a plasma shocked temperatures of 1 - 30 keV or with a bremsstrahlung model of temperature 1 - 30 keV. ∼ ∼ The X-ray spectra of most of the IPs are characterized by the strong Fe emission lines at

6.4, 6.7 and 6.9 keV (Ezuka and Ishida, 1999). This class of pulsators are slow systems

10 1 and the maximum spin-period derivative measured in these systems is 10− s− . The ∼ 30 32 1 IPs have a luminosity range of 10 - 10 erg s− and almost all of them lie within 1 kpc distance from Sun.

3.9 Discussion

We carried out observations with the ACIS-I instrument onboard Chandra and the EPIC instrument onboard XMM-Newton space observatories to investigate the nature of seven X-ray pulsators. In addition, we obtained NIR observations from ESO-NTT archive for ﬁve of these pulsators and 2MASS observations for rest of them.

We determined the position of X-ray pulsators with a sub arc-second accuracy with the help of their Chandra observations and it is the best position measurement made 3.9 Discussion 101 so far for these sources. The position determined this way helped us to find the NIR counterparts of all the sources in their NIR observations. The X-ray observations obtained using XMM-Newton satellite helped us to check the stability of the X-ray pulsations and the pulse profiles. Pulse period derivative of these sources is also measured with the help of XMM-Newton observations. The XMM-Newton observations of these sources also gave the best spectral measurement made so far for these sources and could resolve the different

Fe emission lines present in their X-ray spectrum. With the help of observed NH and the

J, H and Ks magnitudes of NIR counterparts, we broadly identiﬁed the spectral type of the NIR counterparts and hence could draw important conclusions about these sources. In this Section, we have discussed the individual sources in the light of their previous and present observations.

SAX J1324.4–6200 SAX J1324.4–6200, which has a pulse period of 170.84 0.04 s, was discovered ± serendipitously from observations of the low mass X-ray binary (LMXB) XB 1323–619 performed on August 22, 1997 with the BeppoSAX (Angelini et al., 1998). It showed sinusoidal pulse proﬁle with a pulse fractional amplitude of (52 5)%. The X-ray (1.8 ± - 10 keV) spectrum of the source could be ﬁtted with either an absorbed powerlaw with

+2.7 22 2 photon index, Γ = 1.0 0.4 and hydrogen column density, NH = 7.8 1.1 10 cm− or ± − × +3 22 with a blackbody model with a temperature, kT = 2.4 0.4 keV and NH = 4 2 10 ± − × 2 cm− (Angelini et al., 1998). The X-ray ﬂux in 1 - 10 keV energy band was estimated to

12 2 1 be 8.2 10− erg cm− s− . After its discovery with BeppoSAX, a re-analysis of an ASCA × observations performed earlier in 1994 showed that it was also detected in observations of XB 1323–619 performed on August 4, 1994 (Angelini et al., 1998). In addition, pointed ASCA observations (of 187 ks) were performed on SAX J1324.4–6200 on February 2, 2000 and the source was detected with a pulse period of 171.2 s (Lin et al., 2002). ∼ A possible orbital period of 27 hr was found by Lin et al. (2002) and the system was suggested as a LMXB pulsar (Lin et al., 2002). During short observations performed 102 3.ThenatureofsevenfaintX-raypulsators

Table 3.14: Spin-period history of SAX J1324.4-6200

Telescope Date Spin period References (UT) (s) ASCA 04 Aug1994 170.35 0.48 Angelini et al. 1998 ± BeppoSAX 22 Aug1997 170.84 0.04 Angelini et al. 1998 ± ASCA 02Feb2000 171.25 0.01 Linetal.2002 ± Swift 30Dec2007 172.84 0.1 Mereghettiet al.2008 ± XMM-Newton 11Jan2008 172.86 0.02 present work ± using Swift on December 30, 2007, SAX J1324.4–6200 was detected with a pulse period of 172.8 s (Mereghetti et al., 2008) which would imply a spin down rate over the last 10

9 1 years, P˙ = 6 10− s s− . Mereghetti et al. (2008) identiﬁed a possible 2MASS NIR ∼ × counterpart in the Swift error circle of SAX J1324.4–6200 with a K band magnitude of

14.39 0.08 and suggested that the source could be a persistent Be accreting X-ray pulsar. ± During our XMM-Newton observation, SAX J1324.4–6200 was detected with a pulse period of 172.85 0.02 s and with a nearly sinusoidal single peak pulse proﬁle. ± The pulse fractional amplitude of SAX J1324.4–6200 measured during our observations is 52%. The spin-period history of SAX J1324.4–6200 clearly shows a linear increase in spin-period from 170.35 s to 172.86 s over the last 14 years (Table 3.14) and using it, we

9 1 calculated the spin period derivative, P˙ = (6.34 0.08) 10− ss− . The XMM-Newton X- ± × ray spectra of SAX J1324.4–6200is fitted well with both powerlaw and blackbody models with photon index, Γ = 1.0 and temperature, kT = 2.2 keV respectively. No significant Fe 6.4 keV emission line is detected in the X-ray spectrum of SAX J1324.4–6200 and we derived an upper limit of 118 eV on the equivalent width of it (at 90% confidence

12 2 1 level). The observed flux in 2 - 10 keV band is estimated to be 4.5 10− erg cm− s− . × Our ESO-NTT observations of SAX J1324.4–6200 clearly showed that at the location of the NIR counterpart identified by Mereghetti et al. (2008) in the 2MASS data, there are three different stars marked as C1, C2 and C3 in Figure 3.6. However, our accurate position determination with the Chandra observations of SAX J1324.4–6200 helped us 3.9 Discussion 103 to identified star ‘C1’ as the most likely NIR counterpart of SAX J1324.4–6200, with a magnitude in K waveband = 14.97 0.11 mag. Using ‘NIR Method 1’ (refer Section s ± 3.7.4), we found that the NIR counterpart could be a late-type main sequence star (M through A type) at a distance of 1.4 kpc or a main sequence early-type star (B type) at ≤ a distance 8 kpc or a late-type giant at a distance of 9 kpc for the given NIR fluxes. ≤ ∼ ‘NIR Method 2’ indicate the NIR counterpart of SAX J1324.4–6200 to be an early-type star (Figure 3.9). If we assume the NIR counterpart of SAX J1324.4–6200 to be a late-type star at a distance 1.4 kpc, then the X-ray flux of the system would be low as compared to the ≤ other low mass X-ray binaries known and the system would be a IP or a high mass X-ray binary pulsar. The X-ray spectrum of SAX J1324.4–6200 is typical of a high mass X-ray binary with no detection of strong Fe emission lines. The X-ray timing properties reveal

9 1 the pulsar to be spinning down with a spin period derivative of 10− ss− , which is quite high for it to be an IP. The X-ray pulsars with giant stars as a companion are quite rare and usually show large variabilities on longer timescales. This source seems to be emitting at a constant rate for more than 10 years, thus it is not likely that it is a X-ray pulsar with a giant companion star. With the above arguments, we classify it as a high mass X-ray binary pulsar.

SAX J1452.8–5949 SAX J1452.8–5949 was discovered with a spin period of 437.4 1.4 s from the ± observations made on July 20, 1999 using BeppoSAX as a part of ‘Galactic plane survey’

(Oosterbroek et al., 1999). The pulse profile of SAX J1452.8–5949 was measured to be nearly sinusoidal with a pulse fractional amplitude of (75 25)%. The X-ray spectra ± of the source could be fitted well with an absorbed powerlaw model of Γ = 1.4 0.6, ± 13 2 1 and using it, the X-ray flux was estimated in this source to be 7 10− erg cm− s− . × Oosterbroek et al. (1999) suggested that SAX J1452.8–5949 could be an accreting Be

34 1 X-ray pulsar at a distance of 6 12 kpc with the X-ray luminosity, L 10 erg s− . − X ∼ 104 3.ThenatureofsevenfaintX-raypulsators

We did not detect any pulsations in our XMM-Newton observations of SAX

J1452.8–5949 with a pulse fractional amplitude greater than 18% (0.2 - 12 keV). The XMM-Newton X-ray spectrum of SAX J1452.8–5949 fitted well with a powerlaw model of photon index, Γ = 0.83 (Table 3.7) and with no significant detection of Fe emission lines. We derived an upper limit on the equivalent width of Fe 6.4 keV line in this source to be 436 eV. A faint star is detected in the NIR ESO-NTT images at the position of SAX J1452.8–5949 inside the Chandra error circle (Figure 3.6) and is identified as the most likely NIR counterpart of SAX J1452.8–5949.

Both neutron star and a white dwarf X-ray pulsar are consistent with the inferred X-ray spectral parameters of SAX J1452.8–5949, thus it is difficult to distinguish between a neutron star or an accreting white dwarf from the X-ray spectral information. Also with the quite high upper limit on equivalent width of Fe 6.4 keV emission line, we cannot deny the presence of it in the X-ray spectrum. The spectral parameters from our present XMM-Newton observations of SAX J1452.8–5949 are comparable with the one reported by Oosterbroek et al. (1999). ‘NIR Method 1’ suggests the NIR counterpart to be a late spectral type main sequence star (M through A type) at a distance 10 kpc and ‘NIR ≤ Method 2’ indicate a temperature of < 20,000 K (Figure 3.9). With the above arguments, we can say that if SAX J1452.8–5949 is in a binary system, it must have a low mass companion (LMXB or IP), regardless of its pulsation property. The non-detection of pulsations in SAX J1452.8–5949 is in contrast with the previous detection made by Oosterbroek et al. (1999) who detected pulsations with a fractional amplitude of (75 25)% (at 90% confidence level). To explain this discrepancy ± we are left with two possibilities: the fractional amplitude of the pulsations decreased or the pulsations observed by Oosterbroek et al. (1999) were spurious (as the detection significance of pulsations is less than 3σ). In the latter case, SAX J1452.8–5949 can be any non pulsating source in the Galaxy, like LMXB, accreting white dwarf, BY Dra, RS CVn or active star, etc. On the other hand, if the detection of pulsations by Oosterbroek 3.9 Discussion 105 et al. (1999) was real, then the pulse amplitude was reduced by a substantial amount. This behaviour has been observed in other LMXBs also where the pulsed fraction decreased on the time-scale of days (Her X-1; Ramsay et al. 2002), hours (GX1+4; Naik et al. 2005) and minutes (4U 1907+09; in ’t Zand et al. 1997). Given the large value for the upper limit on pulse fractional amplitude we obtained, the possibility that SAX J1452.8–5949 is a slow pulsating LMXB or an IP is still open.

AX J1700.1–4157 AX J1700.1–4157 was discovered with a pulse period of 714 0.3 s from obser- ± vations performed on September 16, 1997 using the ASCA as a part of its ‘Galactic plane survey’ (Torii et al., 1999). The pulse profile was characterized by a broad single peak and the pulse fractional amplitude was measured to be 50% of the source flux. The X-ray ≈ spectrum of AX J1700.1–4157 could be fitted well with a powerlaw model of Γ = 0.7 and

22 2 12 N = 6 10 cm− and using it, the observed X-ray flux was estimated to be 6 10− H × × 2 1 erg cm− s− . The source was detected twice in the previous observations made using the ASCA on March 24, 1994 and September 1, 1996 but no pulse period measurements could be made due to poor statistics (Torii et al., 1999). On the basis of its hard X-ray spectrum and non-detection of an Fe emission line, AX J1700.1–4157 was suggested to be a Be/X-ray pulsar similar to X-Persei (Torii et al., 1999). During our XMM-Newton observations, AX J1700.1–4157 is detected with a pulse period of 723.7 7.4 s and with a pulse fractional amplitude of 51%. The pulse ± profile of AX J1700.1–4157 is nearly sinusoidal with a single peak. The pulse fractional amplitude and pulse profile obtained from our observations is comparable with the previous measurements made by Torii et al. (1999). Using the previous and present pulse

8 1 period measurements, we calculated an upper limit on P˙ = 5.2 10− ss− . The observed × X-ray spectrum of AX J1700.1–4157 is ﬁtted well with both the powerlaw model of Γ =

22 2 0.56 and N = 0.52 10 cm− and a blackbody model of kT = 1.9 keV and N 0.064 H × H ≤ 22 2 10 cm− and is also comparable to the previous measurements. However, during our × 106 3.ThenatureofsevenfaintX-raypulsators observations a strong Fe emission line at 6.7 keV is detected in the X-ray spectrum of AX

1700.1–4157 with an equivalent width of 800 eV (Figure 3.3) which was not detected in the previous observations. The X-ray ﬂux measured from the X-ray spectral ﬁt is 3.7 ∼ × 12 2 1 10− erg cm− s− . We detected a NIR counterpart at the position of AX J1700.1–4157 with the J, H and Ks magnitudes given in Table 3.11. ‘NIR method 1’ suggests that the NIR counterpart of AX J1700.1–4157 is a late-type star and ‘NIR method 2’ indicate the temperature of it to be 20,000 K. ∼ The detection of strong Fe 6.7 keV emission line in X-ray spectrum of AX

J1700.1–4157 and the low mass companion as suggested by NIR observations indicate it to be an IP rather than a high mass X-ray binary pulsar. On the basis of present observations, we suggest AX J1700.1–4157 to be an IP. However the NIR spectroscopic observations of this system would further help us to conﬁrm the nature of its counterpart.

AX J1740.1–2847

AX J1740.1–2847 was discovered with a pulse period of 729 14 s using ASCA ± observations performed on September 7 - 8, 1998 as a part of the ‘Galactic plane survey’ (Sakano et al., 2000). Its pulse period was characterized by a single (main) peak, and a weak (sub) peak at the middle of the pulse phase. The pulse fractional amplitude of main peak was measured nearly 100% of the source ﬂux. Apart from pulsations, this source also showed a signiﬁcant aperiodic variability. The X-ray spectrum of the source could

+0.6 +2.9 22 2 be fitted well with a powerlaw model of Γ = 0.7 0.6 and NH = 2.5 1.8 10 cm− and − − × 12 2 1 the X-ray flux of AX J1740.1–2847 was estimated to be 4.1 10− erg cm− s− . It was × suggested to be a faint X-ray pulsar similar to X-Persei on the basis of its low-luminosity and long pulse period (Sakano et al., 2000). From our XMM-Newton, we detected it with a pulse period of 730.46 3.0 s and ± with a pulse fractional amplitude of 83%. With the present large error on the pulse period of AX J1740.1–2847, it is difficult to find whether AX J1740.1–2847 is spinning-up or spinning down. The pulse profile of AX J1740.1–2847 is characterized by a main peak 3.9 Discussion 107 and a weak sub-peak, similar to its previous ASCA observations. The X-ray spectrum of

AX J1740.1–2847 is well ﬁtted with a powerlaw model of Γ = 0.67 and N = 1.1 1022 H × 2 22 2 cm− and a blackbody model with kT = 2.3 keV and N = 0.24 10 cm− . The spectral H × parameters obtained from our XMM-Newton observations are similar to the one reported using ASCA observations (Sakano et al., 2000). We have also detected a strong Fe 6.4 keV emission line in the X-ray spectrum with an equivalent width of 1.8 keV. The X-ray

12 2 1 ﬂux measured using both powerlaw and blackbody models is 3.1 10− erg cm− s− ∼ × and is similar to the X-ray ﬂux measured using ASCA observations. We detected a NIR counterpart at the Chandra position of this source (Figure 3.7). ‘NIR method 1’ suggests the NIR counterpart of AX J1740.1–2847 to be a late-type star at a distance of 1-2kpc ≤ and ‘NIR method 2’ suggests a star with a temperature of less than 20,000 K. The hard X-ray spectra and detection of Fe emission line at 6.4 keV suggest it to be either a HMXB pulsar or an IP. The NIR observations of this source suggest its counterpart to be a low mass star. With the present observations, we suggest the system to be more likely an IP.

AX J1749.2–2725 AX J1749.2–2725 was discovered with a pulse period of 220.38 0.20 s from ± the ASCA observations performed on March 26, 1995 as a part of its ‘Galactic plane survey’ (Torii et al., 1998). The pulse shape of AX J1749.2–2725 was broad and a single peaked in the 4 - 10 keV energy band while no signiﬁcant pulsations were detected below 4 keV. AX J1749.2–2725 also showed a high aperiodic variability with an rms of 35% ∼ for the timescale of 1024 s. The X-ray spectrum of AX J1749.2–2725 could be ﬁtted well

23 2 with a powerlaw model of photon index, Γ = 1.0 and N 10 cm− . The X-ray flux H ∼ 12 1 2 of AX J1749.2–2725 was estimated to be 0.3 10− erg s− cm− . AX J1749.2–2725 × lies close to another X-ray source GRO J1750–27, thus it has been observed three times during the observations of GRO J1750–27. During the second observations performed in September 1995, AX J1749.2–2725 was detected with almost the same pulse period and 108 3.ThenatureofsevenfaintX-raypulsators spectral parameters as it was detected during first observations. However, during the third and fourth observations performed in September 1996 and September 1997 respectively, AX J1749.2–2725 was detected at much fainter levels as compared to the first and second observations and with no significant detection of pulsations.

