Universitat de - Departament d'Astronomia i Meteorologia ...

UNIVERSITAT DE BARCELONA
Departament d’Astronomia i Meteorologia
Discovery and study of the
microquasar LS 5039 and a
search for new microquasars
Memòria presentada per
Marc Ribó Gomis
per optar al grau de
Doctor en Ciències Físiques
Barcelona, setembre de 2002

Programa de Doctorat d’Astronomia i Meteorologia
Bienni 1996–1998
Memòria presentada per Marc Ribó Gomis per optar al grau de
Doctor en Ciències Físiques
Director de la tesi
Dr. Josep M. Paredes Poy

Per a la meva neboda Blanca,
que va començar l’aventura de viure
tot just al final d’aquesta etapa de la meva vida

Agraïments
Agraïments / Agradecimientos / Acknowledgements
Després d’aquesta llarga aventura que és la realització d’una tesi doctoral, hi ha
molta gent a qui vull donar les gràcies. Abans que res, però, demano disculpes per
si em deixo algú que mereixeria ser-hi i no hi és, doncs el cansament acumulat en
aquests moments fa difícil mantenir-se prou despert.
En primer lloc vull donar les gràcies al Josep Maria Paredes, el meu director de
tesi, no només pel que m’ha ensenyat científicament, sinó també personalment. Vull
agrair epecialment la seva disponibilitat, dedicació i seguiment de la meva feina,
que ha permès realitzar una gran quantitat de treball sense perdre els papers en
cap moment. Sense les seves idees i suggerències aquesta tesi no hagués estat mai
possible.
També vull agrair molt especialment el suport del Josep Martí, que m’ha ensenyat
a analitzar dades ràdio interferomètriques i òptiques. Vull agrair la gran quantitat
de preguntes que ha contestat amablement a través del correu electrònic o el telèfon
al llarg d’aquests anys, fos allà on fos, i que sempre han estat molt útils. Gran part
de la feina que aquí presento no hagués estat possible sense ell.
I would like to thank very much the support I had from Maria Massi when I was
in Bonn in 1999. I learned to reduce and analyze VLBI data with her, and part
of this thesis has been possible thanks to her useful comments and suggestions. I
would also like to thank her for the very interesting and sometimes truly discussions
that we have mantained during these years. It’s great to see your enthusiasm!
Gràcies a la Marta Peracaula, per l’anglès, per com aclareix i explica de manera
planera problemes que de vegades semblen irresolubles, especialment pel que fa a

feina no presentada en aquesta tesi. Gràcies per donar-me, encara que potser no en
siguis conscient, paciència i tranquilitat en determinats moments d’aquests darrers
anys. I gràcies també per ensenyar-me moltes coses de software i administració de
màquines.
Muchas gracias a Eduardo Ros, por todo lo que ha hecho por mi tanto a nivel
científico como personal, y especialmente durante mi estancia en Bonn en 1999 y
durante la semana que pasamos analizando datos en Dwingeloo en el año 2000.
Gracias por enseñarme tantas cosas de VLBI, por tus sabios consejos y por tu
pulcritud en el trabajo que hemos realizado juntos. Gracias también a Ana, tu
mujer, por el apoyo recibido durante mi estancia en Bonn.
Moltes, moltes gràcies a l’Octavi Fors, per ensenyar-me tantes coses innovadores
de software astronòmic, tantes eines, i sobretot astrometria òptica. Estic molt con-
tent d’haver pogut treballar amb tu. A banda d’això has estat un company de
despatx d’aquells que no voldries perdre mai.
Moltes gràcies al Xavier Otazu. En primer lloc perquè ell va ser el primer que
em va donar la idea de realitzar una tesi doctoral. En segon lloc, per ser tan crític
i escèptic en molts dels temes de que hem parlat durant aquests anys, i que han fet
que m’hagués de replantejar moltes coses. D’altra banda, tenir un article amb tu, i
sobre wavelets, és alguna cosa més que un simple article, després de tants anys de
pensar en possibles articles que mai van tirar endavant per falta de temps.
Gràcies també al Joan García, pels seus consells tan útils quan m’he hagut
d’enfrontar amb càlculs d’òrbites. També li vull agrair les diverses vegades que ha
revisat l’anglès d’alguns dels meus escrits, i la manera tan acurada com ho ha fet.
Gràcies al Robert Estalella, per ensenyar-me rigurositat i pulcritud a l’hora d’a-
cabar un article, pels consells d’anglès, d’AIPS i pel seu saber fer.
Moltes gràcies a l’Ignasi Ribas, per la gran quantitat de preguntes de software,
anglès i especialment d’astrofísica estel·lar, que m’ha contestat en aquests darrers
anys. Algú com tu li dóna un valor afegit a un departament.
Gràcies a l’Eduard Masana, pel latex, pel Java, pel Netscape, pels scripts, pel
linux i, com no, pel masanix. Gràcies per la gran quantitat de problemes solucionats.
Moltes gràcies al Valentí Bosch, que encara que fa poc temps que s’ha incorporat

al grup, ja ha realitzat aportacions importants que m’han ajudat en alguns aspectes
d’aquesta tesi.
Gràcies al Pau Reig, per contestar amb detall moltes preguntes sobre anàlisi
temporal i espectral de dades de satèl·lits de raigs X, i per estar a l’altra banda del
correu electrònic o del telèfon quan ha calgut.
Gracias a Ignacio Negueruela, por estar varias veces al otro lado del teléfono o
del correo electrónico cuando me han surgido dudas al respecto de binarias de rayos
X masivas, y por compartir información al respecto de LS 5039.
Moltes gràcies al Jorge Casares. Ha estat un plaer, i espero que ho segueixi sent,
treballar amb tu. Recordo amb alegria molts moments passats a París durant la
nostra setmana de feina intensiva en el projecte MINE durant la primavera de 2002,
així com les interessants explicacions didàctiques sobre espectroscopia òptica tant a
Granada com a Canàries.
Gràcies al Blai Sanahuja, per les potser escasses però interessants i de vegades
llargues converses sobre política científica, investigació i “dinàmica de grups”, que
han fet que pogués assimilar alguns dels mals tràngols passats durant aquests anys.
I sí, suposo que encara segueixo sent un d’aquest “joves impetuosos”.
També gràcies a la Cesca Figueras, per atendre’m quan he tingut dubtes sobre
estrelles joves i fotometria estel·lar, i en especial per la seva acurada dedicació pel
que fa a la disponibilitat de revistes electròniques.
Encara que ja no sigui al departament, vull donar les gràcies al Jaume Soley, el
tècnic informàtic que hem tingut durant molts anys. Gràcies no només per solucionar
els problemes, sinó també per les explicacions, per la seva capacitat de resposta i
pel seu entusiasme quan hi havia problemes.
També vull donar les gràcies al JR, que sempre té a punt tota la paperassa
necessària, i amb l’ajuda del qual moltes coses semblen senzilles quan realment no
ho són. Gràcies especialment per la cura amb que ha dut la paperassa necessària
per dipositar aquesta tesi.
També, com no, vull donar les gràcies a l’ Óscar Morata i l’Inma Sepúlveda que,
a part de proporcionar-me els estils per escriure aquesta tesi, sempre han estat a
prop, encara que fos un cop l’any, per tenir converses i entendre moltes coses, per

desfogar-nos conjuntament, per la seva comprensió i recolzament en molts aspectes
durant aquests anys, i per algunes nits de festa en el món exterior.
Gràcies també al Josep Miquel Girart, per moltes i diferents qüestions que m’ha
solucionat, i per algunes converses interessants que hem tingut en aquests darrers
dos anys.
I encara que ja fa molt temps que va marxar, també vull donar les gràcies a la
Maite, per mantenir la seva frescura i alegria durant els durs primers dos anys de
tesi, i per enviar notícies de tant en tant.
Vull donar les gràcies a tots els incondicionals de l’hora de dinar durant aquests
anys, per les converses i debats, i per compartir amb mi algunes de les penes i
alegries que hem passat durant aquest temps: Maite, David, Teresa, Xavi, Octavi,
Marta, Ricard, Eduard, Albert Domingo, Josep Miquel, Pep, Andreu Raig, Ada, i
en algunes ocasions, especialment en aquest darrer i dur estiu, també l’ Àngels.
Gràcies també a tota la gent del Departament d’Astronomia i Meteorologia que,
d’alguna manera o altra, ha fet més fàcil la meva feina durant aquests anys.
Although he will never read this, I would like to thank Jan van Paradijs, for
pushing the field of X-ray binaries, for the incredible work he did when compiling
the catalog of X-ray binaries, for his friendly character, and for offering me to join
his group on gamma-ray bursts in 1997.
Thanks to Titus Galama, for welcoming me the several times I have been in
Amsterdam, and for his useful and polite comments. Thanks also for contacting
and offering me to work on GRBs.
Gracias también a Mariano Méndez, por acogerme tan bien durante mis visitas
a Amsterdam, por los proyectos de colaboración que al final no cristalizaron, y por
su entusiasmo y simpatía.
Thanks to many people of the VLBI group at Bonn, specially to Giuseppe Cimò,
Mauro Sorgente, Andrea Tarchi, Andrei Lobanov, Yoshiaki Hagiwara, Alan Roy,
Richard Porcas, Alok Patnaik, Walter Alef and also Anton Zensus, for allowing me
to join the group during 5 months in 1999.
Thanks to Simon Garrington for his help during my visit to Jodrell Bank in

March 2000. Thanks to the people that were at JIVE helping me in November
2000, specially to Denise Gabuzda, Cormac Reynolds, Mike Garrett, Ian Avruch
(Max), Bob Campbell, Leonid Gurvits and Huib van Langevelde.
Thanks to Ginny McSwain, for sharing interesting information on LS 5039 while
I was finishing my thesis.
Muchas gracias a Félix Mirabel, por pensar en mi para un postdoc y por ofre-
cerme la posibilidad de trabajar en un proyecto tan interesante como MINE. Siempre
con ideas nuevas, siempre con la cabeza en marcha, consiguiendo así ilusionar con sus
interesantes propuestas. Gracias también por como nos acogió en su casa durante
la dura semana del proyecto MINE en París.
Un abrazo y muchas a gracias a los colaboradores argentinos Gustavo Romero
y Paula Benaglia. Ha sido muy enriquecedor y un placer trabajar con vosotros, y
espero que siga siéndolo.
Como, no, un fuerte abrazo a Jorge Combi, por sus continuos ánimos en esta
etapa final de mi tesis, y por los buenos momentos pasados durante esa dura semana
en París. Gracias por las risas y las charlas.
Gracias a los “VLBIeros” con los que he coincidido en varios congresos durante
estos años, que tan bien me han acogido y con los que he compartido tantos mo-
mentos simpáticos y divertidos, que le hacían a uno olvidar que se encontraba en
cualquier lugar del mundo, y en especial a Eduardo Ros, José Luís Gómez, Iván
Agudo, Miguel
Ángel Pérez Torres, José Carlos Guirado, Lucas Lara, María José
Rioja, Antxón Alberdi y Jon Marcaide.
Thanks to the “VLBIers” Zsolt Paragi and Michele Pestalozzi, for making me
laugh and enjoy some of the conferences during these years.
Since this thesis is based on 7 published papers, I would like to thank all the
people who have made useful comments or suggestions while writing them, in partic-
ular: O. Fors, X. Otazu, I. Ribas, D. Fernández, E. Masana, F. Figueras, J. Colomé.
I would also like to thank all those who have read draft versions and made useful
suggestions: M. Peracaula, J. García-Sánchez, A. Lobanov, R. Porcas, W. Alef, and
J. Bloom. Finally, I would also like to thank the referees for their comments: R.
Fender, H. Falcke, L. F. Rodríguez and F. Colin.

From the scientifical point of view, I would finally like to thank Rob Fender, for
the interesting and stimulating discussions and informal conversations mantained in
conferences during the early years of my PhD, that gave me lots of interesting ideas.
Thanks also for pushing the field of radio jet X-ray binaries (a.k.a. microquasars).
Gràcies als incondicionals d’ASTER en les sortides observacionals de fa uns anys,
per les bones estones passades i per tot el que m’heu ensenyat: Jordi A., Montserrat
G., Jordi B., David, Montse P. i Xavi.
Gràcies a la Rosa Maria Ros, per iniciar-me en l’astronomia en les seves classes
d’EATP de l’institut.
També vull donar les gràcies a totes aquelles persones que algun cop m’han
preguntat sobre astronomia i han escoltat amb interès, perquè fan necessària i útil
la nostra tasca d’investigadors.
Agraeixo les dues beques que m’ha concedit la CIRIT (Generalitat de Catalunya,
referències 1998 BEAI 200293 i 1999 FI 00199).
Des del punt de vista personal, moltíssimes gràcies a tota la colla del Balmes,
pels sopars, caps de setmana, paelles, festes, tertúlies i un llarg i molt llarg etcètera,
i en especial pel suport i els ànims dels darrers mesos: Toni, Bego, Núria, Iolanda,
Mar, Leti,
Àlex, Joan, Jordi, Raquel, Raül, Laura, Jaume, etc... D’entre ells, vull
agrair especialment el suport del Toni i la Bego en els moments en que calia ser allà.
També vull recordar aquí que el Marc Llimós, que mai llegirà aquests agraïments,
va ensenyar-me que calia lluitar per aconseguir aquelles coses que valien la pena.
Gracias también a Simone y Roman. Lo siento, pero todavía no os puedo escribir
en alemán! Muchas gracias por vuestros continuos ánimos (hace dos minutos que
acabo de recibir un mail de Roman animándome en la etapa final).
Com no, moltes gràcies al Jordi Alborch, perquè va fer amenes i/o suportables
certes etapes de les meves estades a Bonn i al JIVE. Per les festes de Brusel·les, Sant
Feliu, Begur, Barcelona i Madrid, entre d’altres. Per l’alegria i per l’entusiasme, per
la capacitat de reacció i diversió, que han conseguit evadir-me en molts casos en què
ho necessitava.
Gràcies també a la gent de la “quinta Alborch”, com diuen alguns: Àlex Sánchez,

Àlex Gómez, Manel, Mercè, Pau, Cuca, Mario i Céline. Per les sortides de cap de
setmana o de festa i per les vegades que m’heu fet oblidar els problemes de la feina.
Gràcies al Francesc, al Pablo i en especial al Pep, els nois de l’Espai (s’Espai), per
l’espontaneïtat, improvització, performances i altres diversions que han aconseguit
fer més fàcils i agradables certs moments durs de la meva vida, especialment durant
els primers anys de tesi.
I aquí arriba l’agraïment als companys de carrera del sopar mensual, per les di-
vagacions, extravagàncies, “puñaladas traperas” que han fet que ens descargoléssim
de riure, tertúlies interessants, etc: David Nofre, David Pino, Toni Hernández, Toni
Magariños (Pato) i Oriol Saladrigas. Moltes gràcies en especial al David Nofre per
les llargues converses que hem mantingut durant aquests anys.
També vull agrair a l’Imma Bastida els ànims que m’ha donat els darrers mesos.
Gracias a Antonio Vallecillos, por los continuos ánimos y por compartir conmigo
mi desahogo deportivo (y también psíquico) semanal.
A tota la gent del nucli dur de D-Recerca, per la feina feta, la il·lusió, les ganes,
i per intentar canviar allò que creiem que no és just: Teresa, Cesca, Elisenda, Ana,
Xavi de Pedro, Eduard Serra, Hèctor, Sílvia, Miquel, María i un llarg etcètera,
d’entre els quals vull afegir l’Italo, per les interessants converses sobre política en el
darrer any.
A toda la gente del núcleo duro de Precarios, por desahogarnos juntos, por
llorar esta política científica que tenemos y por intentar hacer algo por cambiarla, y
también, como no, por lo bien que nos lo hemos pasado juntos en varias ocasiones:
Pastora, Teresa, Ana, Roberto, César, Cristina, Toni, Marta, Xavi de Pedro, Pablo,
y un largo etcétera.
Gracias a Jacinta, por estar ahí cada semana durante los últimos 25 años, por las
charlas de los lunes por la mañana y por los fantásticos platos tantas veces cocinados.
Vull agrair als meus pares, Ramon i Elena, l’educació que m’han donat, el suport
econòmic durant tots aquests anys de sou precari, els ànims i el suport moral en els
moments més durs. No us podré agrair mai prou l’oportunitat que m’heu donat de
realitzar una tesi doctoral. Gràcies.

Gràcies a la meva germana Nàdia, pels anys que hem viscut junts i per ser
una excel·lent companya de pis durant els primers anys de tesi. De vegades encara
trobo a faltar aquelles xerrades de diumenge explicant-nos la setmana. En fi, va ser
important tenir-te al meu costat durant la difícil situació personal/professional dels
primers dos anys de tesi sense beca. Gràcies també al Pau pels ànims i el suport.
I finalment, l’agraïment més important. Gràcies Noemí. Gràcies per cuidar-
me, per animar-me, per escoltar-me en els moments en que més ho necessito, per
recolzar-me durant aquest llarg i dur estiu, per ser com ets i per estar sempre al
meu costat.

Contents
Resum de la tesi:
Descobriment i estudi del microquàsar LS 5039 i
una cerca de nous microquàsars vii
1 Introduction and background 1
1.1 X-ray binaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 High mass X-ray binaries . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Low mass X-ray binaries . . . . . . . . . . . . . . . . . . . . . 6
1.1.3 Radio emitting X-ray binaries . . . . . . . . . . . . . . . . . . 8
1.2 Microquasars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.1 Projection and special relativity effects . . . . . . . . . . . . . 12
1.2.2 Quasars and microquasars . . . . . . . . . . . . . . . . . . . . 16
1.2.3 Known microquasars . . . . . . . . . . . . . . . . . . . . . . . 17
1.2.4 Accretion disk and jet formation . . . . . . . . . . . . . . . . . 21
1.2.5 Black hole states and radio emission . . . . . . . . . . . . . . 22
1.3 Motivation of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . 24
i

ii CONTENTS
I Discovery and study of the microquasar LS 5039 31
2 Multiwavelength approach to LS 5039 33
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2 Chronology of findings on LS 5039 . . . . . . . . . . . . . . . . . . . 34
2.3 An interesting target in the search for new microquasars . . . . . . . 35
2.4 The radio counterpart: NVSS J182614−145054 . . . . . . . . . . . . 36
2.4.1 A radio counterpart in the NVSS . . . . . . . . . . . . . . . . 36
2.4.2 VLA observations, discovery of a REXB . . . . . . . . . . . . 37
2.4.3 Long-term GBI monitoring, a persistent radio source . . . . . 42
2.4.4 VLBA observations, discovery of a microquasar . . . . . . . . 46
2.4.5 EVN and MERLIN observations,
confirmation of persistent relativistic radio jets . . . . . . . . . 50
2.5 The optical/IR star: LS 5039 . . . . . . . . . . . . . . . . . . . . . . 55
2.5.1 Photometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.5.2 Spectral type . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.5.3 Distance and its uncertainty . . . . . . . . . . . . . . . . . . . 59
2.5.4 Radial velocity curve . . . . . . . . . . . . . . . . . . . . . . . 60
2.6 The X-ray counterpart: RX J1826.2−1450 . . . . . . . . . . . . . . . 65
2.6.1 The ASM/RXTE data . . . . . . . . . . . . . . . . . . . . . . 65
2.6.2 PCA/RXTE observations . . . . . . . . . . . . . . . . . . . . 67
2.6.3 BeppoSAX observations . . . . . . . . . . . . . . . . . . . . . 73
2.7 The γ-ray counterpart: 3EG J1824−1514 . . . . . . . . . . . . . . . . 74

CONTENTS iii
2.8 A proposed scenario to explain the multiwavelength behavior . . . . . 79
2.8.1 Spectral energy distribution . . . . . . . . . . . . . . . . . . . 79
2.8.2 A model based on the γ-ray/radio emission . . . . . . . . . . . 80
2.8.3 Energetic considerations . . . . . . . . . . . . . . . . . . . . . 84
2.8.4 Future prospects . . . . . . . . . . . . . . . . . . . . . . . . . 85
2.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3 LS 5039 as a runaway microquasar 93
3.1 Positions and proper motions . . . . . . . . . . . . . . . . . . . . . . 94
3.1.1 Optical positions . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.1.2 Radio positions . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.1.3 Proper motions . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.2 Ejection from the galactic plane . . . . . . . . . . . . . . . . . . . . . 98
3.3 The past trajectory of LS 5039 . . . . . . . . . . . . . . . . . . . . . 99
3.4 SNR G016.8−01.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.4.1 A lower limit of the distance . . . . . . . . . . . . . . . . . . . 103
3.4.2 Particle density estimates . . . . . . . . . . . . . . . . . . . . 104
3.5 The H i surroundings of LS 5039 . . . . . . . . . . . . . . . . . . . . . 106
3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

iv CONTENTS
II A search for new microquasars 119
4 The cross-identification method 121
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.2 Cross-identification between RBSC and NVSS catalogs . . . . . . . . 121
4.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5 Radio and optical observations 131
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.2 Radio observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.2.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.3 Optical observations . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.3.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.4.1 Discussion on individual sources . . . . . . . . . . . . . . . . . 142
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6 EVN and MERLIN observations 149
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.2 Observations and data reduction . . . . . . . . . . . . . . . . . . . . . 149
6.2.1 MERLIN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.2.2 EVN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
6.2.3 Combining EVN and MERLIN . . . . . . . . . . . . . . . . . 153
6.2.4 Flux density measurements at the 100 m antenna in Effelsberg 153