During XMM-Newton observations, AX J1749.2–2725 is detected with a pulse period of 218.08 1.6 s and with a pulse fractional amplitude of 79%. Using the previous ± and present measurements of pulse period, we calculated the pulse period derivative of

9 1 AX J1749.2–2725 to be, P˙ = 1.7 10− ss− . The pulse proﬁle of AX J1749.2–2725 is × characterized by a broad single peak. The X-ray spectrum of this source is ﬁtted well with both the powerlaw and the blackbody model with a Γ = 1.5 and kT = 2 keV respectively,

2 with a reduced χ of 1.2 for 59 degrees of freedom. The NH estimated from both the

22 2 models is in the range of (7.6 - 13.2) 10 cm− and is comparable to the N estimated × H in the previous observations made by ASCA. We detected a faint NIR counterpart of AX

J1749.2–2725 at its Chandra position with a Ks magnitude = 15.15. In this particular source, the NH measured from the X-ray spectral ﬁt is very large as compared to the

Galactic NH towards its direction (Dickey and Lockman 1990; Table 3.12), implies that the material local to this source is contributing eﬀectively to the total NH. Thus, for this particular source we have used the Galactic NH for extinction measurements and is given in Table 3.12. The ‘NIR method 1’ suggests that the NIR counterpart of AX J1749.2–2725 could be a A-type star at a distance of 8 kpc or a late-type star within 1 ∼ kpc distance from Sun. However, due to the large uncertainty involved in the extinction measurement using NH, we would not like to draw any conclusion about the spectral type of the NIR counterpart of AX J1749.2–2725. We could not use ‘NIR method 2’ for AX J1749.2–2725 as the NIR counterpart was not detected in J waveband during our NTT observations. The X-ray pulsator AX J1749.2–2725 is spinning up with a spin period deriva-

9 1 tive, P˙ of 1.7 10− s s− , which corresponds to a typical value of neutron star pulsars × 3.9 Discussion 109 and is rare for IPs. The X-spectrum of AX J1749.2–2725 also suggest that it is a neutron star system rather than a white dwarf system. Thus, the present observations indicate AX J1749.2–2725 to be a neutron star X-ray binary pulsar rather than an IP.

AX J1820.5–1434 AX J1820.5–1434 was detected with a pulse period of 152.26 0.04 s from ± observations made on April 8 - 12, 1997 using ASCA as a part of its ‘Galactic plane survey’ (Kinugasa et al., 1998). The pulse profile of AX J1820.5–1434 was characterized by a broad peak with a phase width of 0.3. The X-ray spectrum of AX J1820.5–1434 could be fitted well with both powerlaw and blackbody model with a Γ = 0.9 0.2 and ± 23 2 kT = 2.6 0.2 keV respectively and N = (0.5 -1) 10 cm− . The observed flux of AX ± H × 11 2 1 J1820.5–1434 was estimated to be 2.3 10− erg cm− s− . This source was suggested ∼ × to be a high mass X-ray binary pulsar (Kinugasa et al., 1998) on the basis of its X-ray timing and spectral properties.

During our XMM-Newton observations, AX J1820.5–1434 was detected with a pulse period of 153.20 0.27 s and a pulse fractional amplitude of 69%. The pulse proﬁle ± of AX J1820.5–1434 was characterized by a nearly sinusoidal single peak. We have

9 1 calculated the pulse period derivative, P˙ of this source to be 2.8 10− ss− using the ∼ × previous and present measurement of pulse period. The X-ray spectrum of AX J1820.5– 1434 is ﬁtted well with a powerlaw and a blackbody model of Γ = 1.41 and kT = 1.9 keV respectively (Table 3.7). Using these models, NH is measured in the range of (4.6 - 8.4)

22 2 10 cm− . The spectral parameters measured from our observations are comparable to × the previous observations. We did not detect any Fe emission line in the X-ray spectrum and an upper limit on the equivalent width of 6.4 keV emission line is estimated to be 90 eV. We detected a NIR counterpart in the 2MASS data at the Chandra position of ∼

AX J1820.5–1434 with a magnitude in Ks waveband = 11.75 mag. It is possible that the NIR counterpart detected in the 2MASS observations is a combination of two or more stars. The Galactic NH measured in the direction of AX J1820.5–1434 by Dickey and 110 3.ThenatureofsevenfaintX-raypulsators

Lockman (1990) suggests that most of the NH measured from the spectral ﬁt is local to the X-ray source and hence for the present source, we have calculated the extinctions using the Galactic NH and is listed in Table 3.12. The ‘NIR method 1’ suggests the NIR counterpart to be a low-mass star at a distance of 1 - 2 kpc. However, we would not like to draw any conclusions on the nature of the NIR counterpart on the basis of present 2MASS observations. The hard X-ray spectrum of AX J1820.5–1434 with a non-detection of strong Fe emission line suggests it to be a neutron star binary system. This system is spinning down

9 1 with a spin period derivative, P˙ of 2.8 10− ss− which is quite high for a typical IP. On × the basis of X-ray spectral and timing properties, we would like to classify this system to be a member of high mass X-ray binary pulsars. However, further NIR imaging and spectroscopic observations would further help us to ﬁnd the exact nature of this star.

AX J1832.3–0840

AX J1832.3–0840 was detected in the X-ray images obtained using ASCA on October 11, 1997 as a part of the ‘Galactic plane survey’ (Sugizaki et al. 2000). The pulsations of 1549.1 0.4 s were detected from follow-up observations performed using ± ASCA on October 17 - 18, 1999 with a pulse fractional amplitude of 63%. The pulse pro-

ﬁle of AX J1832.3–0840 was characterized by a double-peak, independent of the energy bands. The X-ray spectrum of AX J1832.3–0840 could be ﬁtted well with a powerlaw

22 2 model of Γ = 0.76 and N = 1.27 10 cm− . An iron emission line at 6.7 keV was also H × detected in the X-ray spectrum with an equivalent width of 450 eV. The X-ray ﬂux in AX

11 2 1 J1832.3–0840 was estimated to be 1.1 10− erg cm− s− . The X-ray spectrum of this × source could also be ﬁtted well with a thermal-equilibrium plasma model (Raymond and Smith, 1977) with interstellar absorption and a partial covering fraction absorption with the following parameters : plasma temperature = 10.6 keV, a partial covering absorption

22 2 column density = 10.6 10 cm− and a partial covering absorption fraction = 0.65. No × ﬁrm identiﬁcation of the X-ray source AX J1832.3–0840 could be made from its ASCA 3.10 Conclusions 111 observations (Sugizaki et al. 2000).

During our XMM-Newton observations, we detected AX J1832.3–0840 with a pulse period of 1552.33 1.6 and with a pulse fractional amplitude of 67%. The pulse ± proﬁle of AX J1832.3–0840 was characterized by a double-peak proﬁle (shown in Figure

3.1), similar to the previous ASCA observations. Using the previous and present spin

9 1 period measurements, we calculated the spin period derivative, P˙ to be 3.1 10− ss− . × The X-ray spectrum of AX J1832.3–0840 obtained using the XMM-Newton observatory was best ﬁtted with a bremsstrahlung model of temperature, kT = 30 keV, interstellar

22 2 22 2 absorption, N = 0.81 10 cm− , partial absorption of 6.19 10 cm− , and a partial H × × covering fraction of 6.8. We also detected three iron emission lines at 6.4, 6.7 and 6.9 keV corresponding to ﬂuorescent iron, He-like and H-like Fe with equivalent widths of 60, 140 and 190 eV respectively. ‘NIR method 1’ suggests the NIR counterpart to be a low mass star at a distance of 1 kpc and ‘NIR method 2’ suggest the temperature of the ∼ NIR counterpart to be less than 20,000 K. The X-ray spectrum of AX J1832.3–0840 characterized with Fe emission at 6.4, 6.7 and 6.9 keV is typical of an IP. AX J1832.3–0840 is spinning down with a spin down

9 1 rate, P˙ = 3.1 10− ss− , which has also been seen in IPs. The NIR observations suggest × the NIR counterpart to be a low mass star at a distance of 1 kpc. With these arguments, ∼ we classify AX J1832.3–0840 to be an IP.

3.10 Conclusions

In this Chapter, we have investigated the nature of seven X-ray pulsators to ﬁnd their nature with the help of X-ray and NIR observations. The main conclusions of this Chapter are as follows :

1. We detected the NIR counterparts of 7 X-ray pulsators with the help of position

accuracy provided by their Chandra observations. 112 3.ThenatureofsevenfaintX-raypulsators

2. With the help of X-ray spectral and timing analysis, along with the NIR photometric

analysis, we have found that out of seven X-ray pulsators, three pulsators SAX J1324.4–6200, AX J1749.1–2847 and AX J1820.5–1434 are most likely to be X-ray pulsars with high mass companion stars and another three AX J1700.1–4157, AX

J1740.1–2847 and AX J1832.3-0840 are most likely to be IPs and SAX J1452.8– 4959 could be an IP or a LMXB. Chapter 4

Multiwavelength study of the transient X-ray binary IGR J01583+6713

4.1 Introduction

We present, in this Chapter, the optical and X-ray observations of a transient X- ray source IGR J01583+6713 (Section 4.2) discovered with INTEGRAL (Steiner et al., 2005) during its X-ray outburst in December 2005 and the subsequent quiescent phase.

We investigated the nature of the IGR J01583+6713 with the help of multi-wavelength (optical and X-ray) observations (Section 4.3 and 4.4). It was one of the very few transient X-ray sources discovered by INTEGRAL in the northern sky, and suitable for optical observations with Sampurnanand Telescope (ST) at Aryabhatta Research Institute of Obser- vational Sciences (ARIES), Nainital. We carried out a photometric monitoring campaign of the source following its discovery. This was complemented with multiple spectroscopic observations with the Himalayan Chandran telescope (HCT) at Indian Astronomical Ob- servatory (IAO), Hanle. We have discussed the optical photometric variability, the X-ray

Research presented in this Chapter is partly published in Kaur et al. (2008c)

113 114 4.TransientX-raybinaryIGRJ01583+6713 variability and spectral classiﬁcation of the optical counterpart of IGR J01583+6713 in

Section 4.5. The results and conclusions form the last Section (4.6) in this Chapter.

4.2 IGR J01583+6713

The X-ray transient IGR J01583+6713 was discovered by the IBIS/ISGRI imager on board INTEGRAL during an observation of the Cas A region on December 6, 2005 (Steiner et al., 2005). The source was detected with a mean ﬂux of about 14 mCrab in the 20 - 40 keV band, while a null detection was reported in the higher energy band of 40-80 keV. In subsequent INTEGRAL observations of the same ﬁeld during December 8 -

10, 2005, the X-ray ﬂux was found to be decreasing on a timescale of days. The Swift observations of the source on December 13, 2005 identiﬁed IGR J01583+6713 to be a point

h m s object located at RA : 01 58 18 .2 and DEC: +67◦ 13′ 25′′. 9 (J2000) with an uncertainty of 3.5 arc seconds and its spectral analysis revealed that it is highly absorbed with NH

23 2 21 approximately 10 cm− as compared to the estimated Galactic value of NH = 4.7 x 10

2 cm− towards the same direction (Kennea et al., 2005). The optical counterpart of IGR J01583+6713 was proposed to be a Be star on the basis of detection of Hα (EW 70 Å) and weak Hβ (EW 6 Å) emission lines in its optical spectrum (Halpern and Tyagi, 2005).

Subsequently, based on low resolution single epoch optical spectroscopy on December 23, 2006, Masetti et al. (2006) classiﬁed the optical counterpart as an early spectral type (O8 III orO9 V) Galactic ( 6.4 kpc) star and ruled out both the possibility of a supergiant ∼ companion and of the source being a low mass X-ray binary and cataclysmic variable.

4.3 Optical Observations

We have described the broadband photometric and intermediate resolution spectroscopic optical observations of IGR J01583+6713 in the following Sections. 4.3 Optical Observations 115

4.3.1 UBVRI Photometry

We have carried out broadband Johnson UBV and Cousins RI CCD photometric follow- up observations on 13 nights from December 13, 2005 to March 06, 2006 using a 2k × 2k CCD camera of the 1.04-m Sampurnanand Telescope (ST) at ARIES, Nainital, India.

During our observations the seeing varied from about 1.2 to 2.4 arcsec. Table 4.1 lists the log of optical photometric observations of IGR J01583+6713 in UBVRI wavebands along with the number of frames taken and exposure time in the respective filters. Usually more than two exposures were taken in each filter with a typical exposure times of 300 s, 300 s, 300 s, 200 s and 150 s in the UBVRI wave-bands respectively. Bias frames were taken intermittently and flat-field exposures were made of the twilight sky. In addition, as part of other ongoing programs, we could secure observations of Landolt (1992) standard

fields in UBVRI wavebands on four nights (see Table 4.1) near zenith and the same field was also observed at about five different zenith distances for extinction measurements. Figure 4.1 shows the field chart of IGR J01583+6713 observed with the ST on December 13, 2006 in the B waveband. Although all the observations were taken with the entire CCD chip of 13′ 13′, we have shown only a 7′.5 7′.5 CCD frame in Figure × × 4.1 for a closer look at the field. For photometric comparison we selected six stars with similar brightness and these are marked as 1, 2, 3, 4, 5 and 6 in Figure 4.1 while ‘T’ denotes the X-ray transient IGR J01583+6713.

Photometric data reductions were performed using the standard routines in IRAF and DAOPHOT (refer to Chapter 2). The standard magnitudes of all six stars were found to be stable on all four nights and were treated as local standards. Their mean standard magnitudes are listed in Table 4.2. The mean magnitudes and colors for IGR

J01583+6713 are estimated as V = 14.43 0.03, U B = 0.09 0.04, B V = 1.22 0.07, ± − ± − ± V R = 0.99 0.03, V I = 1.88 0.05. The diﬀerence in the measured (B ,V ,R − ± − ± obs obs obs and Iobs) and standard (Bst, Vst,Rst and Ist) BVRI magnitudes of IGR J01583+6713 and comparison star 1 are plotted in Figure 4.2. The variation was observed to be consistent 116 4.TransientX-raybinaryIGRJ01583+6713

Table 4.1: Log of broadband optical photometric observations of the transient source and Landolt (1992) standard ﬁelds Object Dateof Filter Exposuretime Observation (seconds) IGR J01583+6713 13Dec2005 V/R 2 600/2 300 × × 17Dec2005 B/V/R/I 2 300/2 300/3 200/2 150 × × × × 18Dec2005 B/V/R/I 3 300/3 300/3 150/3 150 × × × × 19Dec2005 V/R/I 1 300/1 200/1 150 × × × 20Dec2005 B/V/R/I 1 300/2 300/1 300/3 150 × × × × 21Dec2005 R/I 1 150/2 150 × × 28Dec2005 B/V/R/I 1 300/2 300/2 200/3 150 × × × × 25Jan2006 B/R/I 1 300/2 200/1 150 × × × 26Jan2006 B/V/R/I 1 300/2 300/2 200/2 150 × × × × 24Feb2006 B/V/R 2 300/2 300/2 200 × × × 28Feb2006 V/R 1 300/1 200 × × 02Mar2006 U/B/V/R/I 2 300/2 300/2 300/2 200/2 150 × × × × × 06Mar2006 B/V/R 1 300/1 300/1 200 × × × 24Nov2006 U/B/V/R/I 2 300/2 300/2 300/2 200/2 150 × × × × × 13Dec2006 U/B/V/R/I 3 300/3 300/3 300/2 200/2 150 × × × × × Landolt standard ﬁeld SA104 25Jan2006 B/R/I 11 300/11 60/11 60 × × × SA101 02Mar2006 U/B/V/R/I 9 300/9 180/9 180/9 120/9 120 × × × × × SA92 24Nov2006 U/B/V/R/I 2 300/2 300/2 180/2 130/2 100 × × × × × SA95 24Nov2006 U/B/V/R/I 7 500/7 300/7 150/7 100/7 100 × × × × × SA98 13Dec2006 U/B/V/R/I 7 450/7 300/7 120/7 60/7 60 × × × × × RU149 13Dec2006 U/B/V/R/I 2 300/2 300/2 120/2 60/2 60 × × × × × 4.3 Optical Observations 117

1 2 ’

1 3 ’

1 4 ’

1 5 ’

1 6 ’

1 7 ’

+ 6 7 o 1 8 ’

s s s s s h m s s 5 0 4 0 3 0 2 0 1 0 1 5 8 0 0 5 0

Figure 4.1: Identiﬁcation chart of IGR J01583+6713 taken with the ST in B passband on December 13, 2006. The transient is marked as ‘T’ and comparison stars are marked as 1,2,3,4,5and6 118 4.TransientX-raybinaryIGRJ01583+6713 magnitudes of IGR BVRI ) st and I st , R st , V st ) and standard (B obs and I obs , R obs , V obs erence in the measured (B ﬀ 6713 and Comparison star 1 from JD 2453710 to JD 2453810. + Figure 4.2: The di J01583 4.3 Optical Observations 119

Table 4.2: BVRI magnitudes of comparison stars Starno. B V R I 1 15.55 14.51 13.91 13.37 2 17.57 15.45 14.17 13.00 3 16.54 15.38 14.68 14.02 4 16.25 14.92 14.11 13.36 5 16.13 14.73 13.90 13.20 6 17.71 15.52 14.20 12.99 with the typical photometric uncertainty with standard deviations of 0.01 mag for BVR and 0.02 mag for the I passband for comparison star 1. For IGR J01583+6713 an increase in ﬂux of about 0.05 mag is apparent around December 13, 2005 followed by a constant ﬂux level thereafter.

4.3.2 Spectroscopy

Spectroscopic observations of IGR J01583+6713 were made on eight nights during Au- gust 18, 2006 to October 28, 2006 using the Himalayan Faint Object Spectrograph and Camera (HFOSC) available with the 2-m Himalayan Chandra Telescope (HCT), located at Hanle, India. The log of observations with the respective exposure times is given in

Table 4.3. We have used a slit of dimensions 1′′. 92 11′ for the Hα line proﬁle study × and a slit of dimensions 15′′. 41 11′ for the calibration observations. The spectrum were × obtained with two diﬀerent gratings (Grism 7 & Grism 8) in the wavelength range of 3500

1 1 - 7000 Å and 5800 - 8350 Å at a spectral dispersion of 1.45 Å pixel− and 1.25 Å pixel− respectively. Data were reduced using the standard routines within IRAF as described in Chapter 2. The wavelength calibration was done using FeAr and FeNe lamp spectrum for the Grism 7 and Grism 8 spectrum respectively. We employed the IRAF task identify and typically around 18 emission lines of Fe, Ar and Ne were used to find a dispersion solution. A fifth order fit was used to achieve a typical rms uncertainty of about 0.1 Å. For flux calibration of the IGR J01583+6713 spectrum in both Grism 7 and Grism 8, the 120 4.TransientX-raybinaryIGRJ01583+6713

Table 4.3: Log of optical spectroscopic observations of the X-ray transient IGR J01583+6713 Dateof WavelengthRange Exposuretime Observation (Å) (seconds) 2006Aug18 3500-7000: 5200-9200 1 900/1 900 × × 2006Oct14 3500-7000: 5200-9200 1 900/1 900 × × 2006Oct15 3500-7000: 5200-9200 1 1200/1 900 × × 2006Oct16 5200-9200 1 900 × 2006Oct17 5200-9200 1 900 × 2006Oct18 3500-7000: 5200-9200 1 900/1 900 × × 2006Oct28 5200-9200 1 900 × 2006Oct29 5200-9200 1 600 × instrumental response curves were obtained using spectrophotometric standards (Hiltner

600, Feige 110) observed on the same night, and the star’s spectrum were brought to the relative flux scale. During our two months observations of IGR J01583+6713 with Grism 8 of HCT, Hα is detected with a FWHM of 11.3 Å. ∼ The combined flux calibrated spectrum of X-ray binary IGR J01583+6713, taken with Grism 7 and Grism 8 on October 15, 2006 is shown in Figure 4.3 over a wavelength range of 3800 - 9000 Å. The identified spectral features are marked in the spectrum. The blue region of Grism 7, in the wavelength region 3600 - 3800 Å has poor Signal- to-Noise ratio and is not shown in Figure 4.3. Along with strong Hα and Hβ, we also detected a few weak spectral features, mainly singly ionized Iron, neutral Helium and neutral Oxygen lines, in the IGR J01583+6713 continuum spectrum in the wavelength region 6100 - 7900 Å, shown in Figure 4.4. Hα line profile variations are shown in Figure 4.5 covering a time span of 2 months. No change in line-profile or equivalent width of ∼

Hα is seen during this period. 4.3 Optical Observations 121 e tion are also β and H α use Interstellar bands are marked as "DIB", telluric absorp ff is marked as "NaD". "P" represents Paschen lines. H 6713 taken on October 15, 2006. Grism 7 and Grism 8 spectrum ar 9000Å. Di + Figure 4.3: Flux calibrated optical spectrum of IGR J01583 bands are marked with a filled triangle and the Sodium doublet combined together to show it overa wavelength range of 3800 - marked. 122 4.TransientX-raybinaryIGRJ01583+6713 5, 2006 and October I and OI are marked. 6713 in the wavelength range of 6100 - 7900 Å taken on October 1 + Figure 4.4: Continuum spectrum of IGR J01583 16, 2006. A few weakly identified features like FeII, HeI, SiI 4.4 X-ray Observations 123

Figure 4.5: Hα line proﬁle variations of IGR J01583+6713

4.4 X-ray Observations

The X-ray observations of IGR J01583+6713 were carried out with the X-ray Telescope

(XRT) onboard the Swift satellite and with the Proportional Counter Array (PCA) onboard the RXTE satellite. The Swift observatory carried out observations of IGR J01583+6713 on Decem- ber 13, 2005 for 47 ks and on April 11, 2006 for 37 ks. Both the XRT observations had useful exposures of 8 ks. The standard data pipeline package (XRTPIPELINE v. 0.10.3) ∼ was used to produce screened event files. Only data acquired in the Photon counting (PC) mode were analyzed adopting the standard grade filtering (0 - 12 for PC) according to XRT nomenclature. X-ray events from within a circular region of radius 0.8 arcmin, centered at the X-ray transient, were extracted for timing and spectral analysis. Background data were extracted from a neighbouring source free circular region of the same radius as taken for the source. Source and background lightcurves were generated using the X-ray 124 4.TransientX-raybinaryIGRJ01583+6713 counts from the respective circular regions with the instrumental time resolution of 2.5 s. A final source lightcurve was produced by subtracting the background lightcurve. The observations were made for small segments and no variability beyond statistical variation is seen in the X-ray light curves between the segments. The final source spectrum were obtained by subtracting the background spectrum, and spectral analysis was done using the energy response of the detector for the same day. The RXTE pointed observations were performed on December 14, 2005 with a total effective exposure time of 3 ks. The RXTE-PCA standard 2 data with a time ∼ resolutionof 16 s were used to extract the spectrum of the source inthe energy range 3 - 20 keV. During this observation only two PCUs were operational. The background spectrum for this observation was generated using the task ‘pcabackest’. PCA background models for faint sources were used to generate the background spectrum. The spectral response matrix of the detector was created using the task ‘pcarsp’ and applied for spectral fitting. The spectrum was rebinned to have sufficient signal to noise (S/N) ratio in each bin of the spectrum. The source was also regularly monitored by the All Sky Monitor (ASM) on board the RXTE satellite. The ASM lightcurve is shown in Figure 4.6 from MJD 53660 to MJD 54160, including the outburst detected on MJD 53710. The epoch of detection of the hard X-ray transient IGR J01583+6713 is marked as ‘T’ in the Figure 4.6. The soft X-ray enhancement in the RXTE-ASM light curve is consistent with a 10 - 20 mCrab intensity. The optical and X-ray observations of IGR J01583+6713 taken for the present study are also marked in Figure 4.6. We have used XSPEC12 for X-ray spectral analysis. We fitted both a powerlaw model and a blackbody model with line of sight absorption to the source spectrum obtained from the Swift and the RXTE. The spectral parameters obtained for both the models for the three observations are given in Table 4.4. Both the powerlaw and blackbody models fit the Swift data well with a reduced χ2 in the range of 0.8 - 1.4, while the RXTE spectrum is not well fitted by the blackbody model. For the Swift spectrum the models 4.4 X-ray Observations 125 Swift INTEGRAL ved by the marked are optical photometric observations (by observations (by open circles with a dot inside) and the RXTE p-arrow), the 6713 from MJD 53660 to MJD 54160 including the outburst obser + during MJD 53710 to MJD 53720, marked as "T" in the figure. Also Swift filled triangle), optical spectroscopic observations (by u and the Figure 4.6: ASM lightcurve of IGR J01583 observations (by open star). 126 4.TransientX-raybinaryIGRJ01583+6713 0.60 1.06 0.22 0.10 13 13 / / ± ± ± ± powerlaw) blackbody) + + 9.60 2.58 6.23 0.45 0.56 2.05 0.49 1.34 19 1.1 19 1.2 / / ± ± ± ± observations. RXTE and the 0.22 1.74 0.44 1.81 14 0.8 14 0.8 / / ± ± Swift RXTE Swift 1.4 2.5 1.71 Spectral parameters of model (Absorption Spectral parameters of model (Absorption 6.0(with90%confidence) 21.96 1.5(with90%confidence) 15.16 < < ) 5.8E-12 19.6E-12 3.9E-12 1 ) 5.5e-12 29.7e-12 4.0e-12 − 1 ) 5.1E-12 10.8E-12 3.7E-12 s − 1 ) 5.0e-12 11.6e-12 3.1e-12 s − 2 1 s − − 2 s − 2 − 2 − Table 4.4: Spectral Parameters for the ) ) 2 2 − − dof dof / / (cm (cm 2 2 χ χ 22 22 10 10 × × H H Observed Flux 2-10 keV (erg cm Unabsorbed Flux 2-10 keV (erg cm Observed Flux 2 - 10 keV (erg cm Unabsorbed Flux 2 - 10 keV (erg cm Reduced Reduced ParameterN ParameterN 2005December14 2005December13 2006April05 2005December14 2005December13 2006April05 PhoIndex Blackbody temperature (keV) 1.61 4.5 Results and Discussions 127 are indistinguishable from spectral fit only. Figure 4.7 shows the powerlaw fit to the Swift observations made on December 13, 2005 (top panel), the RXTE observations made on December 14, 2005 (middle panel), and the Swift observations taken on April 05, 2006 (bottom panel).

The Swift spectrum for both the observations made on December 13, 2005 and April 05, 2006 have coarse energy binning, making the Fe emission line at 6.4 keV unde- tectable in the raw spectrum. To find an upper limit on the 6.4 keV emission line in the Swift spectrum taken on December 13, 2005, we fixed the line-center energy at 6.4 keV in the spectrum and fitted a Gaussian to the line. The upper limit on the equivalent width of the 6.4 keV emission line is determined to be 100 eV with 90% confidence limit. We searched for X-ray pulse periods using a pulse-folding technique. The background count rate was subtracted from the IGR J01583+6713 lightcurve and the barycenter correction was done. The time resolution of the instrument and the time span of the continuous data constrained the pulse period search to the range 5s - 800s. We found X-ray pulsations with a pulse period of 469.2 s and with a pulsed fractional amplitude of 22%. Figure 4.8 shows the light curve of IGR J01583+6713, observed with the Swift on

December 13, 2005, folded with the pulse period of 469.2 s. However the evidence of

4 pulsation detection is marginal with a false-alarm-probability of 10− .

4.5 Results and Discussions

4.5.1 Photometric Variability

There were 7 optical photometric observations made in 15 days from December 13, 2005 to December 28, 2005 and later on there were observations witha gapof 20-25daysfor the next 3 months. No variability of more than 2 sigma is found in any optical pass-bands but a larger variation and decreasing trend is seen for the ﬁrst few days of observations in all of them, as shown in Figure 4.2. The scatter in data points of IGR J01583+6713 is more than the comparisons stars by 2 sigma in all the pass-bands and some increase ∼ 128 4.TransientX-raybinaryIGRJ01583+6713

Figure 4.7: X-ray spectrum of IGR J01583+6713. Top - the Swift observations made on December 13, 2005, middle - RXTE observations on December 14, 2005, bottom - the Swift observations made on April 05, 2006. The points with error bars are the measured data points and the histograms are the respective best ﬁtted model spectrum consisting of absorbed powerlaw model components, convolved with the respective telescope/detector responses. 4.5 Results and Discussions 129 in magnitude of about 0.05 mag can be seen around MJD 53725. The decrease in reprocessed optical emission for few days of outburst does indicate that source ﬂux was decreasing during that period. However, we claim no strong variability in this source and we cannot ascertain whether the source was brighter in the optical band during the peak of its X-ray outburst.

4.5.2 Stability of the Hα emission line

The dynamical evolution of Be envelope can be studied with the help of changes in emission line proﬁle (Negueruela et al., 2001). Table 4.3 shows the spectroscopic observations of IGR J01583+6713 made over a period of 2 months. Strong Hα and Hβ emission line features are always found in the spectrum with equivalent width of -74.5 1.6 Å and -5.6 ± 0.3 Å respectively. This ratio is somewhat diﬀerent from that reported by Masetti et al. ± (2006), which may indicate structural changes in the circumstellar disk of Be star in this system.

The line equivalent width and line profile of the Hα line are found to be constant for the observations, listed in Table 4.3. The resolution of our instrument was not good enough to carry out a detailed study of the Hα line shape. The Hα line gives a poor fit with a Gaussian model, leaving the wings of the line unfitted. However, the Hα line, being very strong, of equivalent width -75 Å, confirms that it originated in the circumstellar disk.

4.5.3 X-ray variability and Spectrum

X-ray observations of IGR J01583+6713 were made on December 13, 2005 by the Swift after the detection of its outburst on December 6, 2005 by INTEGRAL. The pulsations at 469.2 seconds were detected with a pulsed fraction amplitude of 22%. Corresponding to 469.2 sec pulse period, we have estimated the X-ray binary IGR J01583+6713 orbital period in the range 216 - 561 days (Corbet, 1986) assuming the maximum eccentricity of the orbit to be 0.47 for Be binaries (Bildsten et al., 1997). 130 4.TransientX-raybinaryIGRJ01583+6713

Figure 4.8: Light curve of IGR J01583+6713 observed using the Swift observatory on December 13, 2005, folded modulo 469.2 s.

Figure 4.7 (top) and (bottom) show the powerlaw ﬁtting to the Swift observations made on December 13, 2005 and April 05, 2006. N is found to decrease from 22.0 H × 22 2 22 2 10 cm− to 2.6 10 cm− for the two Swift observations of IGR J01583+6713, using × the powerlaw model. Using the blackbody model, N is found to decrease from 15.2 H × 22 2 22 2 10 cm− to 0.5 10 cm− . Photon index and blackbody temperature are in the range × 1.7 - 2.0 and 1.3 - 1.8 keV respectively, for all the observations listed in Table 4.4. The

Swift observations clearly show that NH has decreased by about an order of magnitude in a span of four months. The observed flux in 2 - 10 keV band has also decreased by a factor of 4. Within measurement uncertainty, the powerlaw photon index (or the blackbody ∼ temperature in the blackbody emission model) is found to be unchanged. Lack of variability in the optical photometric measurements indicate an absence of changes in the distribution of circumstellar material around the companion star. How- ever, from the X-ray spectral measurements we have detected a significant change in the absorption column density. During the first Swift observations, when the source was

22 2 brighter, we measured an absorption column density of (22 10) 10 cm− and the ± × 4.5 Results and Discussions 131 spectrum shows a upper limit of 100 eV on equivalent width of an Fe K-ﬂuorescence line. If we assume an isotropic distribution of absorbing matter around the compact object, where the X-ray emission originates, we expect the Fe 6.4 keV spectral line to be detected with an equivalent width of 250 eV (Makishima, 1986), a factor of 2.5 more than the upper limit. The absence of a strong Fe emission line indicates that the X-ray absorbing material around the compact object has a non-isotropic distribution, probably related to its disk structure.