CONTENTS v
6.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.3.1 1RXS J001442.2+580201 and its two-sided jet . . . . . . . . . 155
6.3.2 1RXS J013106.4+612035 and its one-sided jet . . . . . . . . . 158
6.3.3 1RXS J042201.0+485610, a non-detected source . . . . . . . . 159
6.3.4 1RXS J062148.1+174736, a compact source . . . . . . . . . . 159
6.3.5 1RXS J072259.5−073131 and its bent one-sided jet . . . . . . 160
6.3.6 1RXS J072418.3−071508, a quasar with a bent one-sided jet . 162
6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
7 Summary 167
III General conclusions 169

vi CONTENTS

Resum de la tesi:
Descobriment i estudi del
microquàsar LS 5039 i
una cerca de nous microquàsars
1. Introducció i antecedents sobre el tema
Els microquàsars són estrelles binàries de raigs X que mostren jets relativistes en
longituds d’ona ràdio. En aquesta secció farem una breu introduccio a les estrelles
binàries de raigs X, per després passar a comentar algunes de les propietats dels
microquàsars, i concloure amb l’explicació de quina va ser la motivació d’aquesta
tesi.
1.1 Estrelles binàries de raigs X
Les estrelles binàries de raigs X són sistemes binaris que contenen un objecte com-
pacte, tant pot ser un forat negre com una estrella de neutrons, que acreta matèria
de l’estrella companya.
Per tal d’explicar l’emissió en raigs X detectada en aquests sistemes podem fer
el següent raonament. La matèria que és acretada per l’objecte compacte és accel-
erada a velocitats relativistes en transformar la seva energia potencial gravitatòria,
deguda a l’intens camp gravitatori de l’objecte compacte, en energia cinètica. D’al-
tra banda, el ritme d’acreció de matèria té un valor màxim teòric, anomenat límit
vii

viii Resum de la tesi
d’Eddington, que té lloc quan la pressió de radiació contraresta la força d’atracció
gravitatòria. Com que la matèria que és acretada té moment angular, habitualment
es forma un disc d’acreció al voltant de l’objecte compacte. La matèria del disc
perd moment angular per dissipació viscosa, la qual cosa produeix un escalfament
del disc, i cau vers l’objecte compacte en una trajectòria espiral. La temperatura
de cos negre de la darrera òrbita estable, en el cas d’un forat negre acretant matèria
en el límit d’Eddington, ve donada per una dependència amb la massa de l’objecte
compacte. Si el forat negre té unes poques masses solars la temperatura obtinguda
és de l’ordre de 10 7 K. A aquesta temperatura l’energia és radiada en el domini de
raigs X de l’espectre electromagnètic. D’altra banda, podem calcular la lluminositat
d’acreció com l’energia cinètica de les partícules en arribar al radi gravitacional o
de Schwarzschild. Així, podem explicar les lluminositats típiques de les estrelles
binàries de raigs X, que són de l’ordre de 10 37 erg s −1 , amb ritmes d’acreció de
massa de l’ordre de 10 −9 M⊙ any −1 . Aquesta matèria prové de l’estrella companya
a l’objecte compacte. Si com a objecte compacte tenim una estrella de neutrons en
comptes d’un forat negre, els ritmes d’acreció de massa necessaris són lleugerament
superiors. En qualsevol cas, podem explicar la lluminositat en raigs X de les binàries
de raigs X amb ritmes d’acreció de massa superiors a 10 −10 M⊙ yr −1 .
Si l’objecte compacte és una estrella de neutrons amb un camp magnètic intens,
de l’ordre de 10 12 G, el disc d’acreció quedarà truncat a alguns milers de quilòmetres
de l’objecte compacte, ja que la matèria seguirà les línies de camp magnètic, i
acabarà impactant en els pols magnètics, produint altre cop emissió en raigs X. Si
l’eix magnètic està inclinat respecte l’eix de rotació, veurem pulsacions de raigs X
si l’emissió en forma de feix dels pols magnètics escombra la nostra visual. Aquest
tipus particular de binàries de raigs X s’anomenen púlsars de raigs X.
Per contra, si l’objecte compacte és una estrella de neutrons amb un camp
magnètic feble, aproximadament inferior a 10 10 G, el disc d’acreció pot arribar a
impactar l’objecte compacte. En aquest cas els raigs X seran produïts en la part
interna del disc d’acreció i en la capa límit entre el disc d’acreció i l’objecte compacte.
El nombre de binàries de raigs X conegudes actualment és de 280 aproximada-
ment. Depenent de la massa de l’estrella companya a l’objecte compacte, se solen
dividir en binàries de raigs X de massa alta i de massa baixa.
En les de massa alta, de les quals se’n coneixen unes 131, la companya és una
estrella jove de tipus espectral O o B, i típicament se’n troben de dos tipus: estrelles

Resum de la tesi ix
Be de la seqüència principal o gegants, i estrelles O o B supergegants. Les primeres
tenen masses d’entre 8 i 20 M⊙, són objectes amb una elevada rotació, i tenen un disc
de decreció equatorial, que proporciona la matèria a ser acretada per part de l’objecte
compacte. Típicament aquest és una estrella de neutrons amb un camp magnètic
intens, i aquests sistemes solen ser púlsars de raigs X. En el cas de les supergegants,
de més de 15 M⊙, aquestes tenen un vent estel·lar intens, que proporciona la matèria
a ser acretada. En alguns casos també s’omple l’anomenat lòbul de Roche, amb la
qual cosa la transferència de matèria és encara més eficient. Com que en ambdós
casos es tracta d’estrelles joves, aquests sistemes se situen principalment en el disc
galàctic, ja que tracen la població estel.lar de tipus I. És important destacar que el
57% de les binàries de raigs X de massa alta a la Galàxia són púlsars de raigs X.
En les de massa baixa, de les quals se’n coneixen unes 149, l’estrella companya
té un tipus espectral posterior a B, i típicament tenen masses inferiors a 2 M⊙. En
aquest cas la transferència té lloc per desbordament del lòbul de Roche. La majoria
són fonts transitòries de raigs X. Algunes d’elles es classifiquen com a fonts Z (6
sistemes) i fonts Atoló (18 sistemes), depenent del patró traçat en diagrames color-
color de raigs X. Es creu que les primeres contenen estrelles de neutrons amb camps
magnètics de l’ordre de 10 10 G i ritmes d’acreció propers al límit d’Eddington,
mentre que les segones tenen camps magnètics febles inferiors a 10 8 G i ritmes
d’acreció d’entre una dècima i una centèssima del límit d’Eddington. Les binàries
de raigs X de massa baixa se situen en el bulb i en l’halo de la Galàxia, ja que són
objectes vells de Població II. De fet, 13 sistemes es troben en cúmuls globulars.
D’altra banda, aproximadament 43 binàries de raigs X també mostren emissió
en la banda ràdio de l’espectre electromagnètic, amb densitats de flux superiors
a 0.1–1 mJy. A més, aquesta emissió és produïda en escales angulars petites, de
manera que no pot tenir un origen tèrmic. En aquest contexte, el mecanisme més
eficient per produir l’emissió ràdio observada és el mecanisme d’emissió sincrotró,
segons el qual electrons altament relativistes interaccionen amb camps magnètics
i produeixen emissió ràdio, un percentatge de la qual està linealment polaritzada.
Com en alguns sistemes, com en SS 433, l’emissió ràdio es va observar formant una
estructura allargada o de tipus jet, com en els quàsars i nuclis actius de galàxies,
es va proposar que fluxos d’electrons relativistes eren ejectats perpendicularment
al disc d’acreció, i eren responsables d’emissió ràdio d’origen sincrotró en presència
de camps magnètics. S’han proposat diversos models per explicar la formació dels
jets, tot i que no hi ha encara un model àmpliament acceptat. El que si sembla

x Resum de la tesi
imprescindible per la formació dels jets és la presència d’un disc d’acreció prou
proper a l’objecte compacte, ja que en els púlsars de raigs X no s’ha detectat mai
emissió ràdio d’origen sincrotró.
El percentatge de sistemes de massa alta que emeten en ràdio (8 objectes) és
aproximadament un 6%, mentre que en els sistemes de massa baixa (35 objectes)
és d’un 23%. Si només considerem els objectes de la Galàxia (excloent els sistemes
situats en els núvols de Magallanes) i excloem els púlsars de raigs X, aquests per-
centatges esdevenen 22% i 24%, respectivament. Per tant, veiem que d’entre les
binàries de raigs X que poden emetre en ràdio (excloent púlsars de raigs X) i on
aquesta pot ser detectada (objectes galàctics), ha estat així en un 20–25% dels casos
aproximadament, sense que hi influeixi la massa de l’estrella companya.
1.2 Microquàsars
Els microquàsars són estrelles binàries de raigs X amb emissió ràdio en forma de
jets relativistes. El nom va sorgir no només a partir de les similituds morfològiques
entre aquestes fonts i els quàsars, sino també degut a les similituds físiques, ja
que quan l’objecte compacte és un forat negre, paràmetres com la temperatura
de cos negre del disc d’acreció, les mides característiques dels jets o les escales
temporals són escalables a partir de la massa de l’objecte central. En aquest sentit,
els discs d’acreció dels quàsars emeten en l’òptic i en l’ultraviolat, mentre que en
els microquàsars ho fan en ragis X. D’altra banda, fenòmens que tenen lloc en
escales de temps d’anys en quàsars, poden ser estudiats en només alguns minuts en
microquàsars.
Actualment es coneixen uns 16 microquàsars, d’entre les 43 binàries de raigs X
amb emissió ràdio.
És interessant remarcar que alguns autors han proposat que
totes les binàries de raigs X amb emissió ràdio són microquàsars, i que serien vistes
com a tals si es disposés d’observacions amb prou sensibilitat i/o resolució angular.
L’emissió ràdio detectada en els jets dels microquàsars forma habitualment núvols
discrets de plasma, que es propaguen a través de l’espai amb velocitats relativistes.
Així doncs, cal tenir en compte efectes de la relativitat especial a l’hora d’analitzar
les observacions. Si considerem un parell de núvols de plasma ejectats perpendic-
ularment al disc d’acreció i a banda i banda de l’objecte compacte en el mateix

Resum de la tesi xi
instant de temps, i considerem que són intrínsecament iguals, les propietats obser-
vades d’un i altre dependran de l’angle que formi la nostra visual amb la direcció
dels jets. Així, es pot demostrar que el moviment propi mesurat en el pla del cel
per la component que s’acosti a nosaltres serà més elevat que no pas en el cas de
la component que s’allunyi de nosaltres. Un efecte interessant és l’anomenat movi-
ment superlumínic, que té lloc quan la velocitat del núvol és propera a la velocitat
de la llum, i la velocitat mesurada és aparentment més gran que la de la llum. Es
tracta d’un efecte de projecció, fàcilment explicable amb les equacions pertinents.
Es pot veure que podem mesurar un moviment superlumínic en una component
que s’apropi amb una velocitat intrínseca de 0.71 vegades la velocitat de la llum
i movent-se en una direcció que formi un angle de 45 ◦ respecte la nostra visual.
Per angles més grans o més petits, calen velocitats intrínseques més elevades per
detectar moviments superlumínics.
D’altra banda, també es pot demostrar que la densitat de flux observada en el
núvol que s’acosta serà també més elevada que no pas la del núvol que s’allunya.
En els casos extrems, quan l’angle que formi el jet amb la nostra visual sigui molt
petit, només detectarem l’emissió de la component que s’apropa, com succeeix en el
cas dels quàsars, en què el seu flux pot arribar a augmentar com el factor de Lorentz
elevat al cub.
Així doncs, mesurant els moviments propis i/o les densitats de flux observades
en les components que s’acosten i s’allunyen respecte l’observador en el cas d’un
microquàsar, es pot obtenir una cota mínima de la velocitat β (velocitat expressada
en unitats de la velocitat de la llum, β = v/c), i una cota màxima de l’angle del jet
respecte la visual, θ.
Sis dels 16 microquàsars coneguts són binàries de raigs X de massa alta, i els
deu restants ho són de massa baixa. La majoria de sistemes de massa alta són
persistents pel que fa a l’emissió ràdio, mentre que la majoria dels de massa baixa
son fonts transitòries. En 4 microquàsars s’han mesurat moviments superlumínics,
i existeixen indicis d’aquest fet en un cinquè cas. Pel que fa als objectes compactes,
és segur que en 6 d’ells es tracta d’un forat negre, mentre que se sospita de la seva
presència en uns altres 6 casos. En dos microquàsars l’objecte compacte és una
estrella de neutrons, i se sospita el mateix en una altres dos casos.
Com ja hem comentat, les escales de temps en els microquàsars són molt més
petites que en el cas dels quàsars. Així, ha estat possible detectar ràpides transicions

xii Resum de la tesi
dels sistemes realitzant observacions multi-longitud d’ona simultànies. Un exemple
són les observacions en escales de minuts de GRS 1915+105. En aquest sistema
s’ha pogut inferir que una disminució sobtada del flux de raigs X corresponia a
la desaparició de la part interna del disc d’acreció, a continuació de la qual es
produïa l’ejecció de plasma en forma de jet a velocitat relativista, ja que primerament
s’observava una erupció en l’infraroig que venia seguida d’una erupció en ràdio.
D’altra banda, observacions de gran abast temporal del sistema GX 339−4 han
permès estudiar la correlació entre el flux de raigs X tous (provinents del disc d’acre-
ció), el flux de raigs X durs (provinents d’una corona per sobre i per sota del disc)
i la densitat de flux en el domini ràdio (provinent del jet). S’ha pogut establir una
correlació un a un entre els raigs X durs i l’emissió en ràdio, i s’ha proposat que els
raigs X durs de la corona podrien ser de fet la base del jet relativista. Els raigs X
tous semblen traçar els canvis d’estat dels forats negres, de manera que quan el flux
augmenta molt (el dics arriba molt a prop de l’objecte compacte), tant l’emissió de
raigs X durs com l’emissió ràdio són suprimides, probablement degut a la desaparició
del jet en el sistema.
1.3 Motivació de la tesi
Tal com hem vist, actualment es coneixen al voltant de 16 microquàsars, tot i
que aquest nombre podria augmentar fins a 43 si tal i com han proposat diversos
autors totes les binàries de raigs X amb emissió ràdio contenen jets relativistes. De
tota manera, la situació encara no és clara. A més, hi ha molts tipus diferents de
binàries de raigs X amb emissió ràdio: d’alta i baixa massa, totes les fonts de tipus Z
i només 5 fonts de tipus Atoló en el cas de baixa massa, amb forats negres i estrelles
de neutrons com a objectes compactes, etc. Així doncs, quan comencem a dividir
el grup de binàries de raigs X que emeten en ràdio segons aquests diferents tipus,
no podem realitzar estudis estadísticament significatius. Aquesta situació empitjora
notablement si només considerem els 16 microquàsars.
A més, s’han proposat algunes correlacions, com per exemple que la velocitat
dels jets relativistes depèn de la massa de l’objecte compacte. De tota manera, hi
ha pocs casos on la velocitat del jet i la natura de l’objecte compacte siguin coneguts.
Així doncs, és impossible realitzar afirmacions d’aquest tipus a partir del nombre
limitat d’objectes coneguts.

Resum de la tesi xiii
D’altra banda, hi ha una pregunta important a contestar: de quin tipus és la
matèria que forma els jets? Fins ara només s’han detectat línies espectrals atòmiques
en el microquàsar SS 433, que indiquen que en aquest cas els jets son bariònics. De
totes maneres, alguns del models proposats per explicar els processos d’ejecció al
voltant d’objectes compactes només preveuen jets leptònics, formats per electrons i
positrons. Així doncs, és fonamental desvetllar el tipus de matèria dels jets per tal
de poder construir models adients.
Un altre aspecte interessant és la possible emissió de raigs gamma d’alta energia
en els microquàsars. Abans d’aquest treball, només el sistema LS I +61 303 podia ser
associat amb una d’aquestes fonts. Actualment, però, el candidat més prometedor és
el microquàsar LS 5039, com es veurà posteriorment. En qualsevol cas, un grup de
dos objectes és realment pobre, i val la pena realitzar un esforç per tal d’ampliar-lo.
Així doncs, van ser bàsicament dues les raons que van motivar l’inici d’aquest
projecte a llarg termini enfocat en la cerca de nous microquàsars a la Galàxia:
• Cada nou microquàsar ha permès nous i interessants descobriments fenomenològics,
que han de ser tinguts en compte pels models proposats per explicar els
fenòmens d’acreció/ejecció que tenen lloc al voltant d’objectes compactes.
• Augmentar el grup reduït de microquàsars coneguts i permetre estudis es-
tadístics en el futur.
En l’apartat 2 presentem el descobriment i subsegüent estudi del microquàsar
LS 5039. Aquesta va ser la primera font descoberta dins del mètode més global
presentat en l’apartat 3, on expliquem com s’han utilitzat els millors catàlegs de
raigs X i de ràdio per tal de buscar noves candidates a binàries de rais X amb emis-
sió ràdio/microquàsars. Finalment, llistem les nostres conclusions després d’aquest
treball en l’apartat 4.
2. Descobriment i estudi del microquàsar LS 5039
Els catàlegs disponibles en diferents rangs de l’espectre electromagnètic (bàsicament
raigs X, òptic i ràdio) proporcionen una eina fonamental en la cerca de fonts amb

xiv Resum de la tesi
un comportament multi-longitud d’ona conegut. Així doncs, com a primer pas en
la nostra cerca de nous microquàsars, vam inspeccionar en longituds d’ona ràdio
aquells objectes que havien estat classificats com a candidats a binàries de raigs X.
En aquest context, Motch et al. (1997) van dur a terme una identificació creua-
da entre el catàleg de raigs X ROSAT All Sky Survey (RASS) i les estrelles de
tipus espectrals O i B presents en la base de dades SIMBAD. Quasi bé 20 nous ob-
jectes candidats a estrelles binàries de raigs X de massa alta van ser proposats fruit
d’aquest treball. En particular, l’estrella LS 5039, situada a una distància de 3 kpc
aproximadament, mostrava un espectre de raigs X dur, compatible amb un sistema
binari on un forat negre o una estrella de neutrons acretés matèria proporcionada
pel vent de l’estrella LS 5039, de tipus espectral O7V((f)).
Per tal de trobar contrapartides ràdio a les estrelles proposades per Motch et al.
(1997) com a noves binàries de raigs X, vam inspeccionar les seves posicions en el
catàleg radio NRAO VLA Sky Survey (NVSS). Aquest catàleg cobreix el cel per
sobre d’una declinació de δ = −40 ◦ (82% de l’esfera celest) a una freqüència de
1.4 GHz (20 cm de longitud d’ona) utilitzant el Very Large Array (VLA) en les
seves configuracions més compactes, D i DC. Conté mes de 1.8 × 10 6 fonts amb
una densitat de flux superior a 2.5 mJy, el seu límit de completesa. L’error en la
posició de les fonts és inferior a un segon d’arc per les fonts amb una densitat de
flux superior a 15 mJy i de set segons d’arc per les fonts més febles detectades.
Com a resultat de la nostra cerca, vam descobrir una font ràdio de 24 mJy
coincident amb la posició de LS 5039, i catalogada com a NVSS J182614−145054
en el NVSS. A continuació passem a detallar l’estudi multi-longitud d’ona realitzat
d’aquest objecte.
2.1 Estudi multi-longitud d’ona de LS 5039
Per tal de confirmar l’associació entre la font de raigs X, l’objecte òptic i la font
ràdio, vam realitzar observacions amb l’interferòmetre VLA en la seva configuració
A (la més estesa i que proporciona una millor resolució angular). Les observacions
van ser realitzades en 4 èpoques diferents entre febrer i maig de 1998 i a 4 freqüències
diferents, per tal d’obtenir informació sobre la variabilitat i l’espectre ràdio, respec-
tivament. Com a resultat, vam poder confirmar que les posicions ràdio i òptiques

Resum de la tesi xv
concordaven entre elles, amb una diferència inferior a 0.1 segons d’arc. La probabil-
itat d’una associació errònia per coincidència aleatòria era inferior a 10 −7 . D’altra
banda, l’espectre obtingut, que patia lleugeres variacions al llarg del temps, podia
ajustar-se amb un índex espectral de l’ordre de α −0.5 (on la densitat de flux
ve donada per Sν ∝ ν +α ). L’elevada temperatura de brillantor, així com l’espectre
observat, suggerien emissió ràdio d’origen sincrotró. L’ús d’equacions d’equipartició
proporcionava una estimació de l’energia màxima dels electrons, E ≤ 4 × 10 41 erg,
i del camp magnètic mínim, B ≥ 0.01 G. Així doncs, vam concloure (Martí et al.
1998) que es tractava d’una nova binària de raigs X de massa alta amb emissió ràdio
d’origen sincrotró.
A continuació vam demanar que aquest objecte fos observat regularment dins
la campanya GBI-NASA Monitoring Program (GBI), que proporcionava mesures
de flux ràdio a dues freqüències (2.25 i 8.3 GHz). Es van obtenir dades durant un
primer període d’onze mesos (setembre 1998–agost 1999) i un segon període d’un
sol mes (setembre 2000). L’anàlisi d’aquestes dades (preliminar a Ribó et al. 1999,
i complet aquí) mostra que la font és persistent, moderadament variable i amb un
índex espectral α −0.5, com s’havia vist anteriorment. D’altra banda, una anàlisi
temporal de les dades mostra que no hi apareix cap senyal periòdica, un fet esperable
tenint en compte la relació senyal-soroll relativament baixa de les dades.
El pas següent per veure si es tractava d’un microquàsar va consistir en realitzar
observacions d’interferometria de molt llarga base (Very Long Baseline Interferom-
etry, VLBI) amb l’interferòmetre Very Long Baseline Array (VLBA). Les observa-
cions es van realitzar el 8 de maig de 1999, durant 4 hores i a una freqüència de
5 GHz, en el mode d’observació v6cm-256-8-2-L, que proporciona un ample de ban-
da de 64 MHz. La imatge resultant d’aquestes observacions es mostra en la Fig. 1,
on es pot veure emissió en forma de jet bipolar sorgint a banda i banda d’una font
central intensa. Com es pot observar, el jet que apunta cap al sud-est (inferior es-
querra) és més brillant i es troba a una distància major respecte el centre que no pas
el jet nord-oest (superior dret). Aquesta asimetria pot ser interpretada segons els
efectes relativistes mencionats anteriorment, que impliquen una velocitat intrínseca
mínima de 0.15c (on c és la velocitat de la llum) pel plasma en els jets, i un angle
màxim de posició dels jets respecte la visual de 81 ◦ . Assumint equipartició, s’obté
una estimació de l’energia dels electrons, E ∼ 5 × 10 39 erg, i el camp magnètic,
B ∼ 0.2 G. Aquests resultats, confirmaven que LS 5039 era un nou microquàsar a
la Galàxia (Paredes et al. 2000).

xvi Resum de la tesi
MilliARC SEC
4
2
0
-2
-4
-6
VLBA
6 4 2 0 -2 -4 -6
MilliARC SEC
Figura 1: Imatge autocalibrada de LS 5039 obtinguda amb el VLBA a 5 GHz.
A partir d’aleshores diversos grups van començar a estudiar aquest objecte,
bàsicament en longituds d’ona òptiques, i es van obtenir resultats com el període
orbital o una classificació espectral més acurada, tal i com s’explica més endavant.
Com que havíem resolt els jets de LS 5039 en un moment d’emissió persistent
en ràdio, sense que hi hagués hagut cap erupció prèvia en raigs X o en ràdio, era
d’esperar que els jets fossin sempre presents en aquest microquàsar. Per tal de
confirmar-ho, vam realitzar observacions amb la xarxa europea de VLBI (European
VLBI Network, EVN) i la xarxa de sis antenes angleses connectades (Multi-Element
Radio-Linked Interferometer Network, MERLIN). Les observacions es van realitzar
l’1 de març de 2000, utilitzant el mateix mode d’observació que l’emprat amb el
VLBA. Els resultats confirmen la presència persistent de jets relativistes a LS 5039,
que s’estenen des de ∼ 10 a ∼ 1000 unitats astronòmiques (Paredes et al. 2002a).
A més també s’observen les asimetries degudes als efectes relativistes, la qual cosa
permet confirmar els resultats anteriors pel que fa als limits de la velocitat mínima