4.5.4 Spectral Classiﬁcation

A low dispersion (4 Å/pix) optical (3500-8700 Å) spectrum taken one day after the outburst was presented by Masetti et al. (2006). However, due to the absence of any photo- spheric absorption feature and poor signal, they could not secure a definite classification of the optical counterpart of IGR J01583+6713. Our spectrum was taken with a better spectral resolution ( 3 Å near H ) as compared to the Masetti et al. (2006) but the MK ∼ α classical region (< 5000 Å) was too weak in our spectrum to identify the spectral features. Figure 4.3 shows the combined flux calibrated spectrum of IGR J01583+6713 on October 15, 2006. The continuum normalized spectrum taken on October 15, 2006 and October

16, 2006 are shown in Figure 4.4. Most of the spectral features in the 3800 - 8800 Å region are identiﬁed and marked. We compare these features with a near infrared spectral library by Andrillat (1988 & 1990, henceforth AND90) of a sample of 70 emission line Be stars with known MK spectral type (B0 - B9) at resolution of 1.2 Å. The spectrum of the optical counterpart of IGR J01583+6713 is described as follows:

Hydrogen lines are seen in emission. Hα and Hβ show single peak while the Paschen lines (P12 - P20) have a double-peak structure with the red (R) peak greater than the blue (V) one. CaII (8498, 8542, 8662 Å) and NI (Multiplet at 8630 Å and 8680Å) are also in emission. In their sample, AND90 found that the HI and CaII emission features are strong in early type (< B5) stars and diminish strongly for later type stars. CaII triplet 132 4.TransientX-raybinaryIGRJ01583+6713

Figure 4.9: Bottom - Flux Normalized optical spectrum of IGR J01583+6713, Top - Flux calibrated optical spectrum of HD 164284 from Andrillat et al. (1988), a comparison. P12 to P20 are the Hydrogen Paschen lines from P12 to P20. FeII, OI, CaII and NI spectral features are also marked. emission was associated with a large IRAS excess. The present spectrum also show OI (8446Å and Triplet 7773Å ) in emission which is seen in most of the cases with stars having spectral type earlier than B2.5. The features of IGR J1583+6713 resemble most closely to HD 164284 (B2 IV-Ve) in Andrillat et al. (1988) and HD 41335 (B2 IVe) in Andrillat et al. (1990). For a comparison, HD 164284 spectrum is shown on the top of IGR J01583+6713 spectrum in Figure 4.9 over a wavelength range of 7500 - 8800Å.

FeII (6248, 6319, 6384, 6456, 6516, 7515 and 7711 Å) and SiII (6347 Å) lines appear in emission. He I lines at 6678 Å and 7281 Å are in absorption with asymmetric 4.5 Results and Discussions 133 proﬁles indicating the presence of an emission region. Based on these characteristics it is suggested that its spectral type lies around B1 to B3, however a later spectral type cannot be completely ruled out and a further high resolution spectrum in the optical and near infrared wavebands would be required to ascertain the true spectral type for the transient.

Furthermore, the weak Fe 6.4 keV line (EW 100 eV) in the X-ray spectrum and a high ≤ hydrogen column density may suggest the presence of a wind-powered accretion disc, hence leaving the possibility of a low-luminosity blue supergiant open. However, the strong Hydrogen lines suggest that it belongs to a main sequence (IV-V) luminosity class

(Leitherer, 1988). We employed the reddening free Q-parameter to further ascertain the spectral type of the star. Using the normal reddening slope, X[E(U-B)/E(B-V)]=0.72, the Q[=(U- B)-X(B-V)] parameter was found to be -0.63 0.06 and this corresponds to a spectral ± type of B2-3 for an early type main-sequence star (Johnson and Morgan, 1954). We derive a color excess E(B - V) = 1.46 0.05 mag by adopting B2 IV as a ± companion and taking the intrinsic color (B - V)0 = -0.24 mag (Schmidt-Kaler, 1982) and a mean observed color of 1.22 0.05 mag. An estimate of reddening at other wavelengths ± is made following Fitzpatrick (1999) and the intrinsic colors are found to be (U - B)0 =

-1.00, (V - R)0 = -0.15 and (V - I)0 = -0.35 quite consistent with intrinsic colors corresponding to a B2 IV star with (U - B)0 = -0.86 (Schmidt-Kaler, 1982), (V - R)0 = -0.10 and (V - I)0 = -0.29 (Wegner 1994). The above color excesses yield a value of visual extinction, A = 4.5 0.2 mag. Using the absolute magnitude from Lang (1992), and V ± from the relation m - M = 5logD - 5 + AV , the distance to the source is estimated to be 4.0 0.4 kpc placing the transient well beyond the Perseus arm of the Milky Way. ± 22 2 For A 4.5, the corresponding column density N is 0.81 10 cm− . V ∼ H ∼ × The X-ray observations made by Swift during the quiescent state of the transient IGR

22 2 J01583+6713 on April 05, 2006 gave N of the order of 0.5 10 cm− for blackbody H × 22 2 22 ﬁt and 3 10 cm− for the powerlaw ﬁt. The Galactic HI column density is 0.4 10 × × 134 4.TransientX-raybinaryIGRJ01583+6713

2 cm− in the direction of the transient. The NH obtained using the blackbody model is closer to the NH calculated using optical extinction measurement, and suggests that there is almost no contribution by material local to the source to the total NH. However, the X-ray spectral ﬁt parameters of this source are similar to the typical spectral parameters of the typical accretion powered X-ray pulsars and suggest it to be a member of it.

4.6 Conclusions

The nature of the hard X-ray transient IGR J01583+6713 and its optical counterpart has been investigated using new photometric and spectroscopic data in the optical as well as the X-ray band. The main conclusions of this Chapter are as follows.

1. No signiﬁcant variability in V-band is seen, however an upper limit of 0.05 mag is set over a time scale of 3 months since the X-ray outburst implies that the source was quite stable during our observations.

2. The spectral characteristics of the optical counterpart were found to be consistent with a B2 IV (classical Be) star showing strong emission lines of Hydrogen (single peak Balmer and double peak Paschen), ionized Calcium, ionized Silicon, Oxygen, Nitrogen and ionized Fe. Thus we classify this source to be a Be/X-ray binary

system. The source is located (l=129◦,b = 5◦) well beyond the Perseus arm of the Milky Way at a heliocentric distance of 4.0 0.4 kpc. ∼ ±

3. The column density in this source decresed by a factor of ten in the quiescent state suggests the formation of accretion disk during the X-ray outburst in this source.

4. The timing analysis of Swift observations of IGR J01583+6713 suggests it to be a possible X-ray pulsar with a pulse period of 469.2 second and an orbital period of 216 - 561 days. Part II

QPOs in X-ray binaries

135

Chapter 5

Discovery of Quasi Periodic Oscillations in XTE J0111.2–7317

5.1 Introduction

Quasi-periodic oscillations (QPOs) have been discovered in a number of systems in low mass X-ray binaries (van der Klis, 2006a) and high mass X-ray binaries (Kaur et al., 2007b). These oscillations are assumed to arise due to inhomogeneities in the inner accretion disk in both these classes of X-ray binaries. As the inner accretion disk in most of the LMXBs extend close to the surface of the neutron star, these oscillations in LMXBs have a potential to probe the atmospheres close to the surface of the neutron star. However, the lack of measurement of magnetic ﬁeld and non detection of pulsations in most of these systems make them not ideal for present QPO models, for which high mass

X-ray binary (HMXB) pulsars suit the best. Among HMXB pulsars, QPOs seem to occur equally frequently in transient and persistent sources. Transient HMXB pulsars from which QPOs have been detected

Research presented in this Chapter is partly published in Kaur et al. (2007)

137 138 5.DiscoveryofQPOsinXTEJ0111.2–7317 are EXO 2030+375 (Angelini et al., 1989), A0535+262 (Finger et al., 1996), XTE

J1858+034 (Paul and Rao 1998; Mukherjee et al. 2006), V0332+53 (Takeshima et al. 1994; Qu et al. 2005), and 4U 0115+63 (Soong and Swank, 1989), while the persistent HMXB pulsars in which intermittent QPO features have been detected are 4U 1907+09

(in ’t Zand et al. 1998; Mukerjee et al. 2001), SMC X-1 (Angelini et al., 1991), Cen X- 3 (Takeshima et al., 1991), LMC X-4 (Moon and Eikenberry 2001b; La Barbera et al. 2001), and X Per (Takeshima, 1997). In this Chapter, we have presented the discovery of QPOs in HMXB pulsar XTE

J0111.2–7317 (Section 5.2) during its X-ray outburst in November 1998, at a frequency 1.27 Hz which is the highest frequency among the QPOs discovered in HMXB pulsars till now. The X-ray observations were made using the RXTE during its X-ray outburst, described in Section 5.3. The discovery of QPOs in XTE J0111.2–7317 is discussed in

Section 5.4. We have also investigated the possibility of the present detection of QPOs from a nearby source SMC X–1 (Section 5.5). In the absence of the detection of a cyclotron feature, the QPOs helped us to estimate the strength of magnetic ﬁeld in this source, discussed in Section 6.5. The main conclusions of this chapter are summarized in

Section 5.7.

5.2 XTE J0111.2–7317

The transient X-ray pulsar XTE J0111.2–7317 was discovered with the Proportional

Counter Array (PCA) on the Rossi X-Ray Timing Explorer (RXTE) in November 1998 (Chakrabarty et al., 1998a) and was simultaneously detected in hard X-rays with the Burst and Transient Source Experiment (BATSE) on board the Compton Gamma Ray Observa- tory (CGRO) with a ﬂux ranging from 18 to 37 mcrab (Wilson and Finger, 1998). Public data from BATSE and the RXTE All Sky Monitor (ASM) revealed that this source was seen in outburst in both hard and soft X-rays from November 1998 to January 1999. Follow-up observations were taken by the Advanced Satellite for Cosmology and Astro- 5.3 RXTE observations 139 physics (ASCA) to study the pulsations and the X-ray spectrum up to 10 keV, detecting

10 1 2 it with a flux of 3.6 10− erg s− cm− in the 0.7 - 10.0 keV band (Chakrabarty et al. × 1998b; Yokogawa et al. 2000). The ASCA observations also revealed the presence of a pulsating soft excess, which subsequently led to detailed investigations of similar features in several accreting X-ray pulsars: Her X-1, LMC X-4, SMC X-1. The excess is now understood to occur due to reprocessing of the hard X-rays from the inner accretion disk (Hickox et al., 2004). ASCA observations during the outburst also provided the opportunity to locate the source within an error circle of 15′′ . The BATSE observations found that the pulsar is spinning up with a short timescale of 20 years (Yokogawa et al., 2000), thus confirming that the compact object as a neutron star. This object lies in the direction of the Small Magellanic Cloud (SMC), and it is very likely that it belongs to the SMC (Yokogawa et al., 2000), as confirmed (Coe et al., 1998) by the finding that the average

1 velocity shift of the optical lines is 166 15 km s− , which is comparable to the veloc- ± 1 ity shift 166 km s− for the SMC (Feast et al., 1961). The optical counterpart of XTE J0111.2–7317 was ﬁrst proposed to be a B star with strong Hα and Hβ emission (Israel et al., 1999) and later conﬁrmed as a B0.5 - B1 Ve star (Covino et al., 2001).

5.3 RXTE observations

XTE J0111.2–7317 went into outburst in November 1998 and was discovered with the RXTE-PCA during scans of the SMC X-1 region (Chakrabarty et al., 1998a). Subse- quently, two short PCA observations of the source were carried out on December 18, 1998 and later the source was monitored frequently from December 22, 1998 to February 19, 1999 as a Target of Opportunity. There were 20 pointed observations during this time, each with an exposure time of 2 - 3 ks. For most of the pointings, all ﬁve Proportional

Counter Units (PCUs) were on, although on some occasions, only three to four PCUs were on. The source was also regularly monitored by the RXTE-ASM. The nearby bright binary X-ray pulsar SMC X-1 is only 30′ away from XTE J0111.2–7317, and it was in 140 5.DiscoveryofQPOsinXTEJ0111.2–7317 of 300 days are shown here along with rescaled –7317. Two time ranges MJD 51165.56 to 51177.32 and MJD e time when the two sources XTE J0111.2–7317 and SMC X-1 were ents ‘A’ and ‘B’ are shown in Figure 5.2. -ASM Lightcurve of XTE J0111.2–7317 and SMC X-1 over a period RXTE -PCA lightcurve from observations made towards XTE J0111.2 51228.00 to 51228.09 marked withbright ‘A’ and respectively. The ‘B’ power correspond density to spectra for th the segm RXTE Figure 5.1: 5.4 QuasiPeriodicOscillationinXTEJ0111.2–7317 141 the PCA ﬁeld of view (FOV) during observations of XTE J0111.2–7317. Thus, we have also used the RXTE-ASM data on SMC X-1 available for the outburst period of XTE J0111.2–7317 to ascertain its contributions to the light curve and the power density spectrum. Long-term light curves of XTE J0111.2–7317 and SMC X-1 measured with the

RXTE-ASM are shown in Figure 5.1 for 300 days, covering the outburst of XTE J0111.2– 7317 and about ﬁve superorbital intensity modulations of SMC X-1. The PCA light curve of XTE J0111.2–7317 (contaminated by SMC X-1 in places) is also shown in the same Figure but with a diﬀerent normalization.

Light curves were extracted from the PCA observations with a time resolution of 0.125 s using the Standard-1 data. The background count rates were simulated and subtracted from these light curves. The 31 s pulsations of XTE J0111.2–7317 are clearly visible in the light curves, except during the last few days. The light curves were divided into small segments each of length 1024 s, and a power density spectrum of each segment was generated. The power spectra were normalized such that their integral gives the squared rms fractional variability, and the expected white-noise level was subtracted. Figure 5.2 shows two power spectra averaged over the time ranges ‘A’ and ‘B’ marked in

Figure 5.1, during which at least one of the two sources was bright. The peak at 0.032 Hz and its harmonics seen in the top spectrum in Figure 5.2 are due to the pulsations of XTE J0111.2–7317.

5.4 Quasi Periodic Oscillation in XTE J0111.2–7317

Along with the pulsations, the power density spectrum averaged over the time range ‘A’ of XTE J0111.2–7317 also shows a small hump at 1.27 Hz in Figure 5.2. This small hump in the power density spectrum is present due to QPOs. The lower spectrum in Figure 5.2

(representing time range ‘B’ of Figure 5.1) shows an absence of 31 s pulsations, while the 0.7 s pulsations of SMC X-1 and its harmonics are clearly detected. Figure 5.3 gives an expanded view of the XTE J0111.2–7317 power spectrum, in the frequency range 0.5 142 5.DiscoveryofQPOsinXTEJ0111.2–7317 ﬁgure has -PCA observations made towards XTE J0111.2– RXTE e ranges ‘A’ and ‘B’ respectively shown in Figure 5.1. The top urves obtained from 7317 are shown here. The top andbeen bottom multiplied spectra by are a for factor the of tim 500 for the sake of clarity. Figure 5.2: Power density spectra generated from the lightc 5.4 QuasiPeriodicOscillationinXTEJ0111.2–7317 143

Figure 5.3: Power density spectrum of XTE J0111.2–7317 generated from the lightcurve over the entire energy band of the PCA. The line represents the best ﬁtted model for the continuum and a Gaussian centered at the QPO frequency.

- 4.0 Hz. The solid curve is the best-ﬁt model, with one component for the continuum and a second Gaussian component for the QPO. From the individual power spectra, we

ﬁnd that the QPO signature was prominent during the time of outburst, from MJD 51165 to 51173, and faded as the outburst decayed. Inclusion of the Gaussian QPO feature in the model reduced the χ2 by 77 for 98 degrees of freedom. The QPO is detected with a signal-to-noise ratio of more than 9. The average QPO frequency was measured to be

1.266 0.018 Hz with an rms fraction of (2.52 0.15)%. The width of the QPO feature ± ± in the Gaussian model was measured to be, σ = 0.07 0.01 Hz, making it one of the ± narrowest such features among accretion-powered X-ray pulsars. 144 5.DiscoveryofQPOsinXTEJ0111.2–7317

5.5 Contamination by a nearby X-ray binary pulsar SMC X-1

Since SMC X-1 lay within the FOV of the RXTE PCA during the XTE J0111.2–7317 observations, the QPO seen in the power spectrum could also be a contribution from SMC X-1. Figure 5.1 shows the ASM light curves for SMC X-1 (ﬁlled circles) and XTE J0111.2–7317 (open circles) along with the scaled-down PCA light curve of XTE J0111.2–7317 (crosses) during the 1999 outburst. The SMC X-1 light curve clearly shows semi-periodic intensity variations on a timescale of about 60 days, which is supposed to be its super-orbital period. As can be seen in Figure 5.1, during the outburst of XTE J0111.2–7317 the ASM count rate of SMC X-1 was quite high. However, as shown in Figure 5.2, the PCA light curve of XTE J0111.2–7317 shows the 0.71 s pulsations due to SMC X-1’s contamination during the later part of the observations, when the ﬂux of XTE J0111.2–7317 had decayed (segment ‘B’ of Figure 5.1). But during the peak of the outburst of XTE J0111.2–7317, from MJD 51165 to 51173 (segment ‘A’ of Figure 5.1), the PCA light curve does not show the 0.71 s pulsations due to SMC X-1. It is also during this time that the QPO feature is most prominent in the power spectra. We have investigated the binary phase of SMC X-1 during the observations of XTE J0111.2– 7317 reported here and found that the observations made in segment ‘A’ were during the eclipse of SMC X-1. However, the segment ‘B’ observations were done when SMC X-1 was coming out of eclipse. Figure 5.4 shows the RXTE-PCA light curve of XTE J0111.2– 7317 during segment ‘A’, along with the 10 year RXTE-ASM light curve of SMC X-1, both folded with the orbital period of SMC X-1 (Paul et al., 2005) during the same time interval.