Resum de la tesi xvii
del plasma i l’angle màxim dels jets respecte la visual.
Poc després del descobriment de l’emissió ràdio de LS 5039, vam realitzar ob-
servacions fotomètriques en l’òptic amb el telescopi de 1.5 m de l’Observatorio As-
tronómico Nacional (OAN), el juny de 1998, que van permetre detectar variacions
de l’ordre de 0.05 magnituds en intervals d’un dia, amb un valor mitjà de V = 11.36
(Martí et al. 1998).
Posteriorment, Clark et al. (2001) van realitzar observacions fotomètriques i
espectroscòpiques, que van demostrar que LS 5039 patia variacions fotomètriques
molt moderades en l’òptic (0.1 magnituds) i més apreciables en l’infraroig prop-
er (0.5 magnituds) en escales temporals d’anys. Gràcies a les observacions espec-
troscòpiques, aquests mateixos autors van concloure que el tipus espectral de l’estrel-
la era O6.5V((f)). Amb aquesta nova classificació i utilitzant noves calibracions per
estrelles joves, nosaltres hem deduït una distància a LS 5039 de 2.9 ± 0.3 kpc (Ribó
et al. 2002a).
D’altra banda, McSwain et al. (2001) van obtenir la corba de velocitat radial
del sistema binari, i van deduir els següents paràmetres orbitals: un període de 4.12
dies, una velocitat radial del sistema de 4.6 km s −1 , una amplitud de variació de
la velocitat radial de 15 km s −1 , una excentricitat de 0.41, una funció de massa de
0.0010 M⊙ i una velocitat rotacional projectada de 131 km s −1 .
Per tal de confirmar aquests resultats, hem realitzat noves observacions espec-
troscòpiques amb el telescopi de 2.5 m Isaac Newton Telescope (INT), del 23 al 31
de juliol de 2002. Aquestes observacions recents confirmen els valors del període
orbital i la velocitat rotacional projectada (Casares et al. 2002).
Pel que fa a l’emissió en raigs X de LS 5039, hem dut a terme una anàlisi dels
sis anys i mig de dades del seguiment realitzat amb el All Sky Monitoring a bord del
satèl·lit Rossi X-ray Timing Explorer (ASM/RXTE) de la contrapartida en raigs X,
anomenada RX J1826.2−1450. Els resultats, igual que en les anàlisis de dos anys
i mig de dades presentades a Ribó et al. (1999), indiquen que no hi ha variabilitat
periòdica en aquestes dades. En particular no trobem el període orbital obtingut
espectroscòpicament. Tenint en compte l’excentricitat de l’òrbita, serien d’esperar
variacions del ritme d’acreció de massa per part de l’objecte compacte amb el període
orbital, i per tant també de la lluminositat en raigs X d’aquest sistema binari. De
tota manera, la relació senyal-soroll de les dades segurament no és prou bona com

xviii Resum de la tesi
per poder detectar aquestes variacions, que també s’haurien d’observar en el domini
ràdio, i no ha estat així en les observacions GBI, tot i que com ja s’ha dit també
tenen una relació senyal-soroll força baixa.
Per tal d’estudiar l’emissió en raigs X amb cura, vam realitzar observacions
amb l’instrument Proportional Counter Array del satèl·lit RXTE (PCA/RXTE), el
febrer de 1998, que van permetre mesurar un espectre de raigs X significativament
dur fins a 30 keV, on destacava la presència d’una línia de ferro a una energia
de 6.6 keV. D’altra banda, no s’observaven pulsacions (com era d’esperar ja que el
sistema presentava emissió ràdio) ni variacions quasi-periòdiques en escales de temps
de 0.02 a 2000 segons. La lluminositat obtinguda a una distància de 2.9 kpc és de
5 × 10 34 erg s −1 , relativament baixa per una binària de raigs X (Ribó et al. 1999).
Posteriorment vam realitzar noves observacions amb el satèl·lit de raigs X Bep-
poSAX l’octubre de 2000. Aquestes observacions van indicar una devallada de la llu-
minositat en quasi bé un ordre de magnitud respecte les observacions PCA/RXTE.
Això pot ser degut a la variació del vent estel·lar de LS 5039, i per tant del seu ritme
de pèrdua de massa, i en conseqüència del ritme d’acreció de matèria per part de l’ob-
jecte compacte. A més, aquestes observacions van cobrir el rang de fases orbitals del
sistema on s’esperaria trobar un eclipsi de raigs X, sense que aquest hagi estat detec-
tat. Aquest fet, dóna suport a la suggerència que el sistema binari té una inclinació
força baixa, al voltant de 30 ◦ , segons semblen indicar observacions d’espectroscopia
òptica recents (McSwain & Gies 2002). Finalment, hem pogut obtenir un valor per
la densitat columnar d’hidrogen a l’objecte, que és de NH = 1.0 ± 0.3 × 10 22 cm −2
(Reig et al. 2002).
Poc després del descobriment de la natura microquàsar de LS 5039, vam in-
speccionar els catàlegs de fonts d’altes energies per tal de buscar-hi una possible
contrapartida. Com a resultat, vam descobrir la presència d’una font de raigs gam-
ma d’alta energia no identificada en una posició concordant amb LS 5039. A més, el
microquàsar és l’única font de raigs X brillant, i l’única font de raigs X que presenta
emissió ràdio, en el camp de la font gamma, anomenada 3EG J1824−1514 segons
les coordenades en el tercer catàleg del detector EGRET. D’altra banda, tant l’e-
missió ràdio de LS 5039 com l’emissió gamma de 3EG J1824−1514 són persistents
i moderadament variables al llarg del temps. A més, una anàlisi de la variabilitat
gamma sembla excloure un púlsar o un remanent de supernova com a origen de la
font gamma. Finalment, l’espectre en raigs gamma no concorda amb els que habit-

Resum de la tesi xix
ualment es troben en els púlsars de raigs gamma, úniques fonts EGRET d’origen
galàctic identificades fins el moment, excloent una erupció solar. Tots aquests fets
suggereixen una associació entre els dos objectes, com vam proposar a Paredes et al.
(2000). A més de la possibilitat de detectar els microquàsars en raigs gamma d’alta
energia, és segurament més interessant el fet que hi ha prop de 170 fonts EGRET no
identificades, 72 d’elles a menys de 10 ◦ del pla galàctic. Així doncs, hem proposat
que algunes d’aquestes fonts podrien ser microquàsars persistents i “silenciosos”,
com LS 5039, que encara han de ser descoberts.
Per tal d’explicar el comportament multi-longitud d’ona de LS 5039 des del
domini ràdio fins als raigs gamma, hem proposat un model que explica l’emissió
gamma com a resultat de l’efecte Compton invers que pateixen els fotons ultraviolats
de l’estrella per part dels electrons relativistes del jet. L’elevat camp de radiació de
fotons ultraviolats fa de fet inevitable la producció de raigs gamma degut a l’efecte
Compton invers si tenim en compte la presència d’electrons altament relativistes
(amb factors de Lorentz de l’ordre de 10 3 –10 4 ), que posteriorment, i a mida que
s’allunyen del camp de radiació, donaran lloc a emissió ràdio d’origen sincrotró. A
més, el model suggereix que el jet és força col·limat fins a les escales detectades de
∼ 1000 unitats astronòmiques (Paredes et al. 2002a).
Finalment, hem realitzat algunes consideracions energètiques i hem suggerit ob-
servacions que es podrien dur a terme en el futur, per tal de poder millorar el
model proposat i entendre millor els fenòmens d’acreció/ejecció en el cas d’aquest
microquàsar.
2.2 LS 5039 com un microquàsar en fuga
La formació d’un objecte compacte en una estrella binària de raigs X, requereix
necessàriament que hagi tingut lloc una explosió de supernova que no hagi trencat
el sistema binari.
És d’esperar que aquest esdeveniment explosiu canviï consider-
ablement les propietats cinemàtiques del sistema.
Per tal de comprovar quina és la situació en el cas del microquàsar LS 5039 hem
recopilat set posicions òptiques, que cobreixen quasi bé cent anys de temps, i dues
posicions ràdio, que només cobreixen dos anys de temps. Hem realitzat ajustos de
moviments propis tant en les posicions òptiques com en les ràdio, i hem obtingut

xx Resum de la tesi
valors força similars en ambdós casos. Així, podem dir que basats només en dades
astromètriques som també capaços de confirmar que ambdues emissions, en l’òptic
i en ràdio, s’originen en el mateix objecte.
Utilitzant totes les dades conjuntament, hem obtingut els següents moviments
propis: µα cos δ = (4.7 ± 1.1) × 10 −3 ′′ any −1 , µδ = (−10.6 ± 1.0) × 10 −3 ′′ any −1 .
Aquests, juntament amb la nova estimació de la distància de 2.9 ± 0.3 kpc i la
velocitat radial del sistema de Vr = 4.6 ± 0.5 km s −1 (McSwain et al. 2001), han
permès calcular una velocitat espacial de (U = 51, V = −71, W = −118) km s −1
respecte el seu entorn local (Regional Standard of Rest). Aquest resultat implica
que LS 5039 és un microquàsar en fuga (runaway), amb una velocitat sistèmica de
vsis 150 km s −1 , escapant del seu entorn local amb una elevada component de la
velocitat perpendicular al pla galàctic. Això és probablement degut a l’explosió de
supernova que va crear l’objecte compacte en aquest sistema binari.
Hem calculat la trajectòria de LS 5039 en el passat, i trobat dues associacions OB
properes al seu camí traçat en el pla del cel. De totes maneres, aquestes associacions
OB són massa properes a nosaltres per estar relacionades amb el microquàsar. D’al-
tra banda, també hem trobat tres remanents de supernova a prop del camí traçat
per LS 5039. Després de descartar dues d’elles basats en les distàncies que ens en
separen, hem centrat la nostra atenció en el remanent SNR G016.8−01.1. L’estudi
d’aquesta font no ha pogut ni confirmar ni refusar una possible associació amb el
microquàsar, bàsicament per les incerteses en l’estimació de la densitat de flux ràdio
del remanent. Aquest fet segurament justifica noves observacions d’alta sensibilitat
per tal de buscar remanents dèbils en aquesta regió del cel.
D’altra banda, també hem trobat una cavitat semi-oberta de H i a prop de la
posició de LS 5039. Encara que l’estrella O((f)) en aquest microquàsar sembla ser
el creador d’aquesta bombolla, no podem descartar una contribució de l’estrella
progenitora de l’objecte compacte.
Basats en càlculs teòrics, hem estat capaços d’explicar tant la velocitat espa-
cial elevada com l’excentricitat observades en un escenari d’explosió de supernova
simètrica, amb una important pèrdua de massa de ∆ M ∼ 17 M⊙.
Finalment, la velocitat espacial elevada i el possible temps de vida del micro-
quàsar indiquen que aquest podria assolir una latitud galàctica de b = −12 ◦ . Per
tant, si l’associació entre LS 5039 i la font EGRET 3EG J1824−1514 és correcta,

Resum de la tesi xxi
podríem detectar microquàsars de raigs gamma fins a latituds galàctiques de l’ordre
de |b| 10 ◦ . En particular, els microquàsars en fuga podrien estar relacionats amb
algunes de les fonts febles, variables i relativament toves encara no identificades del
tercer catàleg EGRET, situades per sobre i per sota del pla galàctic. Tots aquests
resultats van ser publicats a Ribó et al. (2002a).
3. Una cerca de nous microquàsars
Com ja hem comentat anteriorment, el reduït nombre de microquàsars coneguts ens
va donar un bon motiu per començar un projecte a llarg termini per tal de cercar
nous microquàsars.
3.1 Mètode d’identificació creuada
Per tal de descobrir nous microquàsars d’una manera sistemàtica, hem realitzat una
identificació creuada entre els millors catàlegs de raigs X i ràdio disponibles.
Pel que fa als raigs X, hem utilitzat el catàleg de fonts brillants del satèl·lit
ROSAT (ROSAT All Sky Brigth Source Catalog, RBSC), que conté en la seva versió
final un total de 18.806 fonts detectades en el rang d’energia entre 0.1 i 2.4 keV, i
que és un extracte del ROSAT All Sky Survey (RASS). En el RBSC hi ha informació
en quatre bandes d’energia expressades en keV: A=0.1–0.4, B=0.5–2.0, C=0.5–0.9,
D=0.9–2.0. A partir d’aquestes bandes, es calculen dos índexs de duresa: HR1 =
(B − A)/(B + A), HR2 = (D − C)/(D + C). Els errors a 1σ en la posició de les
fonts són típicament de 10–20 ′′ . Pel que fa al domini ràdio, hem emprat el catàleg
NVSS, que ja ha estat descrit en l’apartat 2.
Com que el nostre objectiu era maximitzar la probabilitat de retenir només les
estrelles binàries de raigs X amb emissió ràdio, hem adoptat els següents criteris de
selecció:
1. Hem seleccionat les fonts amb latituds galàctiques |b| < 5 ◦ del catàleg RBSC.
Com que el NVSS té un limit de δ > −40 ◦ , només les fonts per sobre d’aquesta
declinació han estat considerades. Així doncs, hem cobert aproximadament el
75% de l’àrea |b| < 5 ◦ (l 347–259 ◦ i α 17.2–8.6 h ).

xxii Resum de la tesi
2. D’entre les fonts RBSC restants, les que tenien indicadors de fonts properes que
contaminaven les mesures o indicadors de problemes amb les determinacions
de les posicions, han estat excloses.
3. A partir d’estudis estadístics del RBSC, Motch et al. (1998) van concloure
que les binàries de raigs X eren fàcilment identificables pel seu espectre dur
en raigs X. Per tant, en un intent d’evitar nuclis actius de galàxies i variables
cataclísmiques, només hem seleccionat aquelles fonts que tenien un índex de
duresa HR1 + σ(HR1) major que 0.9. Encara que aquest criteri podria evitar
la detecció de microquàsars en erupció, cal notar que aquesta situació dura
un temps molt curt en la vida d’aquests objectes, i que la probabilitat de
descartar-ne algun per aquest motiu és molt baixa. Un total de 241 fonts del
RBSC romanen en la nostra mostra en aquesta fase del procés.
4. Les posicions en raigs X i en ràdio han de concordar dins dels errors. Així,
hem seleccionat les fonts NVSS dins de caixes d’error 2σ (probabilitat del
95%) de les posicions de les fonts RBSC. També hem utilitzat la restricció
d’una diferència màxima de 40 ′′ entre les posicions RBSC i NVSS, ja que per
distàncies majors, no s’espera cap identificació creïble.
5. No s’ha permès continuar en el procés de selecció cap font ràdio resolta, ja que
qualsevol binària de raigs X amb emissió ràdio hauria d’aparèixer compacta
vista amb la resolució del NVSS 1 .
Les fonts així seleccionades, un total de 35, han estat filtrades amb informació
complementària en l’òptic, utilitzant els següents criteris:
1. Hem inspeccionat les bases de dades SIMBAD i NASA/IPAC Extragalactic
Database (NED) per cada font, i hem exclòs aquelles llistades com a objectes
extragalàctics.
2. També hem inspeccionat les imatges del Digitized Sky Survey, DSS1 i DSS2-
red (la primera versió i les plaques en filtre R de la segona versió), i hem
buscat possibles contrapartides òptiques compatibles amb les posicions de les
fonts NVSS. Els candidats pels quals hem trobat un objecte òptic extès, és a
dir, amb aparença de galàxia, també han estat exclosos de la mostra.
1 Les fonts resoltes únicament en un dels dos eixos i amb una mida angular inferior a la resolució
de l’altre eix no han estat excloses.

Resum de la tesi xxiii
Al final d’aquest procés, hem obtingut un total de 17 fonts emissores en raigs
X i en ràdio. Entre elles, hem trobat la ràdio binària de raigs X LS I +61 303 i
els microquàsars àmpliament coneguts LS 5039, SS 433 i Cyg X-3.
És interessant
comentar que totes aquestes fonts són binàries de raigs X de massa alta, i que cap
dels microquàsars coneguts amb companyes de massa baixa no ha estat recuperat
per aquest mètode. Això és fàcilment explicable pel fet que la majoria de ràdio
binàries X de massa baixa són fonts transitòries, i que les persistents són massa
dèbils com per ser presents en el NVSS. A més, hem recuperat tots els microquàsars
amb companyes massives presents en el rang de latituds galàctiques explorat, és a
dir, LS 5039, SS 433 and Cyg X-3, excepte Cyg X-1. La no detecció de Cyg X-1 pel
nostre mètode és deguda a la baixa densitat de flux d’aquesta font en l’època en que
es va realitzar el NVSS. Tot això fa que tinguem confiança en el mètode utilitzat.
Un cop excloses les fonts ja conegudes, hem separat les 13 fonts restants en dos
grups. En el primer grup, format per 8 fonts prioritàries, les diferències de posició
entre el RBSC i el NVSS són inferiors a la incertesa 1σ de la posició RBSC. En el
segon grup, format per 5 fonts, aquestes diferències estan en el rang 1–2σ.
3.2 Observacions òptiques i ràdio
Després de realitzar la identificació creuada de catàlegs presentada en l’apartat an-
terior, el següent pas lògic era obtenir posicions ràdio acurades per tal de buscar
possibles contrapartides òptiques. Amb aquesta finalitat, vam realitzar observacions
ràdio amb el VLA en configuració A en tres èpoques diferents, separades una set-
mana entre elles, el juliol de 1999, i a dues longituds d’ona (3.6 i 6 cm) per tal
d’obtenir informació espectral. Les fonts a observar eren les 8 prioritàries, tot i que
degut a condicions de visibilitat, només vam poder observar les 6 primeres, és a
dir, 1RXS J001442.2+580201, 1RXS J013106.4+612035, 1RXS J042201.0+485610,
1RXS J062148.1+174736, 1RXS J072259.5−073131 i 1RXS J072418.3−071508. Per
totes elles vam obtenir posicions ràdio acurades fins a 0.01 ′′ . D’altra banda, totes
les fonts apareixien puntuals excepte 1RXS J072259.5−073131, que mostrava indicis
d’un jet cap a l’est a escales de segon d’arc.
En el domini òptic, vam realitzar observacions fotomètriques del 24 al 30 de
novembre de 1998 amb el telescopi de 1.5 m de l’OAN, i observacions astromètriques
l’onze de desembre de 1999 amb el telescopi de 2.2 m del Centro Astronómico His-

xxiv Resum de la tesi
pano Alemán (CAHA). Com a resultat d’aquestes observacions, vam obtenir posi-
cions acurades, amb errors a 1σ de 0.3 ′′ , i fotometria en els filtres V , R i I de
Johnson, per cadascun dels objectes.
En la Fig. 2 mostrem les imatges obtingudes amb el telescopi de 2.2 m del
CAHA amb el filtre I de Johnson i les imatges ràdio resultants de concatenar totes
les observacions a 3.6 cm de longitud d’ona per cadascun dels objectes. Com es
pot observar, vam poder trobar contrapartides òptiques de totes les fonts ràdio. De
fet, una anàlisi estadística mostra que la probabilitat d’associació per coincidència
aleatòria es inferior al 0.3% en tots els casos.
Després d’un estudi exhaustiu de les dades obtingudes, concloem que els dos
primers objectes, és a dir, 1RXS J001442.2+580201 i 1RXS J013106.4+612035, són
bons candidats a binàries de raigs X amb emissió ràdio. D’altra banda, l’objecte
1RXS J042201.0+485610 no és tan bon candidat, ja que presenta un índex espec-
tral de fins α = +1.6 a freqüències radio elevades, fet insòlit en una ràdio binària
X, i a més a més, la seva contrapartida òptica és lleugerament estesa (15% més
que estrelles de la mateixa magnitud en la imatge CAHA), la qual cosa suggereix
un possible origen extragalàctic. Un estudi acurat de la contrapartida òptica de
1RXS J062148.1+174736, mostra que és un 30% més estesa que estrelles de magni-
tud similar en la mateixa imatge, apuntant cap a un origen extragalàctic per aquest
objecte. La situació no és gens clara en el cas de 1RXS J072259.5−073131, perquè
les dades òptiques són compatibles amb una natura de ràdio binària X, mentre que
el jet ràdio cap a un sol costat a escales de segon d’arc (que no es mostra en la
Fig. 2) és més comú en objectes extragalàctics. Finalment, la darrera font observa-
da, 1RXS J072418.3−071508, és un quàsar identificat amb posterioritat a la nostra
identificació creuada de catàlegs. Així doncs, després d’aquest estudi, dos objectes
són bons candidats a ràdio binàries de raigs X. Tots aquests resultats, juntament
amb els de l’apartat 3.1, han estat publicats a Paredes et al. (2002b).
El següent pas lògic era l’obtenció d’espectres òptiques d’aquestes fonts, per tal
de desvetllar la seva natura galàctica o extragalàctica. Es van dur a terme diversos
intents durant els anys 2000 i 2001, però les condicions meteorològiques durant les
observacions van impedir l’obtenció de resultats. Un nou intent està programat
per la tardor de 2002. D’altra banda, observacions VLBI d’aquestes sis fonts seran
publicades properament (Ribó et al. 2002b), i es presenten en el següent apartat.

Resum de la tesi xxv
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
1RXS J001442.2+580201 I band
58 03 00
02 45
30
15
00
01 45
30
15
58 02 02.0
00 14 48 46 44 42 40 38 36
RIGHT ASCENSION (J2000)
01.5
01.0
00.5
VLA 3.6cm
00 14 42.25 42.20 42.15 42.10 42.05 42.00
RIGHT ASCENSION (J2000)
17 48 30
15
00
47 45
30
15
00
46 45
17 47 36.0
35.5
35.0
34.5
1RXS J062148.1+174736 I band
06 21 51 50 49 48 47 46 45 44
RIGHT ASCENSION (J2000)
VLA 3.6cm
06 21 47.82 47.80 47.78 47.76 47.74 47.72 47.70 47.68
RIGHT ASCENSION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
1RXS J013106.4+612035 I band
61 21 30
15
00
20 45
30
15
00
19 45
61 20 34.0
01 31 14 12 10 08 06 04 02 00
RIGHT ASCENSION (J2000)
33.5
33.0
32.5
-07 30 45
VLA 3.6cm
01 31 07.35 07.30 07.25 07.20 07.15 07.10
RIGHT ASCENSION (J2000)
31 00
15
30
45
32 00
15
30
-07 31 34.0
1RXS J072259.5-073131 I band
07 23 03 02 01 00 22 59 58 57 56
RIGHT ASCENSION (J2000)
34.5
35.0
35.5
VLA 3.6cm
07 22 59.74 59.72 59.70 59.68 59.66 59.64 59.62
RIGHT ASCENSION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
48 57 00
56 45
30
15
00
55 45
30
15
1RXS J042201.0+485610 I band
04 22 06 04 02 00 21 58 56
RIGHT ASCENSION (J2000)
48 56 04.5
04.0
03.5
03.0
-07 14 30
45
15 00
15
30
45
16 00
15
VLA 3.6cm
04 22 00.60 00.55 00.50 00.45
RIGHT ASCENSION (J2000)
1RXS J072418.3-071508 I band
07 24 21 20 19 18 17 16 15 14
RIGHT ASCENSION (J2000)
-07 15 19.5
20.0
20.5
21.0
VLA 3.6cm
07 24 17.36 17.34 17.32 17.30 17.28 17.26 17.24 17.22
RIGHT ASCENSION (J2000)
Figura 2: Imatges òptiques en filtre I i imatges ràdio a 3.6 cm, agrupades per parelles,
de les sis fonts estudiades. Els cercles en les imatges òptiques, que no són caixes d’error,
indiquen les contrapartides òptiques de les fonts ràdio. Les creus en les imatges ràdio,
indiquen els errors 1σ de les posicions de les contrapartides òptiques.

xxvi Resum de la tesi
3.3 Observacions amb EVN i MERLIN
Per tal de detectar possibles estructures en forma de jets ràdio a escales angulars
de mil·lèssimes de segon d’arc, vam dur a terme observacions VLBI de les sis fonts
estudiades anteriorment. Les observacions es van dur a terme amb la EVN i MER-
LIN, del 29 de febrer de 2000 a l’1 de març de 2000 (23:30–23:00 de temps universal)
a una freqüència de 5 GHz. També es van dur a terme mesures de la densitat de
flux de les fonts amb l’antena de 100 metres de diàmetre que el Max Planck Institut
für Radioastronomie té a Effelsberg, Alemanya.
Les observacions es van realitzar utilitzant la tècnica de referència de fase, im-
prescindible per poder detectar les fonts més febles. Es van detectar 5 de les 6
fonts estudiades (totes excepte 1RXS J042201.0+485610) i es van obtenir imatges
amb MERLIN, amb la EVN i combinant els dos dispositius d’antenes. Aquestes
imatges es mostren en la Fig. 3, on en les imatges EVN+MERLIN hi ha indicats els
coeficients de “disminució” (tapering) utilitzats en aquests casos.
Tal i com es pot veure, la primera font, 1RXS J001442.2+580201, mostra un jet
bipolar que emana d’una font central en la imatge EVN, i on hi ha indicats els noms
de les components nord i sud, numerades segons l’ordre en que van ser ejectades des
de l’objecte compacte. Una anàlisi detallada en el marc de la relativitat especial
mostra que el jet te una velocitat superior a 0.2c, i forma un angle inferior a 78 ◦
respecte la visual.
La segona font, 1RXS J013106.4+612035, apareix bàsicament compacta, tot i
que hi ha indicis d’un jet relativista en un únic sentit, amb velocitat superior a 0.3c,
i un angle inferior a 72 ◦ respecte la visual, tal i com es pot deduir de la imatge EVN.
La tercera font, 1RXS J042201.0+485610, no va ser detectada, per problemes
amb la referència de fase i degut a que és una font molt dèbil.
La quarta font, 1RXS J062148.1+174736, apareix compacta en totes les imatges
(els allargaments visibles són deguts a la convolució amb el feix sintetitzat, que es
mostra en la part inferior esquerra de cada imatge).
La cinquena font, 1RXS J072259.5−073131, presenta un jet en un únic sentit que
implica una velocitat superior a 0.3c i un angle respecte la visual inferior a 73 ◦ . És
remarcable el torcement del jet a mida que s’allunya de la font central, una propietat

Resum de la tesi xxvii
1RXS J001442.2+580201
1RXS J013106.4+612035
1RXS J062148.1+174736
1RXS J072259.5-073131
1RXS J072418.3-071508
Figura 3: Imatges a 5 GHz de les cinc fonts detectades, obtingudes amb MERLIN,
MERLIN+EVN i amb la EVN. Les unitats dels eixos són mil·lèssimes de segon d’arc.
N1
N2
S2
S1

xxviii Resum de la tesi
habitual en objectes extragalàctics com els blàzars.
Finalment, el quàsar 1RXS J072418.3−071508, mostra també un jet tort i en un
únic sentit, que implica una velocitat superior a 0.5c i un angle respecte la visual
inferior a 60 ◦ .
3.4 Sumari
Després d’una anàlisi detallada dels resultats obtinguts, concloem que les dues
primeres fonts de la mostra, 1RXS J001442.2+580201 i 1RXS J013106.4+612035,
són bons candidats a microquàsar. La tercera font, 1RXS J042201.0+485610, no ho
és degut al seu espectre ràdio invertit a altes freqüències, que suggereix un origen
tèrmic per aquesta font, que a més a més apareix lleugerament estesa en l’òptic i
no es detecta en observacions VLBI. La quarta font, 1RXS J062148.1+174736, és
probablement una galàxia, ja que la contrapartida òptica és estesa. La cinquena
font, 1RXS J072259.5−073131, és probablement extragalàctica de tipus blàzar, ja
que el torcement del seu jet ràdio és una propietat comuna en aquests objectes. Fi-
nalment, la sisena font, 1RXS J072418.3−071508, és un quàsar ja identificat. Tots
aquests resultats, juntament amb els de l’apartat 3.3, han estat publicats a Ribó
et al. (2002b).
Cap d’aquestes sis fonts apareix en el tercer catàleg EGRET, de fonts de raigs
gamma d’alta energia. D’altra banda, recordem que un nou intent per tal d’obtenir
espectres òptics, que revelaran la natura galàctica o extragalàctica dels objectes, es
durà a terme la tardor de 2002.
Finalment les dues fonts prioritàries de la identificació creuada que no han estat
posteriorment observades, així com les fonts no prioritàries, s’estudiaran possible-
ment en el futur.
4. Conclusions generals
Aquí resumim breument les conclusions a les quals hem arribat després de realitzar
aquest treball.

Resum de la tesi xxix
Apartat 2:
1. Després de realitzar una acurada inspecció de modernes bases de dades i dur
a terme observacions ràdio interferomètriques, hem descobert el microquàsar
LS 5039. Per tant, concloem que estudis similars poden revelar una població
anteriorment inadvertida de microquàsars “silenciosos”, que fins ara no han
presentat cap explosió que ens fes saber de la seva existència, com LS 5039.
Si això és correcte, el fenomen dels microquàsars podria ser molt més habitual
del que fins ara s’havia considerat.
2. Hem realitzat un estudi en profunditat del comportament multi-longitud d’ona
de LS 5039, cobrint des de longituds d’ona ràdio fins a raigs gamma d’alta
energia. Per tal d’arribar a aquest fi, hem dut a terme observacions amb
l’objectiu d’estudiar el flux, la variabilitat d’aquest i l’espectre en longituds
d’ona ràdio, òptiques i en raigs X. En el domini ràdio hem obtingut imatges
interferomètriques en el continuu i mapes en línies espectrals realitzats amb
antena única. Totes aquestes dades, juntament amb observacions en raigs
gamma presents en la literatura, ens han permès proposar un escenari i un
model per tal d’explicar la fenomenologia observada en LS 5039. Els estudis
multi-longitud d’ona són claus per entendre els processos d’acreció/ejecció de
matèria al voltant d’objectes compactes.
3. Hem proposat una associació entre LS 5039 i una de les ∼ 170 fonts EGRET
no identificades, associació que sembla plausible degut a l’emissió persistent
i moderadament variable de la font tant en ràdio com en raigs gamma. Si
es confirma, aquesta seria la primera associació entre un microquàsar i una
font EGRET, i suggerim que altres fonts de raigs gamma d’alta energia no
identificades podrien ser microquàsars “silenciosos” encara no descoberts.
4. Hem descobert que LS 5039 és un microquàsar en fuga, amb una velocitat
sistèmica de vsis 150 km s −1 , escapant del seu entorn local amb una elevada
component de la velocitat perpendicular al pla galàctic. Hem estat capaços
d’explicar tant la velocitat espacial elevada com l’excentricitat observades en
un escenari d’explosió de supernova simètrica, amb una important pèrdua de
massa de ∆ M ∼ 17 M⊙.
5. Finalment, la velocitat espacial elevada i el possible temps de vida del micro-
quàsar indiquen que aquest podria assolir una latitud galàctica de b = −12 ◦ .

xxx Resum de la tesi
Per tant, si l’associació entre LS 5039 i la font EGRET 3EG J1824−1514
és correcta, podríem detectar microquàsars de raigs gamma fins a latituds
galàctiques de l’ordre de |b| 10 ◦ . Així, els microquàsars en fuga podrien es-
tar relacionats amb algunes de les fonts EGRET febles, variables i relativament
toves encara no identificades, situades per sobre i per sota del pla galàctic.
Apartat 3:
1. Hem desenvolupat un mètode d’identificació creuada per buscar noves ràdio
binàries X, que son microquàsars en potència. Els resultats obtinguts donen
confiança en el mètode, ja que la llista de fonts seleccionades inlcoïa tots els
microquàsars amb companyes massives en la regió |b| < 5 ◦ excepte un. La
mostra final, un cop excloses les fonts conegudes, conté 13 noves ràdio fonts
X, de les quals 8 han estat classificades com a objectes prioritaris.
2. Hem estudiat 6 d’aquests 8 objectes, i hem trobat contrapartides òptiques per
tots ells. Hem obtingut posicions tant en ràdio com en òptic, perfectament
compatibles entre elles. També hem obtingut espectres ràdio i magnituds
òptiques de les fonts.
3. Hem presentat observacions realitades amb la EVN i MERLIN d’aquestes sis
fonts. Cinc d’elles han estat detectades i hem obtingut imatges, que mostren
diferents morfologies: una font té un jet bipolar, tres fonts tenen un jet en un
únic sentit i una font és compacta.
4. Després d’una anàlisi detallada de les observacions, concloem que les fonts
1RXS J001442.2+580201 i 1RXS J013106.4+612035 són bons candidats a mi-
croquàsar. 1RXS J042201.0+485610 no ho és degut al seu espectre ràdio in-
vertit a altes freqüències, que suggereix un origen tèrmic per aquesta font, que
a més a més apareix lleugerament estesa en l’òptic. 1RXS J062148.1+174736
és probablement una galàxia, ja que la contrapartida òptica és estesa. La font
1RXS J072259.5−073131 és probablement extragalàctica de tipus blàzar, ja
que el torcement del seu jet ràdio és una propietat comuna en aquests ob-
jectes. Finalment, la font 1RXS J072418.3−071508 és un quàsar ja identificat.
5. Treball en curs. Observacions espectroscòpiques en l’òptic de les cinc primeres
fonts la tardor de 2002, per tal de desvetllar la seva natura galàctica o extra-
galàctica. D’altra banda, la resta de fonts de la mostra s’estudiaran possible-
ment en el futur.

Bibliografia
Casares, J., Ribó, M., Paredes, J. M., & Martí, J. 2002, en preparació
Clark, J. S., Reig, P., Goodwin, S. P., et al. 2001, A&A, 376, 476
Martí, J., Paredes, J. M., & Ribó, M. 1998, A&A, 338, L71
McSwain, M. V., & Gies, D. R. 2002, ApJ, 568, L27
McSwain, M. V., Gies, D. R., Riddle, R. L., Wang, Z., & Wingert, D. W. 2001,
ApJ, 558, L43
Motch, C., Haberl, F., Dennerl, K., Pakull, M., & Janot-Pacheco, E. 1997, A&A,
323, 853
Motch, C., Guillout, P., Haberl, F., et al. 1998, A&AS, 132, 341
Paredes, J. M., Martí, J., Ribó, M., & Massi, M. 2000, Science, 288, 2340
Paredes, J. M., Ribó, M., Ros, E., Martí, J., & Massi, M. 2002a, A&A, 393, L99
Paredes, J. M., Ribó, M., & Martí, J. 2002b, A&A, 394, 193
Reig, P., Ribó, M., Paredes, J. M., & Martí, J. 2002, en preparació
Ribó, M., Reig, P., Martí, J., & Paredes, J. M. 1999, A&A, 347, 518
Ribó, M., Paredes, J. M., Romero, G. E., et al. 2002a, A&A, 384, 954
Ribó, M., Ros, E., Paredes, J. M., Massi, M., & Martí, J. 2002b, A&A, 394, 983
xxxi

xxxii BIBLIOGRAFIA

Chapter 1
Introduction and background
Microquasars are radio emitting X-ray binaries displaying relativistic radio jets. In
this chapter we will first explain briefly the accretion mechanisms in X-ray binaries,
as well as perform an approach to the catalogued systems, in Sect. 1.1. We will then
focus on the basic properties of microquasars in Sect. 1.2. We will finish by stating
what was the motivation of this work, the search for new microquasars, in Sect. 1.3.
1.1 X-ray binaries
Since X-rays cannot propagate through the atmosphere and cannot reach the Earth
surface, astronomers had to wait until the 1960s to discover the first extrasolar X-
ray source, namely Sco X-1, thanks to observations with detectors onboard rockets.
However, the real take off of X-ray astronomy can be considered to be in 1970, when
the UHURU (freedom in Swahili) satellite was launched. It provided the first all-sky
X-ray map down to a flux of 1 mCrab, which contained 339 sources. It also allowed
the detection of X-ray pulsations and X-ray eclipses in some of these sources. The
UHURU observations allowed to find the optical counterparts of such objects, that
were called X-ray binaries:
• An X-ray binary is a binary system containing a compact object, either a
Neutron Star (NS) or a stellar-mass Black Hole (BH), accreting matter from
the companion star.
1

2 Chapter 1. Introduction and background
Several scenarios have been proposed to explain the X-ray emission of X-ray bina-
ries, depending on the nature of the compact object, its magnetic field in the case
of neutron stars and on the geometry of the accretion flow. In any case, the ac-
creted matter is accelerated to relativistic speeds, transforming its potential energy
provided by the intense gravitational field of the compact object into kinetic energy.
Assuming that this kinetic energy is finally radiated, we can compute the accretion
luminosity as:
Laccr 1
2 ˙
MaccrV 2 =
GMX ˙
Maccr
RX
, (1.1)
where ˙
Maccr is the accretion rate, V is the free fall speed defined as V = 2GMX/RX,
G is the gravitational constant and MX and RX are the mass and radius of the com-
pact object, respectively. On the other hand, there is a maximum theoretical value
for the accretion rate, when the radiation pressure balances gravity, called the Ed-
dington limit and expressed as:
˙
MEdd = 4πmpcRX
σT
, (1.2)
where mp is the proton mass, c is the speed of light and σT is the Thomson cross
section. The corresponding luminosity can be expressed as:
LEdd = 4πGMXmpc
σT
. (1.3)
In an X-ray binary, the accreted matter carries angular momentum, and on its
way to the compact object it usually forms an accretion disk around it. The matter
in the disk looses angular momentum due to viscous dissipation, which produces a
heating of the disk, and falls towards the compact object in a spiral trajectory. The
black body temperature of the last stable orbit in the case of a BH accreting at the
Eddington limit is given by:
T
2 × 10
K
7
−1/4 MX
M⊙
. (1.4)
For a compact object of a few solar masses the obtained temperature is ∼ 10 7 K.
At this temperature the energy will be mainly radiated in the X-ray domain of the
electromagnetic spectrum.
Using Eq. 1.1 we can see that for an accreting BH, taking as final radius the
Schwarzschild radius defined as RS = 2GMX/c 2 , the emitted accretion luminosity
is:
Laccr 1
2 ˙
Maccrc 2 . (1.5)

1.1. X-ray binaries 3
According to this result, a significant fraction of the rest mass of the accreted matter
can be transformed into radiation. If we consider this fraction to be 1/10 instead
of the quoted 1/2 value in Eq. 1.5, and use a typical luminosity of 10 37 erg s −1 for
X-ray binaries, we obtain ˙
Maccr 2 × 10 −9 M⊙ yr −1 . This mass would be given by
the companion or donor star. If instead of an accreting BH we consider a 2M⊙ NS
with a 10 km radius we obtain a slightly larger value of ˙
Maccr 6 × 10 −9 M⊙ yr −1 .
In any case, it is clear that a donor star mass loss rate greater than 10 −10 M⊙ yr −1 ,
available and sustainable over long periods of time, can easily explain the observed
X-ray luminosities in X-ray binaries.
If the compact object is a NS with a strong magnetic field, B ∼ 10 12 G, the
accretion disk will be disrupted at several thousand kilometers from the compact
object, and matter will follow the magnetic field lines until impacting onto the
magnetic poles, producing again X-ray emission. If there is a misalignment between
the rotation and magnetic axes, X-ray pulsations will be observed if the beamed
emission from the magnetic poles rotates through the line of sight. These particular
X-ray binaries are called X-ray pulsars.
On the contrary, if the compact object is a weak magnetic field NS, B < 10 10 G,
the accretion disk may reach it or come close to it. X-rays will be then produced in
the inner accretion disk and the boundary layer between the disk and the compact
object.
The number of known X-ray binaries has increased considerably in the last years.
There were 63 classified systems in 1983, and 188 in 1995. The current number of
catalogued X-rays binaries is ∼ 280 (Liu et al. 2000, 2001). According to the mass
of the donor, X-ray binaries are divided in High Mass X-ray Binaries (HMXBs) and
Low Mass X-ray Binaries (LMXBs).
1.1.1 High mass X-ray binaries
In HMXBs the donor star is an early type star, with an O or B spectral type.
HMXBs are conventionally divided into two subgroups: systems containing a B star
with emission lines, namely Be stars, and systems containing a supergiant (SG) O
or B star. Here we list some basic typical properties:

4 Chapter 1. Introduction and background
• Be stars: B or late O stars with luminosity classes III–V with emission lines,
indicative of an equatorial decretion disk. Masses in the range ∼ 8–20 M⊙.
Orbital periods in the range ∼ 10–250 days, with eccentricities 0.3. Do not
fill their Roche lobe, and accretion onto the compact object is produced via
mass transfer through the decretion disk. Most of these systems are transient
X-ray sources during periastron passage. The compact object is typically a
strong magnetic field neutron star, and most of these systems are X-ray pulsars.
The lifetime of Be/X-ray binaries is around ∼ 10 7 years.
• OB SG stars: O or B stars with luminosity classes I–III. Masses above 15 M⊙.
Orbital periods typically below 10 days, with eccentricities 0. The mass
transfer is due to a strong stellar wind and/or Roche lobe overflow, with mass
loss rates up to 6 × 10 −6 M⊙ yr −1 . The X-ray emission is persistent, and
large variability is usual. The lifetime of SG/X-ray binaries is between ∼ 10 5 –
10 6 years.
The typical X-ray to optical luminosity ratio is in the range LX/Lopt 10 −3 –10,
although highly absorbed systems in the optical have higher values.
The most recent catalog of HMXBs was compiled by Liu et al. (2000), and
contains 130 sources. We have added to this catalog a source previously classified
as a LMXB. We show in Fig. 1.1 the distribution of catalogued HMXBs in galactic
coordinates. As can be seen, these systems are located close to the galactic plane (in
or near spiral arms) because are young stellar systems of Population I. The presence
of HMXBs in the Small and Large Magellanic Clouds (SMC and LMC) is also clearly
seen.
In Table 1.1 we list the number of different types of HMXBs within the SMC,
the LMC, the Galaxy and total values. In the Be and SG types we have considered
all the confirmed cases and also the suspected ones based on X-ray properties or
orbital solutions. NF stands for those sources that do not fit in the Be or SG types.
Under the question mark we list the number of sources without a clear identification
or with a suspected white dwarf (WD) compact object. All of them amount to 131
systems, of which some are X-ray Pulsars (XPs) and/or transient sources, as quoted
also in the table. Finally, the number of Radio Emitting X-ray Binaries (REXBs)
has also been included (see Sect. 1.1.3).

1.1. X-ray binaries 5
Galactic latitude [deg]
90
60
30
0
−30
−60
High Mass X−ray Binaries
−90
180 120 60 0 300 240 180
Galactic longitude [deg]
Figure 1.1: Distribution of catalogued HMXBs in galactic coordinates. SMC and LMC
stand for Small and Large Magellanic Clouds, respectively. The dashed line represents
the galactic plane.
SMC
LMC
Table 1.1: Types and location of catalogued HMXBs.
Location Be SG NF ? Total XPs Transients REXBs
SMC 23 1 0 1 25 16 17 0
LMC 12 2 3 3 20 6 8 0
Galaxy 42 15 3 26 86 49 36 8
Total 77 18 6 30 131 71 61 8

6 Chapter 1. Introduction and background
The galactic latitude distribution of the 86 galactic HMXBs has a mean of −0.5 ◦ ,
and a standard deviation of 4.6 ◦ . In fact, a total of 80 sources are located within
|b| < 10 ◦ , 78 of them in the |b| < 5 ◦ area. Hence, around 93% and 91% of the galactic
HMXBs are located within the previously quoted regions, respectively. Galactic
sources with |b| > 10 ◦ are nearby systems (d 1 kpc), which have in fact normal
distances to the galactic plane when compared to the other ones.
An inspection of Table 1.1 reveals at first sight that among the 60 identified
HMXBs in the Galaxy the most common type is the Be/X-ray binary (∼ 70%),
followed by the SG/X-ray binary type (∼ 25%). Another remarkable fact is the
large number of XPs: ∼ 54% in total and ∼ 57% in the Galaxy.
1.1.2 Low mass X-ray binaries
In LMXBs the donor has a spectral type later than B, and has a mass ≤ 2 M⊙.
Although typically is a non-degenerated star, there are some examples where the
donor is a WD. The orbital periods are in the range 0.2–400 hours, although the
typical value is < 24 hours. The orbits are circular, and mass transfer is due to
Roche lobe overflow. Most of LMXBs are transients, probably as a result of an
instability in the accretion disk or a mass ejection episode from the companion. The
lifetime of LMXBs is estimated to be ∼ 10 8 –10 9 years. The typical ratio between
X-ray to optical luminosity is in the range LX/Lopt 100–1000, and the optical
light is dominated by X-ray heating of the accretion disk and the companion star.
A considerable fraction of LMXBs display X-ray bursts, which are due to ther-
monuclear flashes on the surface of accreting NSs. These sources are usually called
bursters. On the other hand, some LMXBs are classified as ‘Z’ and ‘Atoll’ sources,
according to the pattern traced out in the X-ray color-color diagram. The first ones
are thought to be weak magnetic field neutron stars of the order of 10 10 G with
accretion rates around 0.5–1.0 ˙
MEdd, while the second ones are believed to have
even weaker magnetic fields of 10 8 G and lower accretion rates of 0.01–0.1 ˙
MEdd.
The most recent catalog of LMXBs was compiled by Liu et al. (2001), and
contains 150 sources. We have removed from this catalog a source recently classified
as a HMXB. We show in Fig. 1.2 the distribution of catalogued LMXBs in galactic
coordinates. As can be seen, and contrary to the distribution of HMXBs shown in

1.1. X-ray binaries 7
Galactic latitude [deg]
90
60
30
0
−30
−60
Low Mass X−ray Binaries
−90
180 120 60 0 300 240 180
Galactic longitude [deg]
Figure 1.2: Distribution of catalogued LMXBs in galactic coordinates. LMC stands for
Large Magellanic Cloud. The dashed line represents the galactic plane.
Fig. 1.1, these systems do not trace the galactic disk, and are located in the galactic
bulge and in globular clusters at relatively high galactic latitudes, because are old
stellar systems of Population II. Only two LMXBs are catalogued in the LMC, and
none in the SMC.
In Table 1.2 we list the number of different types of LMXBs within the SMC,
the LMC, the Galaxy and total values. There are only 5 XPs, i.e., ∼ 3% of the
149 catalogued LMXBs, compared to the ∼ 54% in HMXBs. This fact tells us that
most of LMXBs are low magnetic field NSs or BHs. Two of these XPs are also X-ray
bursters. There are 63 X-ray bursters, all of them containing NS compact objects.
A total of 6 Z sources have been found in the Galaxy, while 1 is suspected in the
LMC (omitted here). There are 18 Atoll sources, of which 16 are X-ray bursters,
LMC