We have also used additional PCA data to estimate the possible level of contribution from SMC X-1 by using its pulsed X-rays. SMC X-1 was observed extensively by RXTE from November 24, 1996 to September 5, 1998 over a range of intensity levels. We used the event-mode data from the RXTE-PCA to obtain the light curve of SMC X-1 over 5.5 ContaminationbyanearbyX-raybinarypulsarSMCX-1 145

Figure 5.4: RXTE-ASM light curve of SMC X-1 folded with its orbital period along with RXTE-PCA light curve of XTE J0111.2–7317 during segment ‘A’ of Figure 5.1. this period with a time resolution of 25 ms. We first measured the local spin periods of the pulsar from barycenter-corrected light curves and then created pulse profiles by folding the light curves on the respective spin periods. The individual observations had short time spans of less than 3 ks, and therefore smearing of the pulse profile due to the orbital motion of the pulsar was negligible. The difference between the maximum and minimum count rates in the pulse profile was taken as a measure of the pulsed X-ray intensity for each observation. The average X-ray intensity was measured by fitting a constant to the folded light curve. A plot of the pulsed X-ray intensity versus the average X-ray intensity of SMC X-1 measured from the PCA observations is shown in Figure 5.5. It can be clearly seen that the pulsed X-ray intensity and the average X-ray intensity are very closely correlated, with a formal correlation coefficient of 0.97. Below an average source-

1 plus-background count rate of 42 counts s− per detector, pulsations are not detected in SMC X-1. As we did not detect the SMC X-1 pulsations along with the 1.27 Hz QPO (segment ‘A’ of Figure 5.1), we can separately conclude that the contribution of SMC X-1 146 5.DiscoveryofQPOsinXTEJ0111.2–7317 to the total ﬂux is negligible in segment ‘A’, and therefore the QPO must be a feature of

XTE J0111.2–7317.

5.6 Discussion

We have discovered QPOs from observations of the HMXB pulsar XTE J0111.2–7317 during the second peak and declining phase of its transient outburst in 1998 - 1999. The two peaks during the outburst can be clearly seen in the light curve taken by BATSE during the period MJD 51120 - 51200 (Yokogawa et al., 2000). However, RXTE-PCA observations of XTE J0111.2–7317 were made during the second peak of the outburst, from MJD 51165 to 51228. The 700 ms pulsations of SMC X-1 are detected in part of the light curve near the end of the outburst (segment ‘B’ of Figure 5.1), and the corresponding power spectrum is shown in Figure 5.2. We have found that the observations in segment ‘B’ were made while SMC X-1 was coming out of eclipse. We have ruled out the possibility that the QPOs observed during segment ‘A’ are from SMC X-1, which would have been equally interesting. The QPO frequencies in HMXB pulsars detected so far lie in the range of 1 mHz to 400 mHz. This is the highest frequency QPO ever detected from a HMXB pulsar.

Several models have been proposed to explain the QPO generation mechanism in accretion-powered X-ray pulsars, among which the Keplerian frequency model (KFM) and the beat-frequency model (BFM) are used most often, explained in detail in Chapter 1. In XTE J0111.2–7317, both the KFM and BFM models are applicable, and as the spin frequency (0.032 Hz) is much smaller than the QPO frequency (1.27 Hz), both would give similar values of the radius at which the 1.3 Hz QPOs are produced. Assuming a neutron star mass (Mn) of1.4M , the radius of the QPO production region is calculated to be ⊙

GM 1/3 r = n = 1.4 108cm. (5.1) QPO 2 2 4π νk ! × The ASCA observations of this source during the outburst measured a ﬂux of 5.6 Discussion 147

Figure 5.5: Relation between average count rate and pulsed count rate for SMC X-1 is shown here. The formal correlation coeﬃcient is determined to be 0.97.

10 2 1 3.6 10− erg cm− s− in the energy range 0.7 - 10.0 keV (Yokogawa et al., 2000), × and BATSE on CGRO measured a similar pulsed ﬂux in an energy band of 20 - 50 keV. Assuming a pulse fraction of about 50%, the total X-ray ﬂux of XTE J0111.2–7317 can be estimated to be about 4 times that found from the ASCA observations. As XTE J0111.2– 7317 belongs to the SMC, the distance uncertainty is relatively small compared with

Galactic X-ray binaries, and we assume a source distance of 65 kpc. Therefore, the total

38 1 X-ray luminosity of the source at 65 kpc is calculated to be about 7.3 10 erg s− . The × radius of the inner accretion disk around a magnetized neutron star with mass of 1.4 M ⊙ and a radius of 10 km, can be approximately expressed in terms of its magnetic moment

8 2/7 4/7 and X-ray luminosity as r = 3 10 L− µ (Frank et al., 2002), where, L is the X-ray M × 37 30 37 37 1 30 3 luminosity in the units of 10 erg s− and µ30 is the magnetic moment in units of 10 cm G. Assuming that the QPOs are produced at the inner accretion disk, i.e. equating r with r , the magnetic moment of the neutron star is calculated to be 2.2 1030 cm3 QPO M × 148 5.DiscoveryofQPOsinXTEJ0111.2–7317

G, which for a neutron star radius of 10 km, is equivalent to a magnetic field strength in the range of 2.2 to 4.4 1012 G depending on the magnetic latitude. The magnetic field × strength in this pulsar is quite comparable to most other HMXB pulsars. In the absence of a detected cyclotron absorption feature (Coe et al., 1998) in the X-ray spectrum of this source, the QPO frequency and the X-ray luminosity provide us the only way to estimate the magnetic field strength in this source. For a neutron star magnetic field of 1012 G, the energy of the cyclotron absorption feature on the stellar surface is 11.6 keV, so for this pulsar an absorption feature is expected to lie in the energy range 25 - 50 keV. More sensitive spectroscopic observations in the hard X-ray band during future outbursts of this source will be useful to detect a possible spectral feature due to cyclotron absorption.

5.7 Conclusions

We have discovered the X-ray quasi-periodic oscillations during an outburst of the transient HMXB pulsar XTE J0111.2–7317. The main conclusions of this Chapter are as follows :

1. We have discovered the highest frequency QPOs feature in HMXBs so far.

2. We have ruled out the possibilitythat the feature is associated with the nearby bright X-ray pulsar SMC X-1.

3. Using the X-ray luminosity and the measured QPO frequency and applying models in which the QPOs are produced because of the motion of inhomogeneous matter

in the inner accretion disk, we have estimated the magnetic ﬁeld strength of the neutron star, which is quite comparable to other pulsars of this class. Chapter 6

Long term evolution of QPOs in the X-ray pulsar 4U 1626–67

6.1 Introduction

In high magnetic ﬁeld X-ray pulsars, the QPO frequency is in the range of a few mHz to a few Hz (Kaur et al., 2007b). The QPOs are known to occur sporadically only in a few percent of the X-ray observations e.g., QPOs are detected in only 15% of the out-of- eclipse observations of Cen X-3 (Raichur and Paul, 2008). Our independent investigation of the RXTE-PCA light curves of several persistent sources show that the QPOs are quite rare. Exceptions to this are some of the transient sources, such as 3A 0535+262 (Finger et al., 1996) and XTE J1858+034 (Paul and Rao, 1998), which showed QPOs during most of the observations made during their outbursts. In this Chapter, we present a study of long term QPO frequency evolution in an almost persistent high magnetic ﬁeld X-ray pulsar 4U 1626–67 and discuss its relation to a possible recession of the accretion disk in this system.

Research presented in this Chapter is partly published in Kaur et al. (2008b)

149 150 6. QPOs in the X-ray pulsar 4U 1626–67

In order to study the QPO frequency evolution in 4U 1626–67, we have used the archived data from a number of X-ray observatories like XMM-Newton, RXTE, Bep- poSAX, Swift and EXOSAT, described in Section 6.3. The QPO frequency evolution in 4U 1626–67 is described in Section 6.4 and its possible implications to the recession of accretion disk is discussed in Section 6.5. In the last Section 6.6, we have summarized the conclusions of this Chapter

6.2 4U1626–67

The X-ray source 4U 1626–67 was discovered with the Uhuru satellite (Giacconi et al.,

1972) in 2 - 6 keV band. Pulsations, with a period of 7.68 s, were first discovered by Rap- paport et al. (1977) with SAS-3 observations and has been extensively monitored since then, especially with the BATSE detectors on board the Compton Gamma Ray Observa- tory (CGRO; Chakrabarty et al. 1997; Bildsten et al. 1997). Optical counterpart of the pulsar was identified as KZ TrA, a faint blue star (V 18.5 mag) with little or no redden- ≈ ing (McClintock et al. 1977; Bradt and McClintock 1983). Optical pulsations with 2% amplitude were detected at the same frequency as the X-ray pulsations (Ilovaisky et al., 1978) and are understood to be due to reprocessing of the pulsed X-ray flux by the accretion disk (Chester , 1979). A faint optical counterpart and the observed optical pulsed fraction requires the companion star to be of very small mass (McClintock et al. 1977, 1980). The X-ray light curve does not show any orbital modulation or eclipse. However, from the reprocessed pulsed optical emission and a close sideband in the power spectrum of optical light curve, an orbital period of 42 minutes was inferred (Middleditch et al.,

1981). Therefore, it falls under the category of ultracompact binaries ( Porb < 80 min), which have hydrogen-depleted secondaries to reach such short periods (Paczynski and

Sienkiewicz 1981; Nelson et al. 1986). Despite extensive searches, the orbital motion of this binary has never been detected in the X-ray pulse timing studies (Rappaport et al. 1977; Levine et al. 1988; Jain 6.2 4U 1626–67 151 et al. 2007). A very low mass secondary, in a nearly face-on orbit can possibly account for the lack of pulse arrival time delay. Recently Jain et al. (2007) have also proposed this source to be a candidate for a neutron star with a supernova fall-back accretion disk. From the extensive timing and spectral observations both in optical and X-ray bands, it has not yet been possible to establish the presence of a binary companion, and the upper limit of the companion mass has been determined to be very low. However, the presence of an accretion disk in 4U 1626–67 is beyond any doubt. Optical spectral and timing studies confirm that most of the optical emission is strongly dominated by the accretion disk (Grindlay 1978; McClintock et al. 1980). The X-ray spectrum also shows bright hydrogen-like and helium-like oxygen and neon emission lines with red- and blue-shifted components, a certain sign of accretion disk origin (Schulz et al. 2001; Krauss et al. 2007). Another direct evidence of an accretion disk in 4U 1626–67 is found from the detection of quasi-periodic oscillations, at a frequency of 40 mHz, from Ginga observations (Shinoda et al., 1990) and subsequently at a higher frequency of about 48 mHz from the BeppoSAX, Advanced Satellite for Cosmology and Astrophysics (ASCA), Rossi X-ray Timing Explorer (RXTE), and XMM-Newton observatories (Owens et al. 1997; Angelini et al. 1995; Kommers et al. 1998; Krauss et al. 2007). The QPOs have also been detected in reprocessed optical emission from both ground-based and Hubble Space Telescope observations (Chakrabarty 1998; 2001). For more than a decade since its discovery, 4U 1626–67 was found to be spinning up with a characteristic timescale P/P˙ 5000 years. It was found to be spinning down ≈ at about the same rate by BATSE on board CGRO in the beginning of 1991 (Chakrabarty et al., 1997). Even though the torque reversal was abrupt, the decrease in bolometric X-ray flux has been gradual and continuous over the past 30 year (Chakrabarty et al. 1997; ≈ Krauss et al. 2007). Recently, from a set of Chandra monitoring observations, Krauss et al. (2007) have established that the bolometric X-ray flux and various emission line fluxes have decreased continuously over the last few years, indicating a gradual depletion 152 6. QPOs in the X-ray pulsar 4U 1626–67 of the accretion disk. The X-ray flux and mass accretion rate are directly related, and these are likely to be related to the mass and extent of the material in the accretion disk. Therefore, the observed gradual decrease in X-ray flux indicates a depletion of material in the accretion disk of the pulsar. Another signature of this is seen by Krauss et al. (2007) as a change in the pulse profile of the pulsar as compared to the earlier observations.

6.3 Observations and data analysis

4U 1626–67 has been observed with various X-ray telescopes over diﬀerent epochs of time. Table 6.1 lists the log of observations of 4U 1626–67 that were found to be useful for the present study. Details of individual observations described below are in chronological order. Detection of QPOs at around 48 mHz have been mentioned from some of these observations, sometimes from a diﬀerent instrument also (Ginga [Shinoda et al. 1990], ASCA [Angelini et al. 1995], BeppoSAX [Owens et al. 1997], RXTE [Kommers et al. 1998; Chakrabarty 1998], and XMM-Newton [Krauss et al. 2007]). However, the QPO frequencies measured from these observations are often not reported with good enough accuracy to investigate a slow frequency evolution. For the present study, we have therefore reanalyzed the data and measured the QPO parameters with the highest possible accuracy. An EXOSAT medium energy (ME) proportional counter light curve of 4U 1626– 67 was obtained from HEASARC archive with the time resolution of 0.3125 s for an observations made on August 30, 1983 for 27 ks. The ME light curve of another observation made by EXOSAT on March 30, 1986 for 84 ks and reported earlier by Levine ≈ et al. (1988) is not available in the HEASARC Archive. ASCA observations of 4U 1626–67 were made on August 11, 1993 with the two

Gas Imaging Spectrometers (GIS2 and GIS3) and the two Solid State Imaging Spectrom- eters (SIS0 and SIS1) and light curves with total useful exposures of 40 and 25 ks were obtained for the GIS and SIS, respectively. During the ASCA observation, the GIS detec- 6.3 Observations and data analysis 153

Table 6.1: Log of Observations of 4U 1626–67

Telescope Year ObsIds No.of Obsspan Timeon Pointings (ks) source(ks) EXOSAT/ME 1983 128 1 27 27 ASCA/GIS 1993 40021000 1 72 40 ASCA/SIS 1993 40021000 1 70 25 BeppoSAX/MECS 1996 10017001 1 162 116 BeppoSAX/LECS 1996 10017001 1 128 35 RXTE/PCA 1996 P10101 9 395 147 RXTE/PCA 1996 P10144 1 13 10 RXTE/PCA 1997 P20146 14 33125 13 RXTE/PCA 1998 P30058 3 9167 40 RXTE/PCA 1998 P30060 10 2758 44 XMM-Newton/EPIC-pn 2001 0111070201 1 17 16 XMM-Newton/EPIC-pn 2003 0152620101 1 84 84 tors were operated in Pulse Height mode and SIS detectors were operated in Fast mode and the light curves were extracted from the unscreened high bit mode data with the minimum time resolution of 0.125 s for both GIS and SIS detectors. The light curves from the pairs of GIS and SIS instruments were added and a single power spectra is generated with the summed light curves. 4U 1626–67 was observed with BeppoSAX on August 9, 1996 for 116 ks by the three units of Medium Energy Concentrator Spectrometer (MECS) and for 35 ks by the Low Energy Concentrator Spectrometer (LECS). Light curves were extracted from all the instruments with 0.125 s. A single summed light curve was generated from three light curves of the MECS instruments to increase the signal-to-noise ratio.

RXTE PCA pointed observations of the source were made from February 1996 to August 1998. In 1996, the observations were made in the beginning of the year and at the end of the year under ObsIDs P10101 and P10144, respectively. The observations made under ObsID P10101 covers a time span of almost 5 days from MJD 50123 to 50128. 154 6. QPOs in the X-ray pulsar 4U 1626–67

There were nine observations in this ObsID each lasting for 4 - 8 hr. A single observation was made under ObsID P10144 for 5 hr on MJD 50445. In 1997, all the observations ≈ were made under ObsID P20146 and cover a time range of almost a year, from MJD 50412 to MJD 50795, but individual observations were made only for a few minutes. In

1998, RXTE PCA made observations under two ObsIDs, P30058 and P30060. There were three observations made under ObsID P30058, out of which two observations were made on MJD 50926 and the third observation was made on MJD 51032. In ObsID P30060, there were 10 short observations of about an hour each. For almost all the observations of RXTE, all ﬁve PCUs were on. Light curves were extracted from observations of 4U 1626–67 with a time resolution of 0.125 s using the Standard-1 data, which cover the entire 2 - 60 keV energy range of the PCA detectors. We divided all the RXTE PCA observations from 1996 to 1998 into three segments, from MJD 50123 to 50128, 50412 to 50795, and 50926 to 51032. The signal-to-noise ratio of the power spectra generated from the individual observations made between MJD 50412 and MJD 50795 was too poor to detect QPO except on MJD 50445, so a single power spectrum was produced by combining power spectra of all observations made between MJD 50412 and MJD 50795.