8 Chapter 1. Introduction and background
Table 1.2: Types and location of catalogued LMXBs.
Location Total XPs Bursters Z sources Atoll sources Transients REXBs
SMC 0 0 0 0 0 0 0
LMC 2 0 0 0 0 0 0
Galaxy 147 5 63 6 18 75 35
Total 149 5 63 6 18 75 35
including 5 X-ray transients. Approximately half of LMXBs are transients. REXBs
will be discussed in Sect. 1.1.3.
The galactic latitude distribution of the 147 galactic LMXBs has a mean of +0.4 ◦ ,
and a standard deviation of 12.2 ◦ , to be compared with the 4.6 ◦ value obtained
for HMXBs. It is clear the higher scatter in galactic latitude, compatible with
Population II sources. In fact, 13 LMXBs are located in globular clusters.
1.1.3 Radio emitting X-ray binaries
The first X-ray binary known to display radio emission was Sco X-1 in the late 1960’s.
Since then, many X-ray binaries have been detected at radio wavelengths with flux
densities ≥ 0.1–1 mJy. Moreover, the flux densities detected at cm radio wavelengths
are produced in small angular scales, and cannot be produced by thermal emission
mechanisms.
In this context, the most efficient known mechanism for production of intense
radio emission from astronomical sources is the synchrotron emission mechanism, in
which highly relativistic electrons interacting with magnetic fields produce intense
radio emission which tends to be linearly polarized. The observed radio emission can
be explained by assuming a spatial distribution of non-thermal relativistic electrons,
usually with a power-law energy distribution, interacting with magnetic fields.
Since some REXBs, like SS 433, were found to display elongated or jet-like
features, like in AGNs and quasars, it was proposed that flows of relativistic elec-
trons were ejected perpendicular to the accretion disk, and were responsible for
synchrotron radio emission in the presence of a magnetic field. Several models have
been proposed for the formation and collimation of the jets, including the presence

1.1. X-ray binaries 9
of an accretion disk close to the compact object, a magnetic field in the accretion
disk, a high spin for the compact object, etc. However there is no clear agreement
on what mechanism is exactly at work.
Among high mass REXBs we find very different types of donors, such as Be,
B[e], SG and Wolf Rayet stars, as well as systems that do not fit in the standard
categories. Among low mass REXBs we find all six Z sources, five Atoll sources, two
persistent sources near the Galactic center and a large number of transient sources.
As quoted in Tables 1.1 and 1.2, there are around 8 radio emitting HMXBs and
35 radio emitting LMXBs. The corresponding abundances are ∼ 6% and ∼ 23%,
respectively. There is a significant difference between those values. However, the
sources located in the Large and Small Magellanic Clouds, at ∼ 50 kpc and ∼ 58 kpc,
respectively, are around 5 times farther than the most distant galactic REXBs,
and the flux density of those sources in quiescence is expected to be ∼ 25 times
lower, preventing any radio detection up to now. If we exclude these sources, then
the abundances change into ∼ 9% and ∼ 24%, respectively. Moreover, since the
strong magnetic field X-ray pulsars disrupt the accretion disk at several thousand
kilometers from the neutron star, there is no inner accretion disk to launch a jet
in these systems, and no synchrotron radio emission has ever been detected in any
of those sources. Hence, considering only the galactic HMXBs and LMXBs and
excluding XPs, the previous numbers change into ∼ 22% and ∼ 24%, respectively,
which are indeed quite similar. As pointed out by several authors and as we have
seen, although the division of X-ray binaries in HMXBs and LMXBs is useful for
the study of binary evolution, it is probably not important for the study of the
radio emission in these systems, where the only important aspect seems to be the
presence of an inner accretion disk capable of producing radio jets. However, the 8
radio emitting HMXBs include 6 persistent sources and 2 transient sources, while
among the 35 radio emitting LMXBs we find 11 persistent sources and 24 transient
sources. It is clear that the difference between the persistent and transient behavior
highly depends on the mass of the donor.
In any case we can say that, excluding X-ray pulsars, 20 to 25% of the catalogued
galactic X-ray binaries have been detected at radio wavelengths, regardless of the
nature of the donor. The corresponding ratio of detected/observed sources is proba-
bly much higher. However, it is difficult to give reliable numbers, since observational
constrains arise when considering transient sources observed in the past (large X-ray

10 Chapter 1. Introduction and background
error boxes, single dish and/or poor sensitivity radio observations, etc.), and we do
not know how many non-detections have not been published.
1.2 Microquasars
A microquasar is a radio emitting X-ray binary displaying relativistic radio jets. The
name was given not only because of the observed morphological similarities between
these sources and the distant quasars but also because of physical similarities, since
when the compact object is a black hole, some parameters scale with the mass of
the central object. A schematic illustration comparing some parameters in quasars
and microquasars is shown in Fig. 1.3.
In this context, we can see that the use of Eq. 1.4 leads to temperatures of
∼ 10 7 K in the case of microquasars containing stellar-mass black holes and ∼ 10 5 K
in the case of quasars containing supermassive black holes (10 7 –10 9 M⊙). This
explains why in microquasars the accretion luminosity is radiated in X-rays, while
it is done in the optical/UV domain in the case of quasars.
On the other hand, the characteristic jet sizes seem to be proportional to the
mass of the black hole, since radio jets in microquasars have typical sizes of the
order of light years, while radio jets in quasars reach distances up to several million
light years in giant radio galaxies.
Last, but not least, the timescales appear to be directly scaled with the mass
of the black hole following τ RS/c = 2GMX/c 3 ∝ M. Therefore, phenomena
that take place in timescales of years in quasars can be studied in minutes in micro-
quasars.
In this sense, one can say that microquasars mimic, on smaller scales, many of
the phenomena seen in AGNs and quasars, but allow a better and faster progress to
understand the accretion/ejection processes that take place near compact objects.
The number of currently known microquasars is ∼ 16, among the 43 catalogued
REXBs. It is interesting to note that some authors have proposed that all REXBs
are microquasars, and would be detected as such provided that there is enough
sensitivity and/or resolution in the radio observations.

1.2. Microquasars 11
Figure 1.3: The quasar-microquasar analogy (from Mirabel & Rodríguez 1998).

12 Chapter 1. Introduction and background
1.2.1 Projection and special relativity effects
The detected radio emission in the jets of microquasars is usually in the form of
blobs, i.e., plasma clouds (or plasmons) emitting synchrotron radiation, which prop-
agate through space at relativistic bulk velocities. Therefore, projection and special
relativity effects have to be taken into account when analyzing the observations.
υ
¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
υ t cosθ
¢
¡ ¢ £ £
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦
υ t
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦ ¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦
¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤
θ
towards the observer
υ t sinθ
Figure 1.4: Scheme of an ejection of plasma clouds in opposite senses.
Projection effects: jet parameters and superluminal motion
Let us consider a pair of relativistic plasma clouds ejected simultaneously from a
central source and moving at intrinsic velocity v in opposite senses, as shown in
Fig. 1.4, in a direction forming an angle θ with respect to the line of sight. If
we obtain several images at different epochs, we will measure the following proper
motions for the approaching and receding components, respectively:
υ
β sin θ c
µa =
, (1.6)
(1 − β cos θ) D
β sin θ c
µr =
, (1.7)
(1 + β cos θ) D

1.2. Microquasars 13
where β is the velocity of the clouds in units of the speed of light (β = v/c) and
D is the distance between the observer and the source. Hence, the approaching
component will have an apparent proper motion, projected in the plane of the sky,
higher than the receding one (e.g., see Fig. 1.5). This pair of equations can be
directly transformed into:
β cos θ = µa − µr
µa + µr
, (1.8)
c tan θ (µa − µr)
D = .
2 µaµr
(1.9)
It is interesting to note that if the proper motions are measured, we can directly
compute the product β cos θ. Moreover, since both variables, β and cos θ, take values
between 0 and 1, it is clear that knowing β cos θ allows us to obtain a lower limit for
the velocity and an upper limit for the angle. We can also compute an upper limit
for the distance to the source given by:
D ≤
c
√ µaµr
. (1.10)
On the other hand, if we have a single image and we do not know the epoch
of the ejection, we can neither measure nor infer the proper motions. However, if
we detect the central source, we can cancel the time variable by using the relative
distances to it, da and dr, and transform Eq. 1.8 into:
β cos θ = µa − µr
µa + µr
= da − dr
da + dr
. (1.11)
As in Eq. 1.8, we can obtain a lower limit for the velocity and an upper limit for the
angle, but in this case no upper limit for the distance to the source can be obtained.
Finally, using Eq. 1.6 we can directly solve for the apparent velocity of the
approaching component (va = µaD):
va =
β sin θ
(1 − β cos θ)
c . (1.12)
We can see that if β cos θ approaches 1, the detected apparent velocity will be higher
than the speed of light. In fact, this will always happen if the condition
β >
1
sin θ + cos θ
(1.13)
is satisfied. The minimum β to detect the so-called superluminal motions is ∼ 0.71
for an angle of 45 ◦ . Higher values of β are required for different angles, according
to Eq. 1.13.

14 Chapter 1. Introduction and background
E
N
1"
MAR 27
APR 03
APR 09
APR 16
APR 23
APR 30
Figure 1.5: VLA multiepoch observations at 3.6 cm wavelength of the superluminal
ejection of GRS 1915+105 in 1994. The approaching component, to the left, appeared
to move at 1.25c and be brighter than the receding one, to the right, which moved with
an apparent velocity of 0.65c (from Mirabel & Rodríguez 1994).

1.2. Microquasars 15
Special relativity effects: Doppler shift and boosting
Taking into account that the plasma clouds move at relativistic speeds, there will
be a shift in the observed to emitted wavelengths. The ratios of the observed (νa
and νr) to emitted frequency (ν0) are given by the Doppler factors:
δa = νa
ν0
δr = νr
ν0
=
=
1
Γ(1 − β cos θ)
1
Γ(1 + β cos θ)
, (1.14)
, (1.15)
where Γ is the bulk Lorentz factor given by Γ = (1 − β 2 ) −1/2 . δr is always smaller
than 1, while δa is smaller than 1 for large values of θ and rises above 1 for small
values of θ, i.e., when the jet is pointing to us. Hence, an eventually observed
receding spectral line will always be red-shifted, while the blue or red-shift of the
approaching one will depend on the ejection angle. If for a particular event we were
able to measure a frequency shift in a spectral line produced in the plasmons, and
also compute β cos θ from Eq. 1.8, we could then find β, and hence θ and D by
using Eq. 1.8 and Eq. 1.9, respectively. This would allow us to solve for the whole
parameters of an ejection.
On the other hand, the measured flux densities of the approaching and receding
clouds will also suffer from special relativity effects. The ratios of the observed (Sa
and Sr) to emitted flux density (S0) from a twin pair of optically thin, isotropically
emitting jets are given by:
Sa
S0
Sr
S0
= δ k−α
a , (1.16)
= δ k−α
r , (1.17)
where k equals 2 for a continuous jet and 3 for discrete condensations, and α is the
spectral index of the radio emission (Sν ∝ ν +α ). Typical values for α in the optically
thin regime are between −0.5 and −1. Hence, the exponent in the equations above
will typically be in the range 2.5–4. As a consequence, the detected flux density of
the receding plasma cloud will be diminished (deboosting), while for the approaching
one will depend on the angle, as the Doppler factor does, i.e., we will have deboosting
for large angles and considerable boosting for small angles. Using these equations,
the ratio of approaching to receding flux densities measured at equal distance from
the core is:
Sa
Sr
=
k−α 1 + β cos θ
1 − β cos θ
. (1.18)

16 Chapter 1. Introduction and background
This ratio is always greater than 1, implying that we will detect the approaching
component brighter than the receding one (e.g., see Fig. 1.5). Solving for β cos θ in
Eq. 1.18 we obtain:
β cos θ = (Sa/Sr) 1/(k−α) − 1
(Sa/Sr) 1/(k−α) + 1
. (1.19)
As we have seen, at any given time da > dr. Therefore, if we only have a single
image of an ejection event, the receding component will be closer to the core than
the approaching one. Since the flux density decreases with increasing distance from
the core (due to adiabatic losses, and assuming no interaction with the interstellar
medium), the measured ratio Sa/Sr will be lower than the one that should be used
in the equation above, which is only valid at equal distances from the core. As a
consequence, in such a case Eq. 1.19 only allows us to obtain a lower limit for β cos θ.
Finally, it may also happen that we have a single image where only the approach-
ing component is detected, either because the relativistic deboosting of the receding
one is very strong or because of lack of sensitivity. In any case, we can use the fact
that we do not detect the counter-jet, and replace Sr with the 3σ level of the image
in Eq. 1.19. This will only provide a lower limit to β cos θ, expressed as
β cos θ > (Sa/3σ) 1/(k−α) − 1
(Sa/3σ) 1/(k−α) + 1
1.2.2 Quasars and microquasars
. (1.20)
After the introduction on accretion and the former explanation about special rel-
ativity effects, we can comment on some specific differences between quasars and
microquasars.
In the distant quasars radio emission was detected in the early times of radioas-
tronomy, because the jets in the detected quasars have small angles with respect to
the line of sight, and the flux densities of the approaching components are signifi-
cantly Doppler boosted. Hence, these objects appear in the sky as very bright radio
sources, showing one-sided jets in high resolution radio interferometric observations,
which usually reveal superluminal motions. The probability of having a quasar with
a jet pointing close to our line of sight is small, but provided enough number of
quasars, these objects are easy to detect, because their emission is persistent. As-
tronomers searched for the optical counterparts, and found objects that looked like

1.2. Microquasars 17
stars but without stellar-like spectra. Therefore, these objects were called ’quasi
stellar radio sources’ (quasars). In fact, the detected optical emission has its origin
in an underlying galaxy, which can be clearly imaged for the closer sources. This
galaxy supplies matter for the accretion process taking place in the central super-
massive black hole of 10 7 –10 9 M⊙, responsible for launching the jet that we detect
as a quasar. As commented above, the inner part of the accretion disk radiates in
the optical/UV domains, and quasars do not show strong X-ray emission. Typical
luminosities are around 10 47 erg s −1 , implying accretion rates of ∼ 10 M⊙ yr −1 .
If a quasar has a jet pointing almost directly to us (within a few degrees) it is
called blazar (after the object BL Lac). Blazars show extreme variability, bending
of the jets (due to projection effects) and strongly polarized radio emission.
If a quasar has a jet pointing away from our line of sight the radio emission will
be deboosted. Hence, if it is a distant object it will not be detected. However, if
it is relatively close (∼ 100 Mpc depending on the resolution and sensitivity of the
observations), it may be that both the approaching and receding jet components are
detected.
Since microquasars are nearby sources there is no need of extreme Doppler boost-
ing in order to detect them at radio wavelengths. Hence, it is not necessary to have
ejections with angles close to the line of sight, and most of times both the approach-
ing and receding components are detected. The detected radio emission is most of
times transient, and at optical wavelengths are not surprising objects. Hence, as-
tronomers had to wait until the launch of X-ray satellites, capable of detecting the
persistent/transient strong and usually hard X-ray emission coming from accretion
disks in microquasars, to discover the galactic relatives of the distant quasars.
1.2.3 Known microquasars
The first microquasar to be discovered was SS 433, although it was considered as a
very strange object for several years. In early 1990’s, two microquasars were discov-
ered near the galactic center, and in 1994 it was discovered the first superluminal
source in the Galaxy, namely GRS 1915+105. The current census of microquasars
contains around 16 sources, depending on the specific properties one imposes to
consider a source as a microquasar. In Table 1.3 we list some of the interesting prop-

18 Chapter 1. Introduction and background
erties of the sources we consider as known microquasars in the Galaxy, including
the position, spectral type of the donor and nature of the compact object, distance
to the system, orbital period, optical magnitude in V or other filters, mass of the
compact object, X-ray and radio luminosities (typical for the persistent sources and
peak values for the transient sources), the activity at radio wavelengths (persistent
or transient), apparent and intrinsic velocity, inclination angle of the detected ejec-
tion(s) and jet size. The objects are grouped in HMXBs and LMXBs and ordered in
increasing right ascension. In some objects only hints of relativistic jets have been
seen or are clearly suspected.
Several public databases, such as SIMBAD, GBI/NASA monitoring program,
ASM onboard RXTE have been consulted to compile the information present in
Table 1.3, as well as the catalogs by Liu et al. (2000, 2001) and lots of references
therein. This compilation is not complete nor exhaustive, and is only provided to
give an idea about the observed properties in microquasars.
The table includes well known high mass REXBs: LS I +61 303, that displays
radio outbursts every 26.5 days; SS 433, with their precessing radio jets moving
at 0.26c; Cygnus X-1, the first binary where evidences of the existence of a BH
were found, and recently classified as a microquasar; and Cygnus X-3, famous for
their strong radio outbursts reaching flux densities of several Jy. The other massive
systems are V4641 Sgr, a candidate to be a microblazar source (in an analogy with
the extragalactic blazars) with extreme superluminal motion, and LS 5039, which
will be discussed in Part I. We point out that the only other massive REXBs are
γ Cas, which is a very peculiar system in many aspects, and CI Cam, where radio
emission seems to be produced in a fairly isotropic nebula.
Regarding low mass REXBs in Table 1.3 we find: XTE J1118+480, which con-
tains a black hole and has a Galactic halo orbit; Circinus X-1, which experiences
outbursts every 16.6 days; XTE J1550−564, a system that contains a black hole
and where hints of superluminal motion have been found; Scorpius X-1, an histor-
ical source recently added to the group; GRO J1655−40, the second superluminal
source to be discovered in the Galaxy; GX 339−4, where the jet has not clearly been
imaged but strong evidences of its existence have been found; 1E 1740.7−2942, the
great annihilator near the Galactic center; XTE J1748−288, a superluminal source
where deceleration and brightening of jet components due to the interaction with
the interstellar medium have been observed; GRS 1758−258, a persistent jet source

1.2. Microquasars 19
Table 1.3: Microquasars in the Galaxy.
§¨©
©
©
¨
¨
¨
¨
¨
§
§
¨
§
¨
¨
¨
¨
§
¨
§

20 Chapter 1. Introduction and background
near the Galactic center; and GRS 1915+105, the first superluminal source to be
discovered in the Galaxy, and recently found to have the most massive black hole
as well as the longest orbital period in the microquasar group.
We can see that up to now, four sources have shown superluminal motions,
namely GRS 1915+105, GRO J1655−40, XTE J1748−288 and V4641 Sgr, while
such phenomenon is also suspected in XTE J1550−564.
We show in Fig. 1.6 the distribution in galactic coordinates of the microquasars
listed in Table 1.3, together with their respective names.
Galactic latitude [deg]
90
60
30
0
−30
−60
XTE J1118+480
LS I +61 303
Microquasars in the Galaxy
Cyg X−1
Cyg X−3
GRS 1758−258
GRS 1915+105
SS 433
LS 5039
Sco X−1
V4641 Sgr
XTE J1748−288
GRO J1655−40
Cir X−1
GX 339−4
1E1740.7−2942
XTE J1550−564
LMXB
LMXB (rel. jets ?)
HMXB
HMXB (rel. jets ?)
−90
180 120 60 0 300 240 180
Galactic longitude [deg]
Figure 1.6: Distribution of known microquasars in galactic coordinates. Filled circles
represent those sources where relativistic jets have been imaged, while open circles are
used for those where hints of relativistic jets have been seen or are clearly suspected.

1.2. Microquasars 21
1.2.4 Accretion disk and jet formation
It is currently accepted that the formation of relativistic radio jets requires the
presence of a compact object (a potential well) and an accretion disk surrounding
it. The mechanism of jet formation and collimation is not yet well understood,
but we can gain knowledge and test proposed models after performing observations
like the ones of GRS 1915+105 shown in Fig. 1.7. We can see that after a large-
amplitude quasi-periodic oscillation, the X-ray flux drops substantially, a behavior
interpreted as the emptying of the inner accretion disk. At the same time, the X-
ray spectrum becomes significantly hard, until when an X-ray spike is seen. Then
the X-ray spectrum softens again and we can see the starting of a flare at infrared
wavelengths, which is interpreted in terms of synchrotron emission by an expanding
plasma cloud, as shown above the lightcurves, ejected just when the X-ray spike
took place. As the plasma cloud expands adiabatically, the maximum of the emitted
energy shifts to longer wavelengths, and a subsequent flare is detected in the radio.
Luminosity [arbitrary units]
1.0
0.5
0.0
X−RAYS
Disk
emptying
280.000
INFRARED
Ejection
RADIO
km/s
7.9 8.0 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9
UT Time [hours]
Figure 1.7: Simultaneous multiwavelength behavior at X-ray, radio and infrared wave-
lengths of GRS 1915+105 during 1997 September 9 (from Mirabel et al. 1998).

22 Chapter 1. Introduction and background
1.2.5 Black hole states and radio emission
Black holes mainly display two different kinds of behavior at X-ray energies. On one
hand, when the total X-ray luminosity is found to be low, they display a spectrum
significantly hard up to a few hundreds of keV. This is the so-called low/hard state.
On the other hand, when they exhibit a high X-ray luminosity, their spectrum is
substantially softer. This is the so-called high/soft state. The lower energy photons,
responsible for the enhancement of X-ray luminosity and steepening of the spectrum
in the high/soft state, are produced by a multicolor blackbody in the inner accretion
disk, while the hard X-ray tail in the low/hard state is thought to be the result of a
(inverse) Comptonizing corona above the disk. Apart from these two states, there
is also the off state, when X-ray emission is almost or totally suppressed, and the
hybrids very high and intermediate states, where both spectral characteristics (soft
disk photons and hard tail) coexist.
The accepted idea was that the accretion rate increased from the off state,
through the low/hard state, the intermediate state up to the high/soft state (and
eventually the very high state). However, this picture is not so clear now, because
the same characteristics of a given spectral state (X-ray spectral and timing proper-
ties) have been observed at very different X-ray luminosities (a traditional indicator
of accretion rate).
An important insight into the possible scenario related to the black hole states
has been provided by radio observations. In this sense, long-term multiwavelength
campaigns such as the one shown in Fig. 1.8 are very useful. There we can see two
years of monitoring of GX 339−4 at radio, hard X-rays and soft X-rays. In the
upper part of the figure it is marked the X-ray spectral state of the source along
time. Starting in the low/hard state, a transition is observed into the high/soft
state, which lasted around one year, followed by a transition back to the low/hard
state and ending with an off state. The radio flux is clearly correlated with the
hard X-rays in all spectral states. The correlation between the radio flux and the
soft X-rays is more complicated, but can be shown to behave as follows: during
the low/hard state radio emission is correlated with the soft X-ray photons, while
during the high/soft state radio emission is suppressed. Faint radio emission is
detected during the off state, compatible with it being a lower luminosity low/hard
state. It is interesting to note that the spectral index of the radio emission is flat
all the time, indicative of a compact and partially self-absorbed jet, except when

1.2. Microquasars 23
Low-hard state High-soft state
L-H state Off state
Figure 1.8: Two-year monitoring of the black hole candidate GX 339−4 at radio wave-
lengths (ATCA and MOST), hard X-rays (BATSE, 20–100 keV) and soft X-rays (RXTE
ASM, 1.5–12 keV) (from Corbel et al. 2000).
transitions between spectral states take place. Then, the radio emission is optically
thin, probably as a result of major ejections of relativistic plasma clouds.
After these observational results, it was proposed that compact radio jets may
always be present during the low/hard states (as observed in other sources), while
are suppressed in the high/soft states. The reason is not clear yet, but a coupling
between the radio jet and the hard X-ray emitting corona is firmly established. A
suggested possibility is that the corona is in fact the base of the radio jet.
On the other hand, the synchrotron radiation mechanism, which is at work in
the jet, has been seen up to infrared and possibly optical wavelengths. Moreover,
it has been suggested that it could also generate photons up to X-ray energies in
its optically thin part. If this is correct, then the jets would be extremely powerful,
and certainly any model explaining accretion should account for their presence.

24 Chapter 1. Introduction and background
1.3 Motivation of the thesis
As we have seen, the current number of microquasars is ∼ 16. This number could
increase dramatically if, as it has been proposed by Fender & Hendry (2000), radio
emission from X-ray binaries always arises from relativistic jets. This would imply
that all REXBs, ∼ 40 systems, are microquasars. However, the situation is not
so clear yet and REXBs appear as a rather heterogeneous group including both
high-mass (e.g., Cyg X-3, LS I +61 303) and low-mass companions (e.g., Aql X-1,
GX 339−4). Moreover, in the latter group, radio emission has been detected in all
six Z sources and only in a few Atoll sources. In addition, the detection of radio
emission does not seem to be limited to a particular type of compact object, since
there are REXBs harboring black holes (e.g., GRO J1655−40, Cyg X-1) and neutron
stars (e.g., Aql X-1, Sco X-1). Moreover, when one begins to further divide the group
of REXBs according to their optical or compact companion or whether radio jets are
present or not, the number of systems is no longer enough to perform a meaningful
statistical study. The situation is even worse when we focus on microquasars.
There is an important question to be answered: what is the matter content of
the jets? Up to now, atomic emission lines have only been observed in the jets
of SS 433, indicating that the jets are baryonic. However, some of the proposed
models can only account for leptonic (e + e − ) jets. Therefore, it is crucial to unveil
the matter content if we want to correctly model the accretion/ejection processes
that take place near compact objects.
Another interesting issue is the possibility of microquasars being sources of high
energy γ-rays. Before this work, the 26.5 d periodic REXB LS I +61 303 was
the only X-ray binary that could be associated with a high-energy γ-ray source
(E > 100 MeV) of the third EGRET catalog (Hartman et al. 1999), namely
3EG J0241+6103 (Kniffen et al. 1997). The search for new REXBs could allow
new findings. Indeed, the current most promising candidate is LS 5039 (see Part I),
which appears to be associated with the source 3EG J1824−1514 (Paredes et al.
2000). Although these associations will only be confirmed, or discarded, by future
missions such as GLAST, clearly a sample of two sources is a very poor one, and any
attempt to discover new REXBs/microquasars, and look for possible high energy
γ-ray emission, is warranted.
Hence, two reasons gave us a strong motivation to start a long-term project

1.3. Motivation of the thesis 25
focused on the search for new microquasars in the Galaxy:
1. Every new microquasar has allowed new interesting phenomenological findings,
which have to be taken into account in the proposed models to explain the
accretion/ejection processes taking place near compact objects.
2. To enhance the reduced sample and allow meaningful statistical studies in the
future.
In Part I of this thesis we present the discovery and subsequent study of the
microquasar LS 5039. This was the first source to be discovered within the most
global approach presented in Part II. There we explain how we have used the best
available X-ray and radio catalogs to search for new REXB/microquasar candidates.
We state our general conclusions in Part III.

26 Chapter 1. Introduction and background

Bibliography
Blandford, R. D., & Königl, A. 1979, ApJ, 232, 34
Blandford, R. D., & Payne, D. G. 1982, MNRAS, 199, 883
Blandford, R. D., & Znajek, R. L. 1977, MNRAS, 179, 433
Corbel, S., Fender, R. P., Tzioumis, A. K., et al. 2000, A&A, 359, 251
Drake, S. A., A Brief History of High-Energy Astronomy,
http://heasarc.gsfc.nasa.gov/docs/heasarc/headates/heahistory.html
Fender, R. P. 2001, MNRAS, 322, 31
Fender, R. P. 2002, in Proc. of Relativistic Flows in Astrophysics, ed. A. W. Guth-
mann, M. Georganopoulos, A. Marcowith & K. Manolakou, Lecture Notes in
Physics, 589, 101
Fender, R. P., & Hendry, M. A. 2000, MNRAS, 317, 1
Fender, R. P., & Kuulkers, E. 2001, MNRAS, 324, 923
Fender, R. P., Southwell, K., & Tzioumis, A. K. 1998, MNRAS, 298, 692
Fomalont, E. B., Geldzahler, B. J., & Bradshaw, C. F. 2001, ApJ, 558, 283
Hartman, R. C., Bertsch, D. L., Bloom, S. D., et al. 1999, ApJS, 123, 79
Hjellming, R. M., & Han, X. 1995, Radio properties of X-ray binaries, in X-ray Bi-
naries, ed. W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel (Cambridge
Univ. Press, Cambridge), 308
Koide, S., Shibata, K., Kudoh, T., & Meier, D. L. 2002, Science, 295, 1688
27

28 BIBLIOGRAPHY
Lewin, W. H. G., van Paradijs, J., & Taam, R. E. 1995, X-ray bursts, in X-ray Bina-
ries, ed. W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel (Cambridge
Univ. Press, Cambridge), 175
Liu, Q. Z., van Paradijs, J., & van den Heuvel, E. P. J. 2000, A&AS, 147, 25
Liu, Q. Z., van Paradijs, J., & van den Heuvel, E. P. J. 2001, A&A, 368, 1021
Livio, M. 1999, Phys. Rep., 311, 225
Markoff, S., Falcke, H., & Fender, R. 2001, A&A, 372, L25
Martí, J., Paredes, J. M., & Peracaula, M. 2001, A&A, 375, 476
Meier, D. L., Koide, S., & Uchida, Y. 2001, Science, 291, 84
Mirabel, I. F., & Rodríguez, L. F. 1994, Nature, 371, 46
Mirabel, I. F., & Rodríguez, L. F. 1998, Nature, 392, 673
Mirabel, I. F., & Rodríguez, L. F. 1999, ARA&A, 37, 409
Mirabel, I. F., Dhawan, V., Chaty, S., et al. 1998, A&A, 330, L9
Mirabel, I. F., Rodríguez, L. F., Cordier, B., Paul, J., & Lebrun, F. 1992, Nature,
358, 215
Negueruela, I., The X-ray binaries page,
http://www.astro.soton.ac.uk/∼bexmgr/xbp.html
Paredes, J. M., Martí, J., Ribó, M., & Massi, M. 2000, Science, 288, 2340
Rees, M. J. 1966, Nature, 211, 468
Rees, M. J. 1984, ARA&A, 22, 471
Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337
Stirling, A. M., Spencer, R. E., de la Force, C. J., et al. 2001, MNRAS, 327, 1273
Tanaka, Y., & Lewin, W. H. G. 1995, Black-hole binaries, in X-ray Binaries, ed.
W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel (Cambridge Univ.
Press, Cambridge), 126

BIBLIOGRAPHY 29
van Paradijs, J., & McClintock, J. E. 1995, Optical and ultraviolet observations of
X-ray binaries, in X-ray Binaries, ed. W. H. G. Lewin, J. van Paradijs, & E. P. J.
van den Heuvel (Cambridge Univ. Press, Cambridge), 58
Verbunt, F., & van den Heuvel, E. P. J. 1995, Formation and evolution of neutron
stars and black holes in binaries, in X-ray Binaries, ed. W. H. G. Lewin, J. van
Paradijs, & E. P. J. van den Heuvel (Cambridge Univ. Press, Cambridge), 457
White, N. E., X-ray binaries,
http://lheawww.gsfc.nasa.gov/users/white/xrb/xrb.html
White, N. E., Nagase, F., & Parmar, A. N. 1995, The properties of X-ray binaries,
in X-ray Binaries, ed. W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel
(Cambridge Univ. Press, Cambridge), 1

30 BIBLIOGRAPHY

Part I
Discovery and study of the
microquasar LS 5039
31

Chapter 2
Multiwavelength approach to
LS 5039
2.1 Introduction
After our discovery of the microquasar nature of LS 5039 in year 2000, several
groups have studied this source at different wavelengths. As a consequence, some
parameters of the system have been unveiled since then. Therefore, although we
could have reproduced all our findings 1,2,3,4,5,6,7 in a chronological way, we have
preferred to provide here a multiwavelength approach to the source using the new
available information to reanalyze some of the data.
Nevertheless, we think it is worth to start giving a chronology of findings on
LS 5039, which is briefly done in Sect. 2.2, to provide the reader with a general view.
We then perform the multiwavelength approach in Sects. 2.4–2.7, that contains our
discovery of the LS 5039 microquasar nature, the research done by us and comments
on what has been done by other groups. We propose a scenario to explain the
observational data in Sect. 2.8 and we finally state our conclusions in Sect. 2.9.
1 Published in Martí, J., Paredes, J. M. & Ribó, M. 1998, A&A, 338, L71.
2 Published in Ribó, M., Reig, P., Martí, J., & Paredes, J. M. 1999, A&A, 347, 518.
3 Published in Paredes, J. M., Martí, J., Ribó, M., & Massi, M. 2000, Science, 288, 2340.
4 Published in Ribó, M., Paredes, J. M., Romero, G. E., et al. 2002, A&A, 384, 954.
5 Published in Paredes, J. M., Ribó, M., Ros, E., Martí, J., & Massi, M. 2002, A&A, 393, L99.
6 Casares, J., Ribó, M., Paredes, J. M., & Martí, J. 2002, in preparation.
7 Reig, P., Ribó, M., Paredes, J. M., & Martí, J. 2002, in preparation.
33

34 Chapter 2. Multiwavelength approach to LS 5039
2.2 Chronology of findings on LS 5039
The V = 11.2 star LS 5039, located at an estimated distance of ∼ 3 kpc and close
to the galactic plane (l = 16.88 ◦ , b = −1.29 ◦ ), was initially proposed by Motch et al.
(1997) to be the optical counterpart of the X-ray source RX J1826.2−1450, a likely
High Mass X-ray Binary (HMXB).
Using data from the NRAO VLA Sky Survey and follow-up VLA observations,
Martí et al. (1998) discovered that the source was also a non-thermal radio emitter
with moderate variability.
X-ray observations of RX J1826.2−1450 carried out by Ribó et al. (1999) showed
that the X-ray spectrum was significantly hard (Γ = 1.95) and that emission was
detected up to 30 keV, with a strong Gaussian iron line at 6.6 keV, and neither
pulsed nor periodic emission was found on timescales of 0.02–2000 s and 2–200 days,
respectively. These authors also found that LS 5039 appeared to be a persistent non-
thermal radio emitter based on the GBI-NASA Monitoring Program observations.
Paredes et al. (2000) discovered that the system displays relativistic radio jets
(β > 0.15) at milliarcsecond scales thanks to VLBA observations, revealing the
microquasar nature of LS 5039, and proposed an association with the high-energy
γ-ray source 3EG J1824−1514.
While in the past the mass donor had been classified as an O7V((f)) star, optical
and near-IR spectroscopic observations by Clark et al. (2001) show that it is in fact
an O6.5V((f)) star.
McSwain et al. (2001) obtained the radial velocity curve of the system, with
their fitted parameters being a short period of P = 4.117 days, a high eccentricity
of e = 0.41, a radial velocity of the system of V0 = 4.6 km s −1 and a mass function
of f(m) = 0.00103 M⊙.
Ribó et al. (2002) used positions at optical and radio wavelengths obtained at
several epochs to compute independent optical and radio proper motions, which
were perfectly compatible, therefore confirming that both, the optical and the radio
emission, originate in the same object. These authors also found that LS 5039 is a
runaway X-ray binary with a space velocity ∼ 150 km s −1 , probably due to the mass
loss experienced during the SN event that created the compact object in this binary

2.3. An interesting target in the search for new microquasars 35
system. They also used all the available optical photometry to obtain an improved
distance to the source of 2.9 ± 0.3 kpc.
McSwain & Gies (2002), based on wind accretion models, have recently sug-
gested a neutron star (1–3 M⊙) as the compact object in LS 5039, and proposed an
inclination of i 30 ◦ for this binary system.
Paredes et al. (2002), thanks to EVN and MERLIN observations, have recently
confirmed the persistent nature of the relativistic radio jets in LS 5039, and proposed
a scenario based on the γ-ray/radio emission, in terms of the inverse Compton
mechanism for the γ-ray emission and synchrotron mechanism for the radio emission.
Finally, Casares et al. (2002) have just confirmed the ∼ 4 d orbital period, while
Reig et al. (2002) have determined that NH = 1.0 ± 0.3 × 10 22 cm −2 .
2.3 An interesting target in the search for new
microquasars
X-ray, optical and radio catalogs provide a fundamental tool for the search of new
sources with known multiwavelength behavior. In a first step, the most obvious
objects to inspect at radio wavelengths were those new X-ray binary candidates
already proposed from previous researches.
In this context, a systematic cross-identification of the ROSAT All Sky Survey
(RASS) (Voges et al. 1996, 1999) with OB star catalogs in the SIMBAD database
was carried out by Motch et al. (1997), hereafter M97. Nearly two tens of OB/X-
ray accreting binary candidates resulted from this work with different degrees of
reliability. In particular, LS 5039 was proposed by M97 to be an X-ray binary
system of the massive type with a high degree of confidence (see Fig. 2.1). The
unabsorbed X-ray luminosity in the 0.1–2.4 keV band, at an estimated distance
of 3.1 kpc, amounted to LX 8.1 × 10 33 erg s −1 . This fact, together with the
hardness of its X-ray spectrum were compatible with a neutron star, or a black
hole, accreting directly from the companion’s wind.. The optical counterpart of the
system appeared as a very luminous (V 11.2) star of the main sequence, with an
early O7V spectral type.

36 Chapter 2. Multiwavelength approach to LS 5039
DECLINATION (J2000)
-14 49 00
30
50 00
30
51 00
30
52 00
30
LS 5039 V band
LS 5039
ROSAT
53 00
18 26 20 15 10
RIGHT ASCENSION (J2000)
05
Figure 2.1: CCD image of LS 5039 obtained on 1998 June 7 with the 1.52 m OAN tele-
scope in the V Johnson filter. The circle with 35 ′′ radius represents the 90% confidence
position from the ROSAT PSPC pointed observations quoted in M97.
2.4 The radio counterpart: NVSS J182614−145054
2.4.1 A radio counterpart in the NVSS
The NRAO VLA Sky Survey (NVSS) (Condon et al. 1998) provides a valuable tool
to search for new REXBs. This catalog covers the sky north of δ = −40 ◦ (82% of the
celestial sphere) at a frequency of 1.4 GHz (20 cm wavelength) using the Very Large
Array (VLA) configurations D and DnC. It contains over 1.8 × 10 6 sources stronger
than its 2.5 mJy completeness limit. The rms positional uncertainties are less than
1 ′′ for sources stronger than 15 mJy and 7 ′′ for the faintest detectable sources.
For all the sources listed in M97 we examined the corresponding NVSS maps
at the 20 cm wavelength in a search for possible radio counterparts. We found one
interesting object in the M97 list that deserved special attention, namely LS 5039
(l = 16.88 ◦ , b = −1.29 ◦ ).

2.4. The radio counterpart: NVSS J182614−145054 37
The existence of a radio counterpart to LS 5039 was first suspected after inspec-
tion of the corresponding NVSS image. The 24 mJy source NVSS J182614−145054
lied outside, but very close to, the 90% confidence error circle of RX J1826.2−1450,
with 22 ′′ radius. This was the RASS source identified with LS 5039 by M97. These
authors also quoted another pointed ROSAT observation that yielded a 90% con-
fidence radius of 35 ′′ , again consistent with NVSS J182614−145054. In addition,
the NVSS coordinates were found to agree within 2 ′′ with the optical position of
LS 5039, as derived from the USNO-A1.0 catalog (Monet 1996, Monet et al. 1999).
All these coincidences together stimulated our interest about this source and
lead us to carry out higher resolution radio observations. This was considered to
be the next logical step in order to better measure both its position and spectral
properties. Therefore, we conducted a multiepoch and multifrequency radio study
of LS 5039 with the VLA.
2.4.2 VLA observations, discovery of a REXB
Observations and results
We observed 8 LS 5039 on several epochs at the wavelengths of 20, 6, 3.6 and 2.0 cm
with the Very Large Array (VLA) interferometer of the NRAO 9 . The VLA data were
processed following standard procedures within the aips package of NRAO. 3C 286
was used as the amplitude calibrator, while the phase calibrators observed were
1834−126 at 20 cm, 1820−254 at 6 and 3.6 cm, and 1911−201 at 2 cm, respectively.
The results of the interferometric runs on LS 5039 are summarized in Table 2.1,
where the flux density at several wavelengths is listed for the different observing
dates. Some 20 cm values collected from the literature have been also included. The
VLA data obtained in A configuration were further concatenated at 3.6 and 2.0 cm
in order to obtain very sensitive sub-arcsec resolution maps. These are displayed in
Fig. 2.2, where LS 5039 appears always as an unresolved point source (≤ 0.1 ′′ ). For
the epochs when nearly simultaneous multifrequency observations are available, we
have plotted in Fig. 2.3 the observed radio spectrum of LS 5039.
8 Published in Martí, J., Paredes, J. M. & Ribó, M. 1998, A&A, 338, L71.
9 The National Radio Astronomy Observatory is a facility of the USA National Science Foun-
dation operated under cooperative agreement by Associated Universities, Inc.

38 Chapter 2. Multiwavelength approach to LS 5039
DECLINATION (J2000)
Table 2.1: VLA observations of LS 5039.
Date Julian Day VLA S20 cm S6 cm S3.6 cm S2.0 cm
[JD−2400000] Conf. [mJy] [mJy] mJy] [mJy]
1989 May 02 (a) 47649 B 36 ± 1 — — —
1996 Jun 19 (b) 50254 D 23.7 ± 0.9 — — —
1998 Feb 11 50856.2 D 28 ± 2 23.4 ± 0.1 18.1 ± 0.1 12.3 ± 0.3
1998 Mar 10 50883.1 A 37 ± 1 24.0 ± 0.1 19.1 ± 0.1 14.1 ± 0.2
1998 Apr 09 50913.0 A 40 ± 1 23.6 ± 0.1 16.9 ± 0.1 12.0 ± 0.2
1998 May 12 50946.0 A 44 ± 1 25.7 ± 0.1 20.2 ± 0.1 15.1 ± 0.2
(a) Helfand et al. (1992). (b) NVSS value from Condon et al. (1998).
-14 50 52.0
52.5
53.0
53.5
54.0
54.5
55.0
55.5
56.0
LS 5039 3.6 cm
USNO
18 26 15.15 15.10 15.05 15.00
RIGHT ASCENSION (J2000)
DECLINATION (J2000)
-14 50 52.0
52.5
53.0
53.5
54.0
54.5
55.0
55.5
56.0
LS 5039 2.0 cm
USNO
18 26 15.15 15.10 15.05 15.00
RIGHT ASCENSION (J2000)
Figure 2.2: Left. Self-calibrated map of LS 5039 at 3.6 cm after concatenation of
all VLA A configuration data. Natural weight of the visibilities was used. The thick
cross indicates the optical position of LS 5039 as listed in the USNO-A1.0 catalog.
Contours are −3, 3, 5, 7, 10, 20, 30, 50, 100, 200, 400, 600, 800, 1000 and 1200 times
0.015 mJy beam −1 , the rms noise. The synthesized beam, shown at the bottom left
corner, is 0.40 ′′ × 0.25 ′′ , with position angle in 9.1 ◦ . Right. The same at 2.0 cm.
Contours are −3, 3, 5, 10, 20, 30, 50, 80, 100, 120 and 140 times 0.082 mJy beam −1 ,
the rms noise. The synthesized beam is 0.23 ′′ × 0.14 ′′ , with position angle in 13.4 ◦ .

2.4. The radio counterpart: NVSS J182614−145054 39
Flux density [mJy]
50
40
30
20
15
10
May 12
Apr 09
Mar 10
Feb 11
1 2 5 8 10 15 20
Frequency [GHz]
Figure 2.3: The radio spectrum of LS 5039 on different epochs during the 1998 obser-
vations with the VLA. Error bars not shown are smaller than the symbol size.
Astrometric identification
The J2000.0 ICRS coordinates of the LS 5039 radio position derived from the VLA
A configuration maps in Fig. 2.2 were found to be: α = 18 h 26 m 15.056 s ± 0.001 s ,
δ = −14 ◦ 50 ′ 54.24 ′′ ± 0.01 ′′ . These coordinates agree within 0.1 ′′ with those of
LS 5039 in the optical USNO-A1.0 catalog, whose typical astrometric error is about
0.25 ′′ . Therefore, we concluded that the identification of the LS 5039 radio counter-
part on the basis of astrometrical coincidence at the sub-arcsec level was virtually
certain. From source count analysis (e.g., Ledden et al. 1980), the a priori probabil-
ity of having a background extragalactic source with S20cm 30 mJy, or brighter,
within 0.2 ′′ of the LS 5039 optical position is as low as 10 −7 . To our knowledge,
this discovery increased to four the number of confirmed massive X-ray binaries with
associated radio emission at that epoch, the other three being SS 433, Cygnus X-1
and LS I +61 303 (Cygnus X-3 was a rather strange system with a Wolf rayet com-
panion, γ Cas a very peculiar system in many aspects, and the transients V4641 Sgr
and CI Cam were not yet discovered).

40 Chapter 2. Multiwavelength approach to LS 5039
Radio behavior
The radio spectra of LS 5039 shown in Fig. 2.3 are very suggestive of non-thermal
radio emission. A typical one can be well represented by
Sν
ν
= (52 ± 1)
mJy
GHz
−0.46±0.01
. (2.1)
Given the unresolved angular size (θ ≤ 0.1 ′′ ), the brightness temperature, expressed
as
TB
= 1.76 × 10
K
3
ν
−2
GHz
θ
arcsec
−2
Sν
mJy
, (2.2)
can be estimated, for example at the 20 cm wavelength, as TB ≥ 4 × 10 6 K. Such
a high value, together with the significantly negative spectral index α (where Sν ∝
ν +α ), safely rules out a thermal emission mechanism. Non-thermal synchrotron
radiation remains therefore as the most plausible interpretation for the LS 5039 radio
emission. We can also obtain a lower limit to the source angular size by preventing
catastrophic inverse Compton losses to occur (TB ≤ 10 12 K). This condition yields
θ ≥ 0.2 milliarcsecond (mas), implying that further progress in studying the LS 5039
structure could be feasible by the use of VLBI techniques (Sect. 2.4.4). We ignored
at that epoch if the radio emission was originated in collimated milliarcsec radio jets
or, alternatively, some other mechanism was at work.
Assuming equipartition arguments for synchrotron radio sources (Pacholczyk
1970), the total energy content and magnetic field in LS 5039 can be expressed as:
Etotal = c13(1 + k) 4/7 φ 3/7 R 9/7 L 4/7
rad
, (2.3)
H = 4.5 2/7 (1 + k) 2/7 c 2/7
12 φ −2/7 R −6/7 L 2/7
rad , (2.4)
where Lrad is the integrated radio luminosity, R the linear size of the emitting
region, φ the fraction of its volume covered by the magnetic field, k the electron
to proton energy ratio, and c12 and c13 some special functions of the synchrotron
theory tabulated in Pacholczyk (1970).
For a distance of 2.9 kpc (Ribó et al. 2002), and assuming that the spectrum
extends from 0.1 to 100 GHz, the total radio luminosity of the system can be eval-
uated as Lrad ∼ 1 × 10 31 erg s −1 . The radio to X-ray luminosity ratio is thus
Lradio/LX ∼ 10 −3 . Adopting φ = 1 and k = 1, the range of angular sizes found
above (0.2 mas ≤ θ ≤ 0.1 ′′ ) translates into the following allowed source parameters:
1.4 × 10 38 ≤ Etotal ≤ 4.3 × 10 41 erg and 0.01 ≤ B ≤ 2.0 G.