XMM-Newton has observed 4U 1626–67 four times, but a signiﬁcant amount of science data was present only in two of these observations, made under ObsIDs 0111070201 and 0152620101, listed in Table 6.1. We have analyzed data only from the pn detector of European Photon Imaging Camera (EPIC) on board XMM-Newton, which operates in the energy band of 0.15 - 15 keV. Light curves were extracted with a time resolution of 0.125 s for both the observations. All the light curves were divided into small segments, each of length 1024 s, and a power density spectrum of each segment was generated. The power spectra were normalized such that their integral gives the standard rms fractional variability, and the expected white noise was subtracted. A ﬁnal power spectrum was generated with the average of all the power spectra generated for each of the observations listed in Table 6.3 Observations and data analysis 155

Figure 6.1: Power density spectrum generated from the lightcurve obtained from the EX- OSAT observation made on August 30, 1983.

6.1. Flares with duration of 1000 s are clearly seen in the EXOSAT data, as mentioned by Levine et al. (1988). However, these flares are not detected in rest of the data mentioned in Table 6.1. A QPO at a frequency of 48 mHz is clearly seen in the power spectra of all the data sets except that from EXOSAT observations, during which it is detected at 36 mHz. Figure 6.1 shows the QPO detection from the EXOSAT observations made on August 30, 1983 in the range of 15 - 100 mHz. A Gaussian model is fitted to the QPO feature to determine its central frequency and width (FWHM of Gaussian) for all the data sets. The continuum of the power spectrum in the band of 20 - 80 mHz is fitted with a constant or a linear model. The uncertainty of the Gaussian model peak at 1 σ confidence interval is quoted as an error on the Gaussian center.

The QPO feature detected in the power spectrum of EXOSAT data is quite narrow 2 mHz as compared to the QPOs seen in rest of the data with a widthof4-5mHz. Figure 6.2 shows power spectra in the frequency range 26 - 72 mHz for the observations listed in Table 6.1 except the EXOSAT observations. Different constant numbers were added to 156 6. QPOs in the X-ray pulsar 4U 1626–67 each plot for clarity. A best-fitted Gaussian model for the QPOs and a constant model or a linear model for the continuum is shown on each plot with a solid line. A dotted vertical line at the best-fitted Gaussian center to the ASCA 1993 data is plotted in the same figure. A shift of 2 mHz is clearly seen from bottom to the top plot shown in Figure 6.2.

6.4 QPO evolution in 4U 1626–67

The evolution of the QPO central frequency as observed by various X-ray telescopes in both spin-up and spin-down eras is shown in Figure 6.3. An error bar plotted on each point in Figure 6.3 represents 1 σ error estimates. We could not ﬁnd Ginga observations of 4U 1626–67 made in July 1988 from archival data; thus the central frequency of QPOs and error estimate on it is taken from Shinoda et al. (1990) and is also shown in Figure 6.3. To conﬁrm the consistency of QPO frequency for each data set listed in Table 6.1, the QPO frequencies were measured from smaller segments of the data, 10 each for the 1996

RXTE observation and the 2004 XMM-Newton observation. The values determined from smaller segments have larger uncertainties, but within the uncertainties these values are consistent with the QPO frequency measured using the complete data sets in each case. It can be clearly seen in Figure 6.3 that the QPO central frequency has increased from

1983 to 1993 and that it gradually decreased from 1993 to 2004. However, the lack of observations does not allow us to deﬁne an exact time when the QPO frequency evolution changed from an increasing trend to a decreasing trend. The observations from 1993 to

2004 showed a frequency decrease of 2.3 mHz, while the error bars on all the data points during this era are within 0.4 mHz, except the ASCA 1993 data point, for which the error bar is 0.6 mHz, confirming the real decrease in QPO frequency with time. The QPO frequency derivative during spin-down era is (0.2 0.05) mHz/yr. A linear fit is shown ∼ ± on the data points with a solid line in the spin-down era in Figure 6.3. The reduced χ2 of the linear fit is 1.07 for 5 degrees of freedom. To further confirm the linearity, a constant model is also fitted to the data from 1993 to 2004. The reduced χ2 for a constant model is 6.5 Discussion 157

3.22 for 6 degrees of freedom, indicates poor ﬁt as compared to the linear ﬁt.

6.5 Discussion

The light curves of 4U 1626–67 taken with various observatories over a period of more than 20 years, have shown QPOs in every single observation of sufficient length. This is the first accretion-powered pulsar for which the QPO study has been made over a long timescale. In this regard, 4U 1626–67 is unique among persistent high magnetic field accreting X-ray pulsars. It shows that the accretion disk of the pulsar is quite stable, holding this feature for years. However, in a few cases, the observation duration was not long enough to make accurate measurements of the QPO parameters. QPOs in accretion-powered X-ray sources are widely believed to arise due to inhomogeneities near the inner accretion disk. The QPO frequency is the Keplerian frequency at the inner disk radius and is therefore positively related to the mass accretion rate or the X-ray luminosity. If the compact object is a neutron star, the inner disk is coupled with the central object through the magnetic field lines and QPOs corresponding to the beat frequency between the spin frequency and the Keplerian frequency of the inner disk can also be seen. In accretion-powered high magnetic field X-ray pulsars, the two different QPOs are never seen to occur in the same source. In some of the sources, such as 4U 1626–67, the QPO frequency is lower than the spin frequency, and therefore the QPOs can only be explained by the beat frequency model (refer Chapter 1).

If the QPOs are as per Keplerian frequency model (νk = νqpo, where νqpo is QPO frequency of the pulsar), then we expect ν L3/7 or ν L3/7. The ﬂux of 4U 1626–67 k ∝ 37 qpo ∝ 37 has decreased from 0.32 to 0.15 units from 1993 to 2004 (Krauss et al., 2007), implies that the change in QPO frequency is expected to be 27% from 1993 to 2004. The ∼ present QPO observations have shown only 4 % decrease in QPO frequency during the same time. However, Keplerian frequency model is not valid in this source. In the beat frequency model, (νk = νqpo + νs, where νs is pulsar spin frequency), the inner disk fre- 158 6. QPOs in the X-ray pulsar 4U 1626–67

Figure 6.2: All power density spectrum are generated from the lightcurves obtained from observations listed in Table 6.1. in chronological order. Diﬀerent constant numbers were added to each plot for clarity. The year of observations is written along with each PDS. A vertical line is drawn at 49.77 mHz, QPO frequency of ASCA 1993 observations, to clearly see the decrease in QPO frequency with time. 6.5 Discussion 159

Figure 6.3: QPO frequency evolution history of 4U 1626–67 from 1983to 2004. Thesolid line is a linear fit to the data from 1993 to 2004. Error bars represent the 1σ confidence intervals. quency is higher as compared to Keplerian frequency model, and the relative change in QPO frequency is expected to be even larger. Therefore, we see that the evolution of QPO frequency and the decrease of X-ray flux cannot be explained in the standard QPO generation mechanism and usual relation between inner disk and X-ray luminosity. We can consider two possibilities : One is that the QPOs are not generated from the inner disk, these are generated due to reprocessing in some outer structure of the disk. This is not very likely due to the large (upto 15%) rms in the QPO feature. Second possibility is that the observed X-ray flux change is not due to change of mass accretion rate by the same factor. Many X-ray sources show X-ray flux variation at long time scale upto a few months due to obstruction provided by complex accretion disk mechanism. The earlier study by Chakrabarty et al. (1997) has concluded that there was an abrupt torque reversal in 1990 and the system moved from spin-up to spin-down era with a characteristic time scale P/P˙ of 5000 yr. The two QPO detections with EXOSAT (35 ∼ mHz in 1983) and GINGA (40 mHz in 1988) are during the spin-up era of this pulsar, 160 6. QPOs in the X-ray pulsar 4U 1626–67 with increasing trend while the observations from 1993 to 2004, in the spin-down era, showed a slow decreasing trend in QPO frequency with time, somewhat coincident with the torque reversal in this source, shown in Figure 6.3. QPO frequency is found to be decreasing in the spin-down era with a frequency derivative of (0.2 0.05) mHz/yr. ∼ ± The X-ray spectral and flux evolution study along with pulse profile changes of 4U 1626– 67 by Krauss et al. (2007) have concluded that the accretion disk in this source is depleting with a time scale of 30 - 70 years. Krauss et al. (2007) has also estimated the long term

11 1 average accretion rate to be 3 10− M yr− for a distance 3kpc. However, a gradual × ⊙ ≥ change in mass accretion rate can not explain the unique torque reversal phenomena of this source (Li et al., 1980).

6.6 Conclusions

In this Chapter, we have studied the QPO frequency evolution in a high magnetic ﬁeld

X-ray pulsar 4U 1626–67 and the main conclusions of this Chapter are as follows :

1. We have detected very persistent quasi-periodic oscillations in the unique accretion- powered X-ray pulsar 4U 1626–67.

2. Using data from several observatories, we have detected a gradual evolution of the

oscillation frequency over a period of 22 years.

3. The frequency evolution indicates a possible recession of the accretion disk of the pulsar during the present spin-down era. Chapter 7

Summary and future prospects

In this thesis, we have presented a multi-wavelength analysis of ten X-ray sources, most of which are high magnetic ﬁeld neutron star X-ray binaries and a few are probable white dwarf X-ray binaries. To study these sources, we have performed X-ray, optical and near- infrared observations with both ground based and space based telescopes. This thesis contains seven Chapters. In Chapter 1, we have given a brief introduction about the binary systems, various mass transfer processes in them and the common observational properties of neutron star X-ray binaries. In Chapter 2, we have described all the ground based and space based telescopes used for data acquisition. In addition, this Chapter also contains data reduction and analysis procedures. The rest of the thesis is divided into 2 parts - Part I (Chapters 3 & 4) - ‘True nature of X-ray binaries’ and Part II (Chapters 5 & 6) - ‘QPOs in X-ray binaries’

Part I : True nature of X-ray binaries Majority of the high mass X-ray binaries are Be/X-ray binaries where a neutron star revolves around a Be star in a wide and eccentric orbit (e 0.2 - 0.9). These systems ∼ are mostly transient in nature and display X-ray outbursts, which are assumed to arise due to the passage of the compact object through the Be-star disk near periastron (called

161 162 7. Summary and future prospects

Type-I outburst) or because of sudden mass ejections from the Be star (Type-II outburst).

Recently, over last 10 years, a new class of Be/X-ray binaries has been discovered known as ‘persistent Be/X-ray binaries’ which have a potential to explore a diﬀerent kind of supernova explosion accompanied with a small kick or without a kick to the neutron star.

Till now, only four persistent Be/X-ray pulsars are known in our Galaxy and in Chapter 3, we have investigated the nature of seven X-ray pulsators which are probable candidates of this class, based on their X-ray properties. The most important step towards confirmation of the Be nature of their companion stars is the identification of their optical/near-infrared counterparts and is investigated in this Chapter with the help of their Chandra observations which provided their position with the sub arcsecond position accuracy. The X-ray timing and spectral analysis along with the near-infrared observations of these pulsators further helped us to constrain the nature of these systems. Our present observations have suggested that out of seven selected X-ray pulsators, three are most likely to be neutron star X-ray pulsars with the high mass companions. The further near-infrared spectroscopic observations of these sources if confirm the Be nature of the companion stars of X-ray pulsators, then the number of known persistent Be/X-ray pulsars would double and would heavily contribute to understand the evolution of high mass stars. IGR J01583+6713 was one of the very few transient X-ray binaries discovered by INTEGRAL in the northern hemisphere and suitable for optical observations from ST, Nainital. The detailed multi-wavelength study of this source formed Chapter 4 in this thesis. The X-ray observations of this source showed a clear signature of formation of accretion disk during its X-ray outburst, however, no signatures of reprocessing were seen in the optical waveband. The optical spectroscopic observations of this source suggested it to be a stable system during our observations. With the help of multi-wavelength observations of IGR J01583+6713, we classified this source to be a Be/X-ray binary or a possible Be/X-ray binary pulsar with a pulse period of 469 s. 163

Part II : QPOs in X-ray binaries QPOs have been rarely detected in transient high mass X-ray binary pulsars and XTE J0111.2–7317 is the sixth transient X-ray pulsar to show such a feature during its X- ray outburst, described in detail in Chapter 5. We have discovered QPOs at a frequency of

1.27 Hz in XTE J0111.2–7317, which is the highest frequency QPOs discovered among the high mass X-ray binary pulsars and suggested the presence of accretion disk in this source at a distance of 1400 km from the surface of the neutron star during its X-ray outburst. The QPO detection in this source also helped us to measure the magnetic field of this pulsar. In Chapter 6, we investigated the accretion disk properties through QPO evolution in an almost persistent high magnetic field X-ray pulsar 4U 1626–67. This is the only X-ray pulsar till now which have shown the QPOs signatures for more than 22 years and have suggested the presence of an almost stable accretion disk. A decrease in X-ray flux for more than 22 years and the decrease in the QPO frequency since its torque reversal in 1991 indicated a slow recession of accretion disk in this source. Our present QPO evolution study clearly indicated that the mass accretion rate and the X-ray flux of this source are not directly related to each other by a linear relation in this source and need a more sophisticated relation between them to explain the present QPO evolution in 4U 1626–67. 164 7. Summary and future prospects

Future Scope We wish to take up the following research projects in future to probe the X-ray binaries.

Faint, slow and persistent Be/X-ray pulsators : to understand the evolution of high mass stars

We have been working on faint, slow and persistent X-ray pulsars during my PhD and we have roughly identified the nature of a set of seven candidate pulsars, however, the exact nature of their near infrared counterparts can only be made with the help of their near infrared spectroscopic observations. Our proposal for this purpose has recently been accepted by 8.2-m ESO-very large telescope(VLT) for 26 hours which will be ∼ carried out in next 5-6 months. We would like to strongly confirm this class of pulsars observationally first and then would like to take theoretical challenges to explain their formation.

Thermonuclear Type-I bursts - to understand the accretion disk properties Type-I X-ray bursts are thermonuclear explosions on the surfaces of the accreting neutron stars, triggered by unstable hydrogen or helium burning. They typically have fast rise times of 1 s, exponential-like decays lasting 10 - 100 s, and recurrence times of a few hours to days. These bursts have been seen at longer wavelengths (mainly optical) with a delay of a few seconds which is understood to be due to reprocessing of the X-rays by accretion disk of the neutron star in them. Thus the multi-wavelength observations of these bursts has capability to probe the accretion disk and to some extend the surface of the neutron star also. Till now, only a very few bursts have been seen in X-ray and in longer wavelengths (optical/infrared) simultaneously to probe the accretion disk properties. The main reason behind it is the uncertainty in the occurrence and the short duration of the Type- I bursts because of which it becomes diﬃcult to plan the simultaneous observations in the X-rays and the optical/near-infrared wavebands. The XMM-Newton observatory 165 has the X-ray and the optical instrument onboard to make the observations of the astronomical sources in both these wavebands simultaneously. The archived XMM-Newton observations of the low mass X-ray binaries can be used to study the correlation of X-ray bursts in the X-ray and the optical wavebands which would help us to probe the neutron star interiors and accretion disk properties in these systems.

Quasi-periodic Oscillations in high magnetic field pulsars - understand the evolution of accretion disk QPO evolution study in high magnetic field pulsars is a promising window to understand the accretion disk evolution (Kaur et al. 2008b [ApJ, 676, 1184]). QPOs are considered to be the direct evidence of presence of accretion disk and change in QPO frequency is directly related to the recession of the accretion disk. Recently, we made a QPO evolution study of 4U 1626-67 with the help of archived X-ray observations and found a clear decrease in QPO frequency with the decrease in the X-ray flux during its spin-down era which further confirms the recession of the accretion disk (Kaur et al. 2008b [ApJ, 676, 1184]). The pulsed X-ray emission and the X-ray QPO of 4U 1626-67 are clearly detected in the optical waveband due to reprocessing of the X-rays from the accretion disk (Ilovaisky, Motch, & Chevalier 1978 [A&A, 70, 19], Chakrabarty et al.