2.4. The radio counterpart: NVSS J182614−145054 41
The radio emission level seems also to be persistent on timescales of, at least, few
years. This is deduced from the detections listed in Table 2.1. On shorter timescales,
the 20 cm flux density varies from 27 to 44 mJy between our first and last observing
epochs, that are separated by four months. We also find some indications of radio
variability for the remaining shorter wavelengths. The amplitudes of variations at
6, 3.6 and 2.0 cm are however not larger than ∼ 10%. The overall spectral evolution
in Fig. 2.3 can be conceivably understood in terms of a progressive decrease in
the synchrotron self-absorption opacity. Unfortunately, the time sampling was not
frequent enough to determine what sort of radio events could configure the LS 5039
light curve during our VLA monitoring. In any case, both the non-thermal spectral
indices and the radio variability observed from month to month are usual features
in REXBs (see, e.g., Hjellming & Han 1995).
From the radio point of view, the LS 5039 behavior appeared so far consistent
with the general properties of REXBs, thus supporting the M97 identification. In
particular, the radio emission of LS 5039 was somehow reminiscent of LS I +61 303.
This is another massive REXB whose individual Lradio/LX ratio is remarkably similar
to that of LS 5039 (Taylor et al. 1996; Paredes et al. 1997). The energy and magnetic
field estimates for LS I +61 303 (Massi et al. 1993) would also fit comfortably well
within the limits found above for LS 5039. The analogy between both sources was
however not complete at this moment, i.e., there was no evidence for LS 5039 radio
outbursts in timescales of days or weeks.
Conclusions
After these observations (Martí et al. 1998) the star LS 5039 was proposed to be
a new high mass REXB. The radio counterpart was consistent, within astrometric
errors, with both the optical and X-ray position of the system. Its appearance at
radio wavelengths was that of a compact source, even at the highest VLA resolution.
The high brightness temperature and observed radio spectrum were suggestive of
non-thermal synchrotron emission. LS 5039 also exhibited some evidences of radio
variability during our VLA runs carried out on a monthly basis. The variations
were specially important at the 20 cm wavelength, where the source flux density
increased by almost a factor of two. This kind of behavior, although moderate,
strongly supported the identification of the radio counterpart.

42 Chapter 2. Multiwavelength approach to LS 5039
2.4.3 Long-term GBI monitoring, a persistent radio source
In order to follow the radio behavior of LS 5039, it was included at our request
in the list of radio sources routinely monitored at the Green Bank Interferometer
(GBI) 10 . The source was also included in the RXTE All Sky Monitor (ASM) service
at our request, to follow its behavior in the X-ray domain. The aims of the requested
observations were basically to be able to detect possible flaring events, X-ray/radio
correlations and to allow the search for the orbital period of the binary. The RXTE
ASM data will be discussed in Sect. 2.6.
GBI observations
The GBI observations consist of daily flux density measurements with the GBI
within the GBI-NASA Monitoring Program. LS 5039 was observed at the frequen-
cies of 2.25 and 8.3 GHz (13.1 and 3.6 cm wavelength, respectively) during two
periods. The first one spans 340 days, from 1998 September 17 to 1999 August 23
(MJD 51073.9–51413.9). The corresponding data set, hereafter Set1, contains 284
entries, with 282 successful flux density measurements at both frequencies. We show
the flux density lightcurves of the Set1 data in the top panel of Fig. 2.4, while the
spectral index α (where Sν ∝ ν +α ) computed between the two frequencies is shown
in the bottom panel. The second period of observations spans 24 days, from 2000
September 12 to October 6 (MJD 51799.0–51822.9). The corresponding data set,
hereafter Set2, contains 24 successful flux density measurements at both frequen-
cies. The flux density lightcurves and spectral index for the Set2 data are shown in
Fig. 2.5. Since there is a ∼ 1 year gap between both data sets, we have preferred to
analyze them separately.
Set1 data
An analysis of the first ∼ 4 months of GBI observations was reported in Ribó et al.
(1999), while Clark et al. (2001) already analyzed this data set.
The mean GBI flux densities and standard deviations are S2.25 GHz = 31.3 (±6.2)
10 The Green Bank Interferometer is a facility of the USA National Science Foundation operated
by NRAO in support of the NASA High Energy Astrophysics programs.

2.4. The radio counterpart: NVSS J182614−145054 43
Flux density [mJy]
Spectral index
50
40
30
20
10
0
0
−1
−2
GBI−NASA Monitoring Program (Set1 data)
2.25 GHz
8.3 GHz
−3
51050 51100 51150 51200 51250 51300 51350 51400 51450
Modified Julian Date [JD−2400000.5]
Figure 2.4: Top: GBI radio light curves of LS 5039 Set1 data at the frequencies of 2.25
and 8.3 GHz. Representative ± 1σ error bars have been plotted for the first data point.
Bottom: The corresponding spectral index. Error bars are also ± 1σ.
mJy and S8.3 GHz = 14.8 (±5.7) mJy. It must be noted that the 1σ errors of the
individual data points are 4 and 6 mJy, respectively. Hence, the source is detected
at the ∼ 8σ level at 2.25 GHz, and only at the ∼ 2.5σ level at 8.3 GHz. However,
the GBI is noise dominated below 10 mJy. Therefore, as pointed out by Clark
et al. (2001), the mean and standard deviation at 8.3 GHz should be estimated
with a Gaussian distribution truncated at 10 mJy. The result of this analysis is
S8.3 GHz = 13.7 (±6.7) mJy, indicating that LS 5039 is detected, on average, at the
∼ 2.0σ level at this frequency, preventing any reasonable variability analysis.
A careful inspection of the 2.25 GHz data in Fig. 2.4 reveals that the typical
day-to-day variability does not exceed ∼ 30 %. Ribó et al. (1999) noted that there
could be some exceptions to this behavior, for example around MJD 51075, 51086

44 Chapter 2. Multiwavelength approach to LS 5039
and 51176, when the flux density of LS 5039 seemed to have varied by more than a
factor of ∼ 2 on less than one day. However, inspection of the GBI operational notes
reveals that technical problems such as cryogenics warming, or weather problems
such as snow on the dishes, were responsible of this behavior. The apparent bump
between MJD 51180 and MJD 51200 is also a result of cryogenics warming. Hence,
it seems that most of the features in the lightcurve are due to technical problems,
because of the relatively low flux density of the source.
Nevertheless, since Pooley et al. (1999) could detect the 5.6 d orbital period
of Cygnus X-1 in similar radio lightcurves, we have performed a timing analysis of
the LS 5039 lightcurves at both frequencies. Given the span of the observations,
the search was restricted between 2 and 100 d. The methods employed were the
Phase Dispersion Minimization (PDM) (Stellingwerf 1978) and the CLEAN algo-
rithm (Roberts et al. 1987). Unfortunately, no convincing period was detected in
this process. In particular, we do not detect the 4.117 d orbital period recently
found by McSwain et al. (2001) from radial velocity measurements. Folding of the
data with such a period does not reveal any significant difference on radio emission
along the orbit. Since the orbit is quite eccentric (e=0.41), and this is a wind fed
system (where the primary does not fill its Roche lobe), we expect the accretion rate
onto the compact object to change significantly between periastron and apastron.
Hence, the non detection of the orbital period in the GBI data can be attributed to
its relatively high noise due to several facts (intrinsic measurement noise, technical
problems, weather conditions, etc.).
Although the 8.3 GHz data is dominated by noise, we can extract some infor-
mation about the spectral index. The weighted mean of individual spectral indices
is found to be α = −0.4 ± 0.3. On the other hand, if we use the mean flux densities
at both frequencies we obtain a spectral index of α = −0.6 ± 0.4. These value
are in good agreement with the results presented in Sect. 2.4.2 (Martí et al. 1998)
obtained a few months before, thus suggesting that the non-thermal radio spectrum
is a persistent property of the source.
Set2 data
This data set is not useful for timing analysis or other purposes due to its limited
time span and total number of points. However, we have explicitly included the

2.4. The radio counterpart: NVSS J182614−145054 45
Flux density [mJy]
Spectral index
100
80
60
40
20
0
0
−1
−2
GBI−NASA Monitoring Program (Set2 data)
2.25 GHz
8.3 GHz
−3
51795 51800 51805 51810 51815 51820 51825
Modified Julian Date [JD−2400000.5]
Figure 2.5: Same as Fig. 2.4 for the Set2 data. The flare is probably false (see text).
light curve in Fig. 2.5 to comment on the intriguing flare seen at 8.3 GHz around
MJD 51803. Although there could be an increase in flux density at 2.25 GHz, thus
supporting a possible outburst, the GBI operational notes reveal that spikes were
occasionally occurring at that time, and it is stated that single high scans should be
ignored. Hence, we conclude that the flare shown in Fig. 2.5 is probably false.
Conclusions
LS 5039 is a persistent radio source. In the GBI monitoring it is clearly detected
at 2.25 GHz, with a mean flux density of ∼ 31 mJy, and marginally detected at
8.3 GHz, with a mean flux density of ∼ 14 mJy. The spectral index of the source is
α ∼ −0.5, confirming earlier results. In addition to the negative spectral index, the
brightness temperature estimates for LS 5039 clearly yield to non-thermal values,

46 Chapter 2. Multiwavelength approach to LS 5039
hence supporting the mechanism of synchrotron radiation as the responsible for the
detected radio emission.
Excluding false data due to technical or weather problems, LS 5039 does not
show day-to-day variability larger than ∼ 30%, and no strong radio outbursts have
been detected. No orbital period is detected in the GBI data in the range between
2 and 100 days. In general terms, the observed radio behavior confirms the earlier
suggestions by Martí et al. (1998) concerning the persistent and moderately variable
nature of the radio emission.
2.4.4 VLBA observations, discovery of a microquasar
As we have seen in Sect. 2.4.2, LS 5039 appeared unresolved to the VLA in A
configuration, i.e., with a size ≤ 0.1 ′′ . On the other hand, a lower limit to the source
size of ≥ 0.2 mas was derived to prevent catastrophic inverse Compton losses to
occur. Therefore, we decided to conduct observations using the Very Long Baseline
Interferometry (VLBI) technique, aimed to reveal the mas structure of the source.
Observations and results
We observed 11 LS 5039 at 5 GHz frequency (6 cm wavelength) on 1999 May 8
(MJD 51306) with the 10×25-m antennas of the NRAO 12 Very Long Baseline Array
(VLBA) and the 27×25-m antennas of the VLA in its phased array mode in its D
configuration. In this mode, the VLA has the same sensitivity as a dish of ∼ 115 m
diameter, and provides sensitive baselines with the VLBA antennas. The VLA
phasing process was successful, and it worked without problems as another VLBI
station. On the other hand, in theory the VLA could also record data on its own.
This would allow to perform a very sensitive VLA map and to follow the radio flux
density during the VLBA run. However, there was an air conditioning failure in the
VLA correlator, and it could not record data.
The source 3C 345 was used as a fringe-finder, whereas J1733−1304 was the
phasing source for the VLA. We could not use the VLBI phase-reference technique
11 Published in Paredes, J. M., Martí, J., Ribó, M., & Massi, M. 2000, Science, 288, 2340.
12 The National Radio Astronomy Observatory is a facility of the USA National Science Foun-
dation operated under cooperative agreement by Associated Universities, Inc.

2.4. The radio counterpart: NVSS J182614−145054 47
MilliARC SEC
4
2
0
-2
-4
-6
VLBA
6 4 2 0 -2 -4 -6
MilliARC SEC
Figure 2.6: VLBA self-calibrated image of LS 5039 at 5 GHz. Contours are 6, 8, 10,
12, 14, 16, 18, 20, 25, 30, 40, and 50 times 0.085 mJy beam −1 , the rms noise. The
synthesized beam, shown at the bottom right corner, is 3.4 mas×1.2 mas, with position
angle in 0 ◦ .
because there was no obvious nearby calibrator to be used. Data were recorded
in VLBA mode v6cm-256-8-2-L, i.e., at 6 cm wavelength, at 256 Mbps, using 8
baseband channels, with 2-bit sampling at left hand circular polarization, yielding
a full bandwidth of 64 MHz. The data were correlated at the VLBA correlator in
Socorro, New Mexico, using an integration time of 4 s. The LS 5039 coordinates
used for correlation were those obtained after the VLA A configuration observations,
i.e., α = 18 h 26 m 15.056 s , δ = −14 ◦ 50 ′ 54.24 ′′ , expressed in J2000.0 ICRS. The
output data were then calibrated using standard procedures in unconnected radio
interferometry within the aips and difmap software packages.
The resulting pattern of the observed visibility amplitudes, decaying as a function
of baseline length, indicated that LS 5039 had structure at milliarcsecond scales. The

48 Chapter 2. Multiwavelength approach to LS 5039
Flux density at 2.25 GHz [mJy]
80
60
40
20
VLBA run 1999 May 8
0
51290 51295 51300 51305 51310 51315 51320
Modified Julian Date [JD−2400000.5]
Figure 2.7: GBI radio monitoring of LS 5039, at 2.25 GHz, during the weeks before and
after the date of our VLBA observations, indicated by the vertical bar.
final synthesis map (see Fig. 2.6) shows a two-sided jet emerging from a central core.
A deconvolved angular size of about 2 mas is estimated for the core. The jets extend
over 6 mas on the plane of the sky, oriented along a position angle of ∼ 125 ◦ with
respect to the North, and they account for 20% of the total 16 mJy flux density. At
a distance of ∼ 3 kpc the total projected length of the jets is ∼ 20 AU.
Discussion
To obtain some order of magnitude estimates, we will assume that the overall size
of the radio source is approximately 6 mas×2 mas. This implies a high brightness
temperature of ∼ 9.4 × 10 7 K, indicative of synchrotron radiation. As has been seen
in Sects. 2.4.2 and 2.4.3, the LS 5039 radio spectrum as a function of frequency
often displays a negative spectral index which is in agreement with a non-thermal
optically thin emission mechanism (Martí et al. 1998, Ribó et al. 1999).
The detection of jets occurred at a time when the source was at its typical
persistent level of radio emission, and only moderately variable, as inferred from
concurrent radio monitoring by the Green Bank Interferometer (GBI) (see Fig. 2.7).
The absence of any precursor outburst for the radio jets strongly suggests that they
are always present and continuously emanating from the core.

2.4. The radio counterpart: NVSS J182614−145054 49
The data can be model fitted with three components: a central core of 12.8 mJy,
a southeast component located at an angular distance of 3.8 ± 0.2 mas from the
core and with a flux density of 2.1 ± 0.1 mJy, and a northwest component located
at 2.8 ± 0.2 mas from the core having 1.0 ± 0.1 mJy. It seems reasonable to assume
that the different distance from the components to the core and the different flux
densities reflect relativistic effects (Rees 1966, Blandford & Königl 1979, see also
Sect. 1.2.1). Hence, using the distances to the core and
β cos θ = µa − µr
µa + µr
= da − dr
da + dr
(2.5)
we obtain β cos θ = 0.15 ± 0.04. On the other hand, based on the flux density
asymmetry we can use the equation
β cos θ > (Sa/Sr) 1/(k−α) − 1
(Sa/Sr) 1/(k−α) + 1
(2.6)
to obtain a lower limit on β cos θ. Assuming that the jet flow is continuous, k = 2,
and using α = −0.5±0.4 we obtain β cos θ 0.15±0.03. Both results are in excellent
agreement. If we consider that the jet is composed by discrete condensations, k = 3,
then β cos θ 0.11±0.02. The value deduced by the use of the distance asymmetry,
β cos θ = 0.15 ± 0.04, allows us to find a lower limit for the jet velocity of β ≥
0.15 ± 0.04 and an upper limit for the ejection angle of θ ≤ 81 ◦ ± 2 ◦ .
Additional information on the source energetics can be obtained by assuming
energy equipartition between the relativistic electrons and the magnetic field (Pa-
cholczyk 1970). We are forced to use the overall source parameters observed, because
not enough information is yet available for appropriate calculations in the rest frame
of the ejecta. The corresponding results are nevertheless expected to be within an
order of magnitude for a mildly relativistic system. Under these assumptions, the
observed radio properties of LS 5039 imply a total energy content in relativistic
electrons of Ee ∼ 5 × 10 39 erg, with an equipartition magnetic field of ∼ 0.2 G.
Conclusions
The VLBA observations showed that LS 5039 exhibited a two-sided jet morphology.
The brightness and distance to the core asymmetry of the detected jet components
implies, assuming that they are intrinsically equal, that the jets have a bulk motion
higher than 0.1c, thus revealing the microquasar nature of LS 5039.

50 Chapter 2. Multiwavelength approach to LS 5039
2.4.5 EVN and MERLIN observations,
confirmation of persistent relativistic radio jets
The persistent radio emission of LS 5039 suggests that the jet is always present.
With the aim to confirm this hypothesis, and also to detect the jet at larger angular
scales than those imaged with the previous Very Long Baseline Array (VLBA) mea-
surements, we observed this microquasar with the European VLBI Network (EVN)
and the Multi-Element Radio-Linked Interferometer Network (MERLIN).
Observations and results
We observed 13 LS 5039 simultaneously with the EVN 14 and MERLIN 15 on 2000
March 1 (3:20–7:10 UT, MJD 51604.2) at 5 GHz. Single dish flux density measure-
ments were carried out with the MPIfR 100 m antenna in Effelsberg. We scheduled
the observation introducing the phase-reference calibrator J1823−1228, the fringe-
finder DA 193 and the flux density calibrator 3C 286. Unfortunately, the phase-
reference calibrator was not detected and could not be used during the correlation.
The EVN observations were performed with the array described in Table 2.2.
Data were recorded in MkIV mode with 2-bit sampling at 256 Mbps at left hand
circular polarization, yielding a full bandwidth of 64 MHz (v6cm-256-8-2-L mode).
The data were correlated at the Joint Institute for VLBI in Europe (JIVE). An inte-
gration time of 4 s was used. Interferometer fringes for LS 5039 were detected in all
baselines. A later fringe fitting of the residual delays and fringe rates was performed
within aips. A first a priori visibility amplitude calibration was performed using
antenna gains and system temperatures measured at each antenna. We averaged
in frequency the data and exported them to be imaged and self-calibrated into the
difference mapping software difmap. The final imaging was carried out on those
data after editing and averaging of the visibilities in 32 s blocks.
MERLIN recorded data with 2-bit sampling at dual polarization and a total
13Published in Paredes, J. M., Ribó, M., Ros, E., Martí, J., & Massi, M. 2002, A&A, 393, L99.
14The European VLBI Network is a joint facility of European, Chinese and other radio astronomy
institutes funded by their national research councils.
15MERLIN is operated as a National Facility by the University of Manchester at Jodrell Bank
Observatory on behalf of the UK Particle Physics & Astronomy Research Council.

2.4. The radio counterpart: NVSS J182614−145054 51
Table 2.2: EVN array used in the LS 5039 observations.
Antenna Code Location Diameter DPFU (a) Tsys (b)
[m] [K Jy −1 ] [K]
Effelsberg EB Germany 100 1.47 27
Jodrell Bank JB U.K. 25 0.11 40
Cambridge CM U.K. 32 0.21 38
Westerbork WB Netherlands 14×25 1.0 67
Medicina MC Italy 32 0.15 53
Noto NT Italy 32 0.16 47
Toruń TR Poland 32 0.15 31
(a) Degrees per flux unit. (b) Best value during the experiment.
32 MHz bandwidth. We analyzed the left hand circular polarization data excluding
1 MHz channel at both edges of the band, yielding a final bandwidth of 14 MHz.
The correlator integration time was of 4 s. Standard reduction and imaging analysis
were carried out using aips and difmap.
To complement the amplitude calibration, we interleaved cross-scans (in azimuth
and elevation) with the 100 m Effelsberg antenna to measure the radio source flux
densities (A. Kraus, private communication). We fitted a Gaussian function to the
flux-density response for every cross-scan, and we averaged the different Gaussians.
We linked the flux density scale by observing primary calibrators such as 3C 286,
3C 48, or NGC 7027 (see e.g. Kraus 1997 or Peng et al. 2000).
Unfortunately, the flux density monitoring with the Effelsberg antenna was not
reliable due to contamination effects by the galactic plane diffuse emission and/or by
the close SNR G016.8−01.1. However, inspection of the shortest MERLIN baselines
reveals a constant flux density during the full run. That allowed us to image the
source directly without splitting the data in time blocks. We present the obtained
EVN and MERLIN images in Figs. 2.8 and 2.9, respectively. The MERLIN image is
presented convolved with a circular beam, equivalent in solid angle to the interfero-
metric synthesized beam of 142×46 mas at position angle (P.A.) 1.15 ◦ . Our images
clearly show that LS 5039 has a two-sided jet emanating from a central core. In
both images the southeast (SE) jet is brighter and larger than the northwest (NW)
one, as can be seen in Table 2.3, where we have quoted the flux densities of the core
and the jets and the lengths and P.A. of the jets. Their total lengths are ∼ 60 mas
in the EVN image and ∼ 300 mas in the MERLIN one.

52 Chapter 2. Multiwavelength approach to LS 5039
Figure 2.8: EVN self-calibrated image of LS 5039 at 5 GHz obtained on 2000 March 1.
Axes units are in mas. The synthesized beam, plotted in the lower left corner, has a size of
7.60 mas×6.96 mas in P.A. of −14 ◦ . The first contour corresponds to 0.3 mJy beam −1 ,
while consecutive ones scale with 3 1/2 .
Table 2.3: Flux densities at 5 GHz, length and P.A. of the LS 5039 structure features
as detected by EVN and MERLIN.
EVN MERLIN
S 5 GHz Length P.A. S 5 GHz Length P.A.
[mJy] [mas] [ ◦ ] [mJy] [mas] [ ◦ ]
Core 29.3 — — 31.6 — —
NW jet 2.6 24 −42 4.0 128 −29
SE jet 3.3 34 140 4.2 174 150
Note: the errors in flux density, lenght and P.A. are 0.1 mJy beam −1 , 2 mas and 4 ◦
for the EVN features and 0.4 mJy beam −1 , 12 mas and 5 ◦ for the MERLIN ones.