2001 [ApJ, 562, 985]). To further study the evolution of accretion disk and the relation between mass accretion rate and the ﬂux, we would like to take up fast photometric optical observations of 4U 1626-67 with a ground based telescopes. 166 7. Summary and future prospects Appendix A

Magnetic Cataclymic Variables

A magnetic cataclysmic variable star (CV) is a semi-detached binary system comprising a magnetic accreting white dwarf and its mass donating companion star. The latter, is usually, but not always, a late-type star near or on the main sequence. The binary orbital periods for such systems are mostly lies between 80 min and several hours ∼ (Ritter 1990 and Ritter & Kolb 1995). Only in a few CVs, the companion is a giant star or another degenerate dwarf. These source are weak X-ray emitters with the maximum

33 1 X-ray luminosity, LX = 10 erg s− and most of them are detected within 1 kpc distance from Earth. Magnetic CVs are subdivided into three main classes : Polars (or AM Her systems), Intermediate Polars and DQ Herculis systems (which usually are a subset of intermediate polars) depending on the geometry of the accretion ﬂow (driven by B) and coherent pulsation time scales as well (Copper 1990; Warner 1995; Cordova 1995). The three sub-divisions of CVs are described below:

Polars • These systems are strongly aﬀected by the magnetic ﬁeld as the magnetospheric radius, rm in most of these sources are larger than the separation between the binary stars (a). Polars are distinguished by the presence of strong linear and circular optical polarization varying on the orbital period (Schmidt and Stockman, 2001). Modeling of this

167 168 A. Magnetic Cataclymic Variables and the detection of the Zeeman features in the states of low accretion rate, show that the white dwarfs have magnetic fields 1-6 107 G and thus the magnetic moment, µ ∼ × ≥ 1034 G cm3. In eclipsing systems, it can be ascertained that the spin and orbital periods differ by less than a few parts per million, and indeed all radiation, from infrared to X-rays show the same period. This locking of the spin and orbital rotations is probably a direct consequence of the inequality, rm > a, which is unique in these systems. These systems also satisfies the condition, r R , so it is not surprising that there is no evidence of m ≥ circ accretion disc in them. Figure A.1 shows a schematic diagram of a Polar, with a closer view at the pole of the magnetic white dwarf where most of the high energy emission mechanisms are taking place. In these systems, the matter from donor star falls on the white dwarf in the form of the gas stream whose ram pressure is much greater than the magnetic pressure over much of the binary separation (a). As a result of that the accretion stream moves freely under gravity until a distance 1010 cm from the white dwarf. At this distance, magnetic ∼ field takes over and the matter follows the magnetic field lines down to the surface of the white dwarf (magnetic pole). Until impact on the white dwarf surface, material essentially free falls, thus reaching substantial velocities which are seen in the optical spectra. The impact of matter on the white dwarf surface generates a shock wave, which becomes the source of hard X-rays in these sources. Hard X-rays emitted in the direction of the white dwarf from the shock wave above its surface heat the local area around the pole sufficiently for the pole to become a source of intense soft X-rays. These sources are generally much stronger sources of soft X-rays than hard X-rays which is understood to be due to the energy liberation deep within the atmosphere of the white dwarf due to clumps in the accretion flow.

X-ray spectrum

The ﬂux distribution from the accretion stream in polars comprises three components : hard X-ray bremsstrahlung (free-free, typically kT 30 keV) emission by free ∼ 169

Figure A.1: The schematic of the magnetic cataclysmic variable having B 107 G (polar). ∼ (Credit : http://heasarc.gsfc.nasa.gov/images). 170 A. Magnetic Cataclymic Variables electrons in a post-shock region where the ﬁeld lines impact the white dwarf atmosphere which is optically thin to the hard X-rays; cyclotron emission by semi-relativistic electrons spiraling around the magnetic ﬁeld lines (shock region is optically thick for low harmonics but thin for higher harmonics) and approximately the black-body spectrum in the UV or soft X-rays with kT 40 eV (Ezuka and Ishida, 1999). ∼

X-ray timing X-ray lightcurves show strong amplitude modulation to the spin-orbital period (from 80 min to 300 min). Pulse shapes vary from single peak proﬁle to more com- ∼ ∼ plicated shapes such as double asymmetric peaked proﬁle with a dip at phase 0.4. Soft

X-ray pulse proﬁles change substantially from one epoch to another and sometimes become anti-phased with respect to the earlier epochs (Heise et al. 1985, Sambruna et al. 1992). The latter case is probably due to a two poles accretion conﬁguration, one pole of which dominates the soft X-rays and the other always producing both hard and soft

X-rays. Despite the complexity showed by polars in the soft X-ray band, there is no dependence of amplitude modulation on energy (within 2 - 10 keV). Flickering on time-scales of seconds to minutes is commonly observed in soft and hard X-rays from polars which varies around the orbit in a way associated with the emission from accretion zones. Quasi-periodic optical and X-ray luminosity variations are also commonly seen in polars, in the range of 10 s to a few minutes, possibly arising in diﬀerent regions of the accretion ﬂows.

Intermediate Polars •

Intermediate polars (IPs) have got weak magnetic ﬁeld (B 106 G) as compared ∼ to the Polars. The weak magnetic ﬁeld in IPs cannot force the synchronous rotation as seen in Polars, thus these objects show a number of coherent pulsations in their lightcurve. Most IPs have spin periods of the order of 10 -20 min and there is a strong tendency to ∼ have spin period, P 0.1P , where P is the orbital period of the binary (Barrett et al., s ≈ o o 171

Table A.1: Classiﬁcation of cataclysmic variables.

Properties Polars Intermediate Polars DQ Her stars

Lx/Lopt 1-100 0.01-10 transient Accretiondisk no partial yes X-ray spectrum soft hard soft (in outburst) and hard (in quiescence) Orbitalperiod 100-500min 1-50hr 1-10hr Magnetic ﬁeld Strong (107 G) 106 G weak 105 G ∼ X-raypulsations Orbitallocked 2-120min 10-70s X-rayQPOs 0.1-10Hz rare 0.001-0.1Hz Polarization common,highlevel rare common

1988). This relation is equivalent to the condition, R r . In most of these systems, circ ≈ co the magnetospheric radius, rm < a, thus the formation of accretion disk takes place in these systems. A schematic diagram of an IP is shown in Figure A.2. As the material migrates inward in the accretion disk, it eventually encounter a magnetic field strong enough to control the flow of the material, at which point matter follows the magnetic field lines and free-falls onto the surface of the white dwarf (magnetic poles). The shock region above the white dwarf surface heats the material upto 10 - 40 keV and becomes a source of hard X-rays.

X-ray spectrum

11 12 The typical X-ray ﬂuxes of these systems lies in the range 10− to 10− erg

2 1 31 33 1 cm− s− and corresponding luminosities are in the range 10 to 10 erg s− . IPs are generally hard X-ray emitter and their energy spectra can be well described by thermal bremsstrahlung with kT 30 keV. Few cases show a two component model with a 10 ∼ ∼ keV thermal source and a 1 keV optically thin plasma. The hotter component possibly ∼ arises in the accretion zone; the softer one in the post or pre shock region (Singh and

Swank, 1993). Recently, a few IPs have also been detected with a strong soft X-ray component, typically a blackbody with kT 40 - 100 eV. The X-ray spectrum of IPs ∼ 172 A. Magnetic Cataclymic Variables

Figure A.2: A schematic of an intermediate polar showing a white dwarf accreting matter from its companion star. The formation of accretion disk takes place in intermediate polars and the inner accretion disc is usually truncated by the magnetic field of the white dwarf. (This figure is taken from http://heasarc.gsfc.nasa.gov/images). also display strong Fe emission lines between 6.4 - 6.9 keV, providing an evidence of a significant electron scattering in the accretion column in these systems (Norton et al.,

1991).

X-ray timing These sources show an increase in the depth of the orbital modulation at lower energies, indicating photoelectric absorption by obscuring gas (similar to dips which occur in the X-ray lightcurves of the LMXBs, where they are caused by the material above the orbital plane subtending angles 10◦ to 15◦ at the neutron star). The pulse shapes are sinusoidal and energy dependent (with decreasing pulsed fraction for increasing energy). These energy dependent pulse proﬁle cannot be explained only by means of geometrical eﬀect models (such as polar cap occultation by white dwarf body) but required photoelectric absorption models or both mechanisms. Besides the spin and orbital motion, there exist a third type of pulsations, beat pulsations which are generally seen in optical, ultra- 173

voilet (UV) and X-ray wavebands of IPs at Pbeat due to interaction between Ps and Porb (Kuulkers et al., 2006) and is given as follows

1 1 1 = (A.1) Pbeat Ps − Porb DQ Her systems •

These sources form a subset of IPs. Observationally they are defined as rapidly rotating accreting white dwarfs of weak magnetic field. The magnetic moments in the DQ Her white dwarfs are weak enough (B 105 G) that the Keplerian disk accretion ∼ flow is broken up only a few stellar radii above the white dwarf surface. The infall is still energetic enough that accretion columns emit radiation significantly at soft X-ray and UV wavebands. These sources are observed to be lacking in hard X-ray emission. In most of these sources, the circular and linear polarization at optical wavebands (0.1 - 1 %) was detected (Howell et al., 1997). The physical reason behind the rapid and coherent pulsations in these sources is still an open question. The properties of all three classes of magnetic CVs are summarized in Table A.1. 174 A. Magnetic Cataclymic Variables Bibliography

Alpar M.A. and Shaham J. Nature, 316, 239 (1985).

Andrillat A., Jaschek M. and Jaschek C. A&AS, 84, 11 (1990).

Andrillat Y., Jaschek M. and Jaschek C. A&AS, 72, 129 (1988).

Angelini L., Church M.J., Parmar A.N., Balucinska-Church M. and Mineo T. A&A, 339, L41 (1998).

Angelini L., Stella L. and Parmar A.N. ApJ, 346, 906 (1989).

Angelini L., White N.E., Nagase F. et al. ApJ Lett, 449, L41 (1995).

Angelini L., White N.E. and Stella L. ApJ, 371, 332 (1991).

Apparao K.M.V. Space Sci. Rev., 69, 255 (1994).

Arnaud K.A. in Astronomical Society of the Paciﬁc Conference Series. 101, Astronom-

ical Data Analysis Software and Systems V (eds.) Jacoby G.H. and Barnes J., (San Francisco, CA: ASP), 17 (1996).

Babushkina O.P., Kudriavtsev M.I., Melioranskii A.S. et al. Soviet Astronomy Letters, 1, 32 (1975).

Balbus S.A. and Hawley J.F. ApJ, 376, 214 (1991).

175 176 BIBLIOGRAPHY

Barrett P., O’Donoghue D. and Warner B. MNRAS, 233, 759 (1988).

Basko M.M. and Sunyaev R.A. A&A, 42, 311 (1975).

Basko M.M. and Syunyaev R.A. Soviet Astronomy, 20, 537 (1976).

Bhattacharya D. and Srinivasan G. in X-ray binaries, (eds.) Lewin W.H.G., van Paradijs J. and van den Heuvel E.P.J., (Cambridge: Cambridge Univ. Press), 495 (1995).

Bhattacharya D. and van den Heuvel E.P.J. Phys. Rep., 203, 1 (1991).

Bildsten L., Chakrabarty D., Chiu J. et al. ApJS, 113, 367 (1997).

Boroson B., O’Brien K., Horne K. et al. ApJ, 545, 399 (2000).

Bradt H., Levine A.M., Remillard R.A. and Smith D.A. astro-ph/0001460 (2000).

Bradt H.V.D. and McClintock J.E. Annual Review of A&A, 21, 13 (1983).

Burrows D.N., Hill J.E., Nousek J.A. et al. Space Sci. Rev., 120, 165 (2005).

Carroll B.W. and Ostlie D.A. An Introduction to Modern Astrophysics (Reading, MA:

Addison-Wesley) (1996).

Chakrabarty D. ApJ, 492, 342 (1998).

Chakrabarty D., Bildsten L., Grunsfeld J.M. et al. ApJ, 474, 414 (1997).

Chakrabarty D., Homer L., Charles P.A. and O’Donoghue D. ApJ, 562, 985 (2001).

Chakrabarty D., Levine A.M., Clark G.W. and Takeshima T. IAU circ, 7048, 1 (1998a).

Chakrabarty D., Takeshima T., Ozaki M., Paul B. and Yokogawa J. IAU circ, 7062, 1 (1998b).

Chester T.J. ApJ, 227, 569 (1979).

Clark G.W., Woo J.W., Nagase F., Makishima K. and Sakao T. ApJ, 353, 274 (1990). BIBLIOGRAPHY 177

Clarkson W.I., Charles P.A., Coe M.J. et al. MNRAS, 339, 447 (2003).

Coburn W., Heindl W.A., Gruber D.E. et al. ApJ, 552, 738 (2001).

Coburn W., Heindl W.A., Rothschild R.E. et al. ApJ, 580, 394 (2002).

Coe M.J. in IAU Coll. 175, The Be Phenomenon in Early-Type Stars, (eds.) Smith M.A.,

Henrichs H.F. and Fabregat J., (Astronomical Society of the Paciﬁc Conference Series. 214; San Francisco: ASP), 656 (2000).

Coe M.J., Stevens J.B., Buckley D.A.H., Charles P.A. and Southwell K.A. MNRAS, 293,

43 (1998).

Corbet R.H.D. MNRAS, 220, 1047 (1986).

Córdova F.A.D. in X-ray binaries, (eds.) Lewin W.H.G., van Paradijs J. and van den

Heuvel E.P.J., 331 (1995).

Covino S., Negueruela I., Campana S. et al. A&A, 374, 1009 (2001).

Cox A.N. in Allen’s Astrophysical Quantities, (ed.) A. N. Cox (4th ed.; New York: Springer), 388 (2000).

Cutri R.M., Skrutskie M.F., van Dyk S. et al. The IRSA 2MASS All-Sky Point Source Catalog, NASA/IPAC Infrared Science Archive, http://irsa.ipac.caltech.edu/ applica- tions/Gator/ (2003)

Dachs J., Hanuschik R., Kaiser D. and Rohe D. A&A, 159, 276 (1986).

Dickey J.M. and Lockman F.J. Annual Review of A&A, 28, 215 (1990).

Ebisawa K., Day C.S.R., Kallman T.R. et al. PASJ, 48, 425 (1996).

Eggleton P.P. ApJ, 268, 368 (1983).

Elsner R.F. and Lamb F.K. ApJ, 215, 897 (1977). 178 BIBLIOGRAPHY

Ezuka H. and Ishida M. ApJS, 120, 277 (1999).

Fabian A.C. Journal of the British Interplanetary Society, 28, 343 (1975).

Feast M.W., Thackeray A.D. and Wesselink A.J. MNRAS, 122, 433 (1961).

Ferrigno C., Segreto A., Santangelo A. et al. A&A, 462, 995 (2007).

Finger M.H., Wilson R.B. and Harmon B.A. ApJ, 459, 288 (1996).

Fitzpatrick E.L. PASP, 111, 63 (1999).

Frank J., King A. and Raine D.J. Accretion Power in Astrophysics (3rd ed.; Cambridge: Cambridge Univ. Press (2002)

Gehrels N., Chincarini G., Giommi P. et al. ApJ, 611, 1005 (2004).

Ghosh P. and Lamb F.K. ApJ, 234, 296 (1979).

Giacconi R., Gursky H., Kellogg E., Schreier E. and Tananbaum H. ApJ Lett, 167, L67 (1971).

Giacconi R., Gursky H., Paolini F.R. and Rossi B.B. Physical Rev. Lett., 9, 439 (1962).

Giacconi R., Murray S., Gursky H. et al. ApJ, 178, 281 (1972).

Grindlay J.E. ApJ, 225, 1001 (1978).

Haberl F., Angelini L., Motch C. and White N.E. A&A, 330, 189 (1998).

Halpern J.P. and Tyagi S. The Astronomer’s Telegram, 681, 1 (2005).

Hardie R.H. and Ballard C.M. PASP, 74, 242 (1962).

Heindl W., Coburn W., Kreykenbohm I. and Wilms J. The Astronomer’s Telegram, 200,

1 (2003). BIBLIOGRAPHY 179

Heindl W.A. and Chakrabarty D. in Highlights in X-Ray Astronomy, (eds.) B. Aschen-

bach &. M. J. Freyberg (MPE Rep. 272; Garching: MPE), 25 (1999)

Heindl W.A., Coburn W., Gruber D.E. et al. ApJ Lett, 521, L49 (1999).

Heindl W.A., Coburn W., Gruber D.E. et al. ApJ Lett, 563, L35 (2001).

Heise J., Bleeker J.A.M., Brinkman A.C., Gronenschild E. and Paerels F. Advances in Space Research, 5, 61 (1985).

Henrichs H.F. in Accretion-driven Stellar X-Ray Sources, (ed.) W. H. G. Lewin and E. P. J. van den Heuvel (Cambridge: Cambridge Univ. Press), 419 (1983)

Hickox R.C., Narayan R. and Kallman T.R. ApJ, 614, 881 (2004).

Horne K. PASP, 98, 609 (1986).

Howell S.B. Handbook of CCD astronomy (2nd ed.; Cambridge: Cambridge Univ. Press), (2006)

Howell S.B., Sirk M.M., Ramsay G. et al. ApJ, 485, 333 (1997).

Ilovaisky S.A., Motch C. and Chevalier C. A&A, 70, L19 (1978). in ’t Zand J.J.M., Baykal A. and Strohmayer T.E. ApJ, 496, 386 (1998). in ’t Zand J.J.M., Strohmayer T.E. and Baykal A. ApJ Lett, 479, L47 (1997). in’t Zand J.J.M. A&A, 441, L1 (2005).

Israel G.L., Stella L., Covino S., Campana S. and Mereghetti S. IAU circ, 7101, 1 (1999).

Jahoda K., Swank J.H., Giles A.B. et al. in Proc. SPIE 2808, EUV, X-Ray, and Gamma-

Ray Instrumentation for Astronomy, Vol. VII, ed. O. H. Siegmund & M. A. Gummin (Bellingham, WA: SPIE), 59 (1996) 180 BIBLIOGRAPHY

Jain C., Paul B. and Dutta A. Submitted to ApJ (2009).

Jain C., Paul B., Joshi K., Dutta A. and Raichur H. Journal of Astrophysics and Astron- omy, 28, 175 (2007).

Jansen F., Lumb D., Altieri B. et al. A&A, 365, L1 (2001).

Johnson H.L. and Morgan W.W. ApJ, 119, 344 (1954).

Karzas W.J. and Latter R. ApJS, 6, 167 (1961).

Kaur R., Kumar B., Paul B. and Sagar R. The Astronomer’s Telegram, 1807, 1 (2008a).

Kaur R., Paul B., Kumar B. and Sagar R. ApJ, 676, 1184 (2008b).

Kaur R., Paul B., Kumar B. and Sagar R. MNRAS, 386, 2253 (2008c).

Kaur R., Paul B., Raichur H. and Sagar R. ApJ, 660, 1409 (2007b).

Kaur R., Wijnands R., Patruno A. et al. astro-ph/0809.1268 (2008d).

Kellogg E., Baldwin J.R. and Koch D. ApJ, 199, 299 (1975).

Kendziorra E., Kretschmar P., Pan H.C. et al. A&A, 291, L31 (1994).

Kendziorra E., Mony B., Kretschmar P. et al. NASA Conference Publication, 3137, 217 (1992).

Kennea J.A., Racusin J.L., Burrows D.N. et al. The Astronomer’s Telegram, 673, 1

(2005).

Kinugasa K., Torii K., Hashimoto Y. et al. ApJ, 495, 435 (1998).

Kirk J.G. and Galloway D.J. MNRAS, 195, 45P (1981).

Kluzniak W. and Abramowicz M.A. astro-ph/0105057 (2001).

Koenigsberger G., Georgiev L., Moreno E. et al. A&A, 458, 513 (2006). BIBLIOGRAPHY 181

Kommers J.M., Chakrabarty D. and Lewin W.H.G. ApJ Lett, 497, L33 (1998).

Krauss M.I., Schulz N.S., Chakrabarty D., Juett A.M. and Cottam J. ApJ, 660, 605 (2007).

Kuulkers E., Norton A., Schwope A. and Warner B. in Compact Stellar X-ray Sources, (eds.) W. Lewin & M. van der Klis (Cambridge: Cambridge Univ. Press), 421 (2006)

La Barbera A., Burderi L., Di Salvo T., Iaria R. and Robba N.R. ApJ, 553, 375 (2001).

Lamb F.K., Shibazaki N., Alpar M.A. and Shaham J. Nature, 317, 681 (1985).

Lamb P. Planetary and Space Science, 27, 1525 (1979).

Landolt A.U. AJ, 104, 340 (1992).

Lang K.R. Astrophysical Data: Planets and Stars (New York: Springer)(1992)

Langer S.H. and Rappaport S. ApJ, 257, 733 (1982).

Leahy D.A. MNRAS, 251, 203 (1991).

Leitherer C. ApJ, 326, 356 (1988).

Levine A., Ma C.P., McClintock J. et al. ApJ, 327, 732 (1988).

Levine A.M., Bradt H., Cui W. et al. ApJ Lett, 469, L33 (1996).

Levine A.M., Delgado-Marti H., Pfahl E. and Rappaport S. Strohmayer T.E., (eds.)

Rossi2000: Astrophysics with the Rossi X-ray Timing Explorer (2000).

Li F.K., McClintock J.E., Rappaport S., Wright E.L. and Joss P.C. ApJ, 240, 628 (1980).

Lin X.B., Church M.J., Nagase F. and Bałucinska-Church´ M. MNRAS, 337, 1245 (2002).

Liu Q.Z., van Paradijs J. and van den Heuvel E.P.J. A&AS, 147, 25 (2000).

Liu Q.Z., van Paradijs J. and van den Heuvel E.P.J. A&A, 442, 1135 (2005). 182 BIBLIOGRAPHY

Liu Q.Z., van Paradijs J. and van den Heuvel E.P.J. A&A, 455, 1165 (2006).

Liu Q.Z., van Paradijs J. and van den Heuvel E.P.J. A&A, 469, 807 (2007).

Longair M.S. High energy astrophysics. Volume 2. Stars, the Galaxy and the interstellar medium. (Cambridge: Cambridge University Press), 1 (1994).

Makishima K. in the Physics of Accretion onto Compact Objects. (eds.) K. O. Mason, M. G. Watson, & N. E. White (Berlin: Springer), 1, (1986).

Makishima K. and Mihara T. in Frontiers of X-Ray Astronomy, Proc, XXVIII Yamada Conf., (eds.) Y. Tanaka, & K. Koyama, Frontiers Science Series, 2, 23 (1992)

Makishima K., Mihara T., Ishida M. et al. ApJ Lett, 365, L59 (1990).

Maraschi L., Traversini R. and Treves A. MNRAS, 204, 1179 (1983).

Maraschi L., Treves A. and van den Heuvel E.P.J. Nature, 259, 292 (1976).

Masetti N., Bassani L., Bazzano A. et al. A&A, 455, 11 (2006).

Masetti N., Parisi P., Palazzi E. et al. A&A, 495, 121 (2009).

McClintock J.E., Bradt H.V., Doxsey R.E. et al. Nature, 270, 320 (1977).

McClintock J.E., Li F.K., Canizares C.R. and Grindlay J.E. ApJ Lett, 235, L81 (1980).

McClintock J.E. and Remillard R.A. ApJ, 308, 110 (1986).

Mereghetti S., Romano P. and Sidoli L. A&A, 483, 249 (2008).

Meszaros P., Harding A.K., Kirk J.G. and Galloway D.J. ApJ Lett, 266, L33 (1983).

Middleditch J., Mason K.O., Nelson J.E. and White N.E. ApJ, 244, 1001 (1981).

Mihara T., Makishima K., Kamijo S. et al. ApJ Lett, 379, L61 (1991). BIBLIOGRAPHY 183

Milligan S., Cranton B.W. and Skrutskie M.F. Proc. SPIE, (eds.) Fischer R.E. and Smith

W.J., 2863, 2 (1996)

Mitsuda K., Inoue H., Koyama K. et al. PASJ, 36, 741 (1984).

Moon D.S. and Eikenberry S.S. ApJ Lett, 552, L135 (2001a).

Moon D.S. and Eikenberry S.S. ApJ Lett, 549, L225 (2001b).

Motch C., Stella L., Janot-Pacheco E. and Mouchet M. ApJ, 369, 490 (1991).

Mukerjee K., Agrawal P.C., Paul B. et al. ApJ, 548, 368 (2001).

Mukherjee U., Bapna S., Raichur H., Paul B. and Jaaﬀrey S.N.A. Journal of Astrophysics

and Astronomy, 27, 25 (2006).

Nagase F. PASJ, 41, 1 (1989).

Nagase F. X-ray Astronomy: Stellar Endpoints, AGN, and the Diﬀuse X-ray Background, 599, 254 (2001).

Nagel W. ApJ, 251, 288 (1981a).

Nagel W. ApJ, 251, 278 (1981b).

Negueruela I. astro-ph/0411335 (2004).

Negueruela I., Okazaki A.T., Fabregat J. et al. A&A, 369, 117 (2001).

Negueruela I., Roche P., Fabregat J. and Coe M.J. MNRAS, 307, 695 (1999).

Negueruela I., Smith D.M., Reig P., Chaty S. and Torrejón J.M. Wilson A., (eds.) The X-ray Universe 2005, volume 604 of ESA Special Publication, 165 (2006).

Nelson L.A., Rappaport S.A. and Joss P.C. ApJ, 311, 226 (1986). 184 BIBLIOGRAPHY

Norton A.J., Watson M.G. and King A.R. Treves A., Perola G.C. and Stella L., (eds.) Iron

Line Diagnostics in X-ray Sources, volume 385 of Lecture Notes in Physics, Berlin Springer Verlag, 155 (1991).

Ogilvie G.I. and Dubus G. MNRAS, 320, 485 (2001).

Oosterbroek T., Orlandini M., Parmar A.N. et al. A&A, 351, L33 (1999).

Orlandini M., Fiume D.D., Frontera F. et al. ApJ Lett, 500, L163 (1998).

Owens A., Oosterbroek T. and Parmar A.N. A&A, 324, L9 (1997).

Paczynski B. and Sienkiewicz R. ApJ Lett, 248, L27 (1981).

Parmar A.N., White N.E. and Stella L. ApJ, 338, 373 (1989).

Paul B., Raichur H. and Mukherjee U. A&A, 442, L15 (2005).

Paul B. and Rao A.R. A&A, 337, 815 (1998).

Pfahl E., Rappaport S., Podsiadlowski P. and Spruit H. ApJ, 574, 364 (2002).

Podsiadlowski P., Langer N., Poelarends A.J.T. et al. ApJ, 612, 1044 (2004).

Pottschmidt K., Kreykenbohm I., Wilms J. et al. ApJ, 634, 97 (2005).

Predehl P. and Schmitt J.H.M.M. A&A, 293, 889 (1995).

Prinja R.K., Massa D. and Searle S.C. A&A, 430, L41 (2005).

Psaltis D. in Compact Stellar X-ray Sources, (eds.) W. Lewin & M. van der Klis (Cam- bridge: Cambridge Univ. Press), 1 (2006)

Qu J.L., Zhang S., Song L.M. and Falanga M. ApJ Lett, 629, L33 (2005).

Raichur H. and Paul B. ApJ, 685, 1109 (2008). BIBLIOGRAPHY 185

Ramsay G., Zane S., Jimenez-Garate M.A., den Herder J.W. and Hailey C.J. MNRAS,

337, 1185 (2002).

Rappaport S., Markert T., Li F.K. et al. ApJ Lett, 217, L29 (1977).

Rappaport S.A. and Joss P.C. Lewin W.H.G. and van den Heuvel E.P.J., (eds.) Accretion-

Driven Stellar X-ray Sources, 1 (1983).

Raymond J.C. and Smith B.W. ApJS, 35, 419 (1977)

Reig P. and Roche P. MNRAS, 306, 100 (1999).

Remillard R.A. and McClintock J.E. Annual Review of A&A, 44, 49 (2006).

Ritter H. A&AS, 85, 1179 (1990).

Ritter H. and Kolb U. in Lewin W.H.G., van Paradijs J. and van den Heuvel E.P.J., (eds.) X-ray binaries, (Cambridge: Cambridge Univ. Press), 578 (1995).

Sakano M., Torii K., Koyama K., Maeda Y. and Yamauchi S. PASJ, 52, 1141 (2000).

Salpeter E. E. ApJ, 140, 796 (1964).

Sambruna R.M., Chiappetti L., Treves A. et al. ApJ, 391, 750 (1992).

Santangelo A., del Sordo S., Segreto A. et al. A&A, 340, L55 (1998).

Schmidt G.D. and Stockman H.S. ApJ, 548, 410 (2001).

Schmidt-Kaler T. Landolt-Bornstein: numerical data (eds.) K. Schaifers, & H. H. Voigt (Springer-Verlag), (1982).

Schulz N.S., Chakrabarty D., Marshall H.L. et al. ApJ, 563, 941 (2001).

Shakura N.I. and Syunyaev R.A. A&A, 24, 337 (1973).

Shibazaki N. and Lamb F.K. ApJ, 318, 767 (1987). 186 BIBLIOGRAPHY

Shinoda K., Kii T., Mitsuda K. et al. PASJ, 42, L27 (1990).

Singh J. and Swank J. MNRAS, 262, 1000 (1993).

Skrutskie M.F., Cutri R.M., Stiening R. et al. AJ, 131, 1163 (2006).

Slettebak A. and Snow T.P. in Physics of Be stars. Proceedings of the 92nd Colloquium

of the IAU, held at Boulder, CO, USA, 18 - 22 August 1986. (1987)

Soong Y.and Swank J.H. in Two Topics in X-Ray Astronomy, Volume 1: X Ray Binaries. Volume 2: AGN and the X Ray Background, (eds.) Hunt J. and Battrick B., ESA Special Publication, 296, 617 (1989).

Steiner C., Eckert D., Mowlavi N., Decourchelle A. and Vink J. The Astronomer’s Tele- gram, 672, 1 (2005).

Stella L. and Vietri M. ApJ Lett, 492, L59 (1998).

Stella L., White N.E. and Rosner R. ApJ, 308, 669 (1986).

Stetson P.B. PASP, 99, 191 (1987).

Stetson P.B. PASP, 102, 932 (1990).

Strüder L., Briel U., Dennerl K. et al. A&A, 365, L18 (2001).

Sugizaki M., Kinugasa K., Matsuzaki K. et al. ApJ Lett, 534, L181 (2000).

Takeshima T. Bulletin of the American Astronomical Society, 29, 1391 (1997).

Takeshima T., Dotani T., Mitsuda K. and Nagase F. PASJ, 43, L43 (1991).

Takeshima T., Dotani T., Mitsuda K. and Nagase F. ApJ, 436, 871 (1994).

Tauris T.M. and van den Heuvel E.P.J. in Compact Stellar X-ray Sources, (eds.) W. Lewin & M. van der Klis (Cambridge: Cambridge Univ. Press), 623 (2006) BIBLIOGRAPHY 187

Taylor J.H. Philosophical Transactions of the Royal Society of London, 341, 117 (1992).

Torii K., Kinugasa K., Katayama K. et al. ApJ, 508, 854 (1998).

Torii K., Sugizaki M., Kohmura T., Endo T. and Nagase F. ApJ Lett, 523, L65 (1999).

Truemper J., Pietsch W., Reppin C. et al. ApJ Lett, 219, L105 (1978).

Turner M.J.L., Abbey A., Arnaud M. et al. A&A, 365, L27 (2001). van den Berg M., Tagliaferri G., Belloni T. and Verbunt F. A&A, 418, 509 (2004). van den Heuvel E.P.J. and Rappaport S. in Physics of Be Stars (Cambridge: Cambridge Univ. Press), 291 (1987) van der Klis M. Bulletin of the American Astronomical Society, 19, 746 (1987). van der Klis M. Advances in Space Research, 38, 2675 (2006a). van der Klis M. in Compact Stellar X-ray Sources, (eds.) W. Lewin & M. van der Klis (Cambridge: Cambridge Univ. Press), 39, (2006b). van Paradijs J. astro-ph/9802177 (1998).

Vasyliunas V.M. Space Sci. Rev., 24, 609 (1979).

Wang Y.M. and Frank J. A&A, 93, 255 (1981).

Wang Y.M. and Welter G.L. A&A, 102, 97 (1981).

Warner B. in Astronomical Society of the Paciﬁc Conference Series, 261, The Physics of Cataclysmic Variables and Related Object, (ed.) B. T. Gansicke, K. Beuermann, & K. Reinsch (San Francisco: ASP), 406 (1995)

Waters L.B.F.M. A&A, 162, 121 (1986).

Weisskopf M.C., Brinkman B., Canizares C. et al. PASP, 114, 1 (2002). 188 BIBLIOGRAPHY

Wheaton W.A., Doty J.P., Primini F.A. et al. Nature, 282, 240 (1979).

Wheaton W.A., Dunklee A.L., Jacobsen A.S. et al. ApJ, 438, 322 (1995).

White N.E. in Astronomical Society of the Paciﬁc Conference Series. 121, Accretion Phenomena and Related Outﬂows, IAU Coll. 163, (ed.) D. T. Wickramasinghe, G. V.

Bicknell, & L. Ferrario (San Francisco: ASP), 142 (1997)

White N.E., Angelini L., Ebisawa K., Tanaka Y. and Ghosh P. ApJ Lett, 463, L83 (1996).

White N.E., Kallman T.R. and Swank J.H. ApJ, 269, 264 (1983).

White N.E., Nagase F. and Parmar A.N. in X-ray binaries, (eds.) Lewin W.H.G., van

Paradijs J. and van den Heuvel E.P.J., (Cambridge: Cambridge Univ. Press), 1 (1995).

Wilson C.A. and Finger M.H. IAU circ, 7048, 2 (1998).

Wojdowski P., Clark G.W., Levine A.M., Woo J.W. and Zhang S.N. ApJ, 502, 253 (1998).

Yokogawa J., Paul B., Ozaki M. et al. ApJ, 539, 191 (2000).

Zhang F., Li X.D. and Wang Z.R. ApJ, 603, 663 (2004).

Zhang W., Morgan E.H., Jahoda K. et al. ApJ Lett, 469, L29 (1996).