2.4. The radio counterpart: NVSS J182614−145054 53
Figure 2.9: MERLIN self-calibrated image of LS 5039 at 5 GHz obtained on 2000
March 1. Axes units are in mas. The convolving circular beam, plotted in the lower left
corner, has a diameter of 81 mas. The first contour corresponds to a flux density of
1 mJy beam −1 , while consecutive ones scale with 3 1/2 .
Discussion
These results confirm the existence of a two-sided radio jet in LS 5039 reported in
previous VLBA 5 GHz observations (Paredes et al. 2000, see also Sect. 2.4.4). This
source does not present any strong outburst or, at least, none has been detected
during the eleven month monitoring carried out by the Green Bank Interferome-
ter between 1998 September 17 and 1999 August 23 (Clark et al. 2001, see also
Sect. 2.4.3). On the other hand, inspection of the RXTE All Sky Monitor data
(Levine et al. 1996) reveals that the X-ray flux was at the typical low level (Ribó
et al. 1999) at the epoch of the EVN and MERLIN observations and several weeks
before. All this suggests that the jets are persistent, as the VLBI images obtained
up to now seem to indicate.
In the observations reported here, the jets extend further away than the ∼ 6 mas

54 Chapter 2. Multiwavelength approach to LS 5039
of the VLBA observations at the same frequency. In all the images the jets have
similar position angles, ∼ 125 ◦ in the VLBA image, ∼ 140 ◦ in the EVN image, and
∼ 150 ◦ in the MERLIN image. These results suggest a bending of the jets with
increasing distance from the core and/or precession.
The brightness and length asymmetry of the jet components may involve special
relativity effects (Rees 1966, Blandford & Königl 1979, see also Sect. 1.2.1). Hence,
assuming that this is the reason for the detected length asymmetry of the jets, we
can estimate some parameters by using the following equation:
β cos θ = µa − µr
µa + µr
= da − dr
da + dr
. (2.7)
Although we do not know the epoch of ejection of the terminal plasma, we can
identify da and dr with the lengths of the approaching and receding jet, respectively.
Using Eq. 2.7 and the EVN values in Table 2.3 we obtain β cos θ = 0.17 ± 0.05,
and hence β > 0.17 ± 0.05 and θ < 80 ◦ ± 3 ◦ . For the MERLIN values we obtain
β > 0.15 ± 0.06 and θ < 81 ◦ ± 3 ◦ . These values are similar to those previously
derived from the VLBA image, of β > 0.15 ± 0.04 and θ < 81 ◦ ± 2 ◦ . We have not
considered an eventual larger size for the NW jet with a brightness below the image
noise level. McSwain & Gies (2002) have recently proposed an inclination of i 30 ◦
for LS 5039. If we assume that the jet is perpendicular to the accretion disk, and
that the disk lies in the orbital plane of the binary system, then θ = i = 30 ◦ , and
using the values from the EVN image, we obtain β = 0.20 ± 0.06, which indicates
a mildly relativistic jet. In this case, the apparent velocity and proper motion of
the approaching components would be 0.12c and 7.2 mas day −1 . For the receding
components we would have 0.085c and 5.1 mas day −1 .
The total length of the EVN and MERLIN jets is ∼ 60 and ∼ 300 mas, respec-
tively. Considering that the source is located at 2.9 kpc (Ribó et al. 2002), these
angular lengths translate into linear lengths in the plane of the sky of ∼ 175 and
∼ 870 AU, respectively. Assuming that θ = 30 ◦ we obtain intrinsic total lengths
of ∼ 350 and ∼1740 AU, respectively, and lengths of the SE jet arm of ∼ 200 and
∼1000 AU, respectively. Moreover, the jet width is smaller than one synthesized
beam even in the EVN image. This implies a jet half opening angle ≤ 6 ◦ .
Further discussion on the LS 5039 jets is reported in Sect. 2.8.

2.5. The optical/IR star: LS 5039 55
Conclusions
The EVN and MERLIN observations have confirmed the existence of relativistic
radio jets in LS 5039, as previously found in VLBA observations (Paredes et al.
2000, see also Sect. 2.4.4). Moreover, the recurrent detection of the jets strongly
suggests that in LS 5039 relativistic radio jets are always present.
2.5 The optical/IR star: LS 5039
The star LS 5039 was originally classified as the star number 5039 in the catalog
Luminous stars in the southern Milky Way, by Stephenson & Sanduleak (1971). It
is an early type star, O6.5V((f)), with a visual magnitude ∼ 11.2. In this section
we report photometric and spectroscopic results that have allowed to obtain the
spectral type, the distance and the orbital parameters of the binary system.
2.5.1 Photometry
Optical photometry of LS 5039 has been carried out by several groups. Here we
start by reporting photometry obtained by us, and then provide a compilation of
optical and near infrared photometry obtained by other groups.
Optical observations with the 1.5 m OAN telescope
Soon after the VLA runs presented in Sect. 2.4.2, we carried out optical CCD ob-
servations 16 at Calar Alto observatory (Spain). Our main goal was to search for
photometric variability that could evidence the active nature of LS 5039 as an X-
ray binary. We used the 1.52 m telescope of the Spanish Observatorio Astronómico
Nacional (OAN) from 1998 June 1 to 8. The Ritchey-Chrétien focus was available
together with a Tektronics TK1024AB chip. This provided a scale factor of 0.4 ′′ per
pixel and a 6.9 ′ × 6.9 ′ field of view. Images were acquired through the V RI Johnson
filters and they were reduced using standard procedures based on the iraf image
processing system. In Fig. 2.10 we show a 30 s exposure of the LS 5039 field in
16 Published in Martí, J., Paredes, J. M. & Ribó, M. 1998, A&A, 338, L71.

56 Chapter 2. Multiwavelength approach to LS 5039
DECLINATION (J2000)
-14 49 00
30
50 00
30
51 00
30
52 00
30
LS 5039 V band
LS 5039
ROSAT
53 00
18 26 20 15 10
RIGHT ASCENSION (J2000)
05
Figure 2.10: CCD image of LS 5039 obtained on 1998 June 7 with the 1.52 m OAN
telescope in the V Johnson filter. The circle with 35 ′′ radius represents the 90% confi-
dence position from the ROSAT PSPC pointed observations quoted in M97. The two
comparison stars used for differential photometry are labeled as C1 and C2.
the V band. Differential photometry was performed against two nearby comparison
stars (C1 and C2 in Fig. 2.10).
Based on several observations of Landolt (1992) standards, the adopted magni-
tudes of the comparison stars are V = 12.53, R = 12.01, I = 11.55 and V = 10.10,
R = 9.36, I = 8.68 for C1 and C2, respectively. This absolute photometry is accu-
rate to ±0.03 mag. Relative variations in the final photometric results of Table 2.4
can be nevertheless traced at the ±0.01 mag level.
From our OAN observations we propose that there may be day-to-day variations
with amplitudes of a few hundredths of magnitude. This appears specially evident
in the latest two days of Table 2.4. Here, the LS 5039 brightness decreased by about
0.04 mag in V and 0.05 mag in R and I from one day to the next. In doing so,
all the differences between the two comparison stars remained constant within less
than 0.01 mag, thus supporting the reality of this variation.
C1
C2

2.5. The optical/IR star: LS 5039 57
Table 2.4: Optical photometry of LS 5039.
Date Julian Day V R I
[JD−2400000]
1998 Jun 2 50966.5 11.35 — —
1998 Jun 3 50967.6 11.33 — —
1998 Jun 7 50971.6 11.35 10.64 9.90
1998 Jun 8 50972.5 11.39 10.69 9.95
Compilation of optical and near infrared observations
Optical and/or near infrared photometry of LS 5039 has been published in the past
by Drilling (1975, 1991), Lahulla & Hilton (1992), Kilkenny et al. (1993), Spencer
Jones et al. (1993), Martí et al. (1998, see Table 2.4) and Clark et al. (2001).
We have compiled all simultaneous broad band (at least 3 filters) measurements in
Table 2.5, where the initials and numbers refer to the cited papers. As can be seen,
all values are consistent with a long-term variability smaller than ∼ 0.1 mag in all
filters except in V , where it varies ∼ 0.2 mag. This suggests that no strong optical
variations appear to be present on timescales of years.
Nevertheless, Clark et al. (2001) also report broad band JHK photometry in the
period 1995–2000 that indicates variability up to ∼0.4 mag in H band and ∼0.5 mag
in K band. These authors also searched for medium-term variability in the B and V
bands during 21 nights spanning 43 days. They only could find aperiodic variability
at the hundredth of magnitude level, on single nights and from night to night, that
can be attributed to variations in mass loss rate or to non-radial pulsations.
Table 2.5: Compilation of optical and near infrared photometry of LS 5039.
Observation U B V R I J H K L
D91 12.02 ± 0.02 12.18 ± 0.01 11.23 ± 0.01 — — — — — —
LH92 11.97 ± 0.02 12.15 ± 0.02 11.20 ± 0.02 — — — — — —
K93, SJ93 — — 11.24 ± 0.01 10.59 ± 0.01 9.88 ± 0.01 9.02 ± 0.01 8.79 ± 0.02 8.57 ± 0.03
1996 Oct, C01 — 12.18 ± 0.02 11.33 ± 0.02 10.65 ± 0.02 9.87 ± 0.02 9.05 ± 0.02 8.75 ± 0.02 8.60 ± 0.02 8.69 ± 0.05
1998 Jun 7 M98 — — 11.35 ± 0.03 10.64 ± 0.03 9.90 ± 0.03 — — — —
1998 Jun 8, M98 — — 11.39 ± 0.03 10.69 ± 0.03 9.95 ± 0.03 — — — —
2000 Sep, C01 — 12.17 ± 0.03 11.32 ± 0.01 10.61 ± 0.01 9.91 ± 0.01 — — — —

58 Chapter 2. Multiwavelength approach to LS 5039
Apart from optical photometry in the Johnson filters, Strömgren and Hβ pho-
tometry have also been conducted. Kilkenny & Whittet (1993) reported the fol-
lowing Strömgren photometry for LS 5039: V = y = 11.223 ± 0.014, (b − y) =
0.766 ± 0.008, (v − b) = 0.624 ± 0.010, (u − b) = 1.204 ± 0.014, m1 = −0.142 ± 0.014
and c1 = −0.044 ± 0.022. They also computed the following de-redenned val-
ues: V = y = 11.223 ± 0.014, (b − y) = 0.743 ± 0.008, (v − b) = 0.597 ± 0.010,
(u − b) = 1.202 ± 0.014, m1 = −0.146 ± 0.014 and c1 = −0.008 ± 0.022. Finally,
Kilkenny (1993) reports the following Hβ photometry: β = 2.576 ± 0.010.
2.5.2 Spectral type
Kilkenny (1993) was the first one to suggest a spectral type for LS 5039. He classified
it as an O7IV star based on Strömgren and Hβ photometry. Later on, Motch et al.
(1997) took a low resolution λλ 3800–4700 (all λ are expressed in ˚A) spectrum and
a medium resolution red spectrum. Among O stars, those showing N iii in emission
at λλ 4630–34 and He ii strongly in emission at λ 4686 are classified as Of stars.
When the N iii is in emission and the He ii is weakly in absorption or emission they
are classified as O(f). Finally, when the N iii is in emission and the He ii is strongly
in absorption they are classified as O((f)). Motch et al. (1997) reported an O7V
star with well marked He ii λ 4684 absorption and some evidence for weak N iii
λλ 4634–4642 emission, and therefore suggested an O7V((f)) classification. No Hα
emission was present in their medium resolution red spectrum.
Clark et al. (2001) obtained higher resolution optical and near infrared spectroscopy
of LS 5039. All Balmer lines were in absorption, preventing hence a classification
as an Oe/Be star. The red-end spectrum (λλ 5650–7500) contained interstellar
lines and helium lines in absorption. The blue-end spectrum (λλ 4050–4950)
showed hydrogen Balmer series and He ii lines. The ratio between He ii λ 4541 and
He i λ 4471 corresponded to a spectral type O6.5, while the presence of He ii 4686
strongly in absorption indicated that LS 5039 is still in the main sequence. Finally,
the presence of very weak emission from N iii λλλ 4634–40–42 allowed them to add
the ((f)) classification. Therefore, they classified LS 5039 as an O6.5V((f)) star. Furthermore,
their K band (2.05–2.20 µm) and H band (1.49–1.72 µm) spectroscopy
was perfectly consistent with such classification, and they stated that there was no
evidence from any additional emission feature that could be attributed to emission
from a jet or accretion disk.

2.5. The optical/IR star: LS 5039 59
2.5.3 Distance and its uncertainty
Motch et al. (1997) proposed a distance of 3.1 kpc to LS 5039 based on the star
color excess, and stated that this estimate could have a large uncertainty. This value
was computed assuming an O7V((f)) spectral type. However, as we have just seen,
recent observations by Clark et al. (2001) show that the optical companion has a
spectral type of O6.5V((f)). On the other hand, old calibrations for the intrinsic
color index and absolute magnitude (Johnson 1966, Deutschman et al. 1976) were
used by Motch et al. (1997). Hence, we have performed a new estimate of the
distance taking into account the new spectral type and more recent calibrations.
The optical photometry available up to now and containing at least B and V
magnitudes (see Table 2.5) comes from Drilling (1991), Lahulla & Hilton (1992)
and Clark et al. (2001), and is listed in Table 2.6. Using an intrinsic color index
of (B − V )0 = −0.30 ± 0.02 for an O6.5V star (Schaerer et al. 1996, Lejeune &
Schaerer 2001), we can compute the color excess E(B − V ) for all the observations.
Finally, using the relationship AV = [3.30 + 0.28 (B − V )0 + 0.04 EB−V ] E(B − V )
(Schmidt-Kaler 1982) and MV = −4.99 ± 0.3 for an O6.5V star (Vacca et al. 1996)
we obtain the distance estimates listed in Table 2.6. The weighted mean of these
values is 2.9 ± 0.3 kpc, and is the distance to LS 5039 used throughout this work.
Finally, assuming that the Sun is at 8.5 kpc from the galactic center, we obtain a
galactocentric distance of 5.8 ± 0.3 kpc for LS 5039.
Table 2.6: Distance estimates to LS 5039 derived from optical photometry.
V B − V E(B − V ) AV d [kpc]
11.23 ± 0.01 (a) 0.95 ± 0.01 (a) 1.25 ± 0.02 4.1 ± 0.1 2.68 ± 0.06
11.20 ± 0.02 (b) 0.95 ± 0.02 (b) 1.25 ± 0.03 4.1 ± 0.1 2.64 ± 0.06
11.33 ± 0.02 (c) 0.85 ± 0.02 (c) 1.15 ± 0.03 3.8 ± 0.1 3.26 ± 0.07
11.32 ± 0.01 (d) 0.85 ± 0.02 (d) 1.15 ± 0.03 3.8 ± 0.1 3.25 ± 0.07
(a) Drilling (1975, 1991).
(b) Lahulla & Hilton (1992).
(c) 1996 October observations by Clark et al. (2001).
(d) 2000 September observations by Clark et al. (2001).

60 Chapter 2. Multiwavelength approach to LS 5039
2.5.4 Radial velocity curve
Optical spectroscopy aimed to reveal the radial velocity curve of LS 5039 was carried
out by McSwain et al. (2001). They performed the observations during three runs
in 1998 August, 1999 June and 2000 October using the 0.9 m coudé feed telescope at
Kitt Peak National Observatory, that provided a resolution R = λ/δλ = 9500. The
obtained radial velocity data appeared to have minima every ∼ 4 days. Therefore,
the data set was then analyzed in the search for an orbital period, and the following
orbital elements were derived after a least-squares fit: P = 4.117 ± 0.011 d, T =
JD 2451822.12±0.09, K = 14.7±0.9 km s −1 , V0 = 4.6±0.5 km s −1 , e = 0.41±0.05,
w = 217 ± 9 ◦ , rms = 3.3 km s −1 , f(m) = 0.00103 ± 0.00020 M⊙, and a1 sin i =
1.09 ± 0.07 km s −1 . They also obtained a projected rotational velocity of V sin i =
131 ± 6 km s −1 . The epoch T corresponds to the time of periastron, and the epoch
of inferior conjunction of the optical star (which would correspond to the time of an
X-ray eclipse if one occurs) is 0.25 days later, corresponding to a phase of 0.06. We
show in Fig. 2.11 the radial velocity curve obtained by McSwain et al (2001).
RADIAL VELOCITY (km s -1
)
30
20
10
0
-10
-20
-30
0.0 0.5 1.0
ORBITAL PHASE
Figure 2.11: Radial velocity measurements and orbital solution plotted against orbital
phase. Phase zero corresponds to periastron (from McSwain et al. 2001).

2.5. The optical/IR star: LS 5039 61
Since no Hα emission was seen, the star could not fill its Roche lobe even at
periastron, and considering a radius of R = 10 R⊙ for an O7V star, they directly
inferred a minimum mass of 15 M⊙ for the optical star, compatible with the expected
mass of 36 M⊙. On the other hand, in order to avoid breakup speed of the star, the
lowest inclination acceptable would be i 9 ◦ , and this would limit the maximum
mass for the compact object to be MX 8 M⊙.
We note that then it was unknown the improvement on the spectral type classifi-
cation as an O6.5V((f)) star. However, the derived values do not change significantly.
In fact, considering a mass of 40 M⊙ for an O6.5V((f)) star, as we have discussed
in Sect. 2.5.3, we obtain a maximum mass for the compact object of 9 M⊙.
Spectroscopic observations with the INT
With the aim to confirm the results obtained by McSwain et al (2001), we have
conducted new medium resolution spectroscopic observations of LS 5039 during 9
consecutive nights. We have observed 17 LS 5039 using the Intermediate Dispersion
Spectrograph (IDS) attached to the 2.5 m Isaac Newton Telescope (INT) 18 on the
nights of 2002 July 23–31. A total of 96 spectra were obtained with the combination
of the 235 mm camera and the R900V grating, which provided a useful wavelength
coverage (free from vignetting) of λλ 3900–5500. The seeing was variable (1 ′′ –2 ′′ )
during our run and we used a 1.2 ′′ slit which resulted in a resolution of 83 km s −1
(FWHM). In addition, we also obtained five spectra of LS 5039 with the holografic
grating H2400B for the purpose of measuring the rotational broadening (V sin i)
of the companion’s absorption lines. The spectral resolution of these spectra was
30 km s −1 and we varied the central wavelength in order to fully cover the range
λλ 3900–5100, where most of the prominent He i and He ii lines lie. The spectral
type standards HR 8622 (O9V spectral type) and HD 168075 (O7V((f)) spectral
type) were also observed with the same instrumental configurations. A complete log
of the observations is provided in Table 2.7.
The images were de-biased and flat-fielded, and the spectra subsequently ex-
tracted using conventional optimal extraction techniques in order to optimize the
signal-to-noise ratio of the output (Horne 1986). CuAr and CuNe comparison lamp
17 Casares, J., Ribó, M., Paredes, J. M., & Martí, J. 2002, in preparation.
18 The INT is operated on the island of La Palma by the Royal Greenwich Observatory in the
Spanish Observatorio del Roque de Los Muchachos of the Instituto de Astrofísica de Canarias.

62 Chapter 2. Multiwavelength approach to LS 5039
Table 2.7: Log of the spectroscopic observations carried out with the INT.
Date Object Exp. Time Wave. Range Dispersion
(2002 July) [s] [λλ] [˚A pix −1 ]
23 LS 5039 1×600 4045–4910 0.85
24 ,, 1×600 ,, ,,
25 ,, 1×750 ,, ,,
26 ,, 1×600 ,, ,,
27 ,, 5×300 3900–5500 0.63
” ,, 7×600 ,, ,,
” HR 8622 2×60 ,, ,,
28 LS 5039 3×300 ,, ,,
” ,, 6×600 ,, ,,
29 LS 5039 1×300 ,, ,,
” ,, 12×600 ,, ,,
” ,, 11×900 ,, ,,
30 ,, 1×300 ,, ,,
” ,, 17×600 ,, ,,
” ,, 4×900 ,, ,,
” HD 168075 1×600 ,, ,,
31 LS 5039 1×300 ,, ,,
” ,, 23×600 ,, ,,
” ,, 3×600 3900–4550 0.23
” HD 168075 1×200 ,, ,,
” LS 5039 2×600 4550–5100 0.22
” HD 168075 1×200 ,, ,,
images were obtained every 15–30 minutes, and the λ-pixel scale was derived through
4/6th-order polynomial fits to 53/86 lines (depending on the set-up), resulting in an
rms scatter < 0.04 ˚A. The calibration curves were interpolated linearly in time.
All the spectra were rectified, by fitting a low-order spline to the continuum, and
rebinned into a uniform velocity scale of 83 km s−1 . The continuum was subsequently
subtracted and every individual spectrum of the target was then cross-correlated
with the template HD 168075, after masking the interstellar absorption lines and
bands at λ 3934 (CaII K), λ 4430, λ 4501, λ 4726, λ 4762, λ 4885, λ 5449 and
λλ 5487–5550. The resulting velocities are presented in Fig. 2.12, where it is clearly
seen a night-to-night variability that suggests a periodicity of 4 d, in agreement
with the results of McSwain et al. (2001).

2.5. The optical/IR star: LS 5039 63
Radial velocity [km/s]
60
40
20
0
−20
−40
52479.0 52481.0 52483.0 52485.0 52487.0
Modified Julian Date [JD−2400000.5]
Figure 2.12: Radial velocity measurements obtained with the INT on 2002 July 23–31.
Although our data only span 9 nights, we have performed a timing analysis
in order to search for possible periods, in particular the proposed ∼ 4 d orbital
period. The methods employed were the Phase Dispersion Minimization (PDM)
(Stellingwerf 1978) and the CLEAN algorithm (Roberts et al. 1987). The two
obtained periodograms show a clear period at 4.0 ± 0.2 d, and the corresponding
1-day alias at 1.33 ± 0.3 d with a similar significance level.
We note that an O6.5V star would not fit in the Roche lobe for a period below
∼ 2 d. A possibility for the binary to have a ∼ 1.3 d orbital period would be if
LS 5039 were a subdwarf. However, with such an orbital period, we would expect
to find a completely circularized orbit (i.e., e = 0), which is not the case if we fold
our measurements with the ∼ 1.3 d period. Therefore, the 4.0 ± 0.2 d period seems
to be the correct one, thus confirming the earlier results by McSwain et al. (2001).

64 Chapter 2. Multiwavelength approach to LS 5039
Radial velocity [km/s]
30
20
10
0
−10
−20
−30
0.0 0.5 1.0
Orbital phase
Figure 2.13: Radial velocity measurements obtained with the INT on 2002 July 23–31,
folded using the ephemeris of McSwain et al. (2001).
We show in Fig. 2.13 our obtained data with the INT folded with the ephemeris
of McSwain et al. (2001), and plotted using the same X, Y intervals as in Fig. 2.11.
While with those ephemeris the minimum takes place at phase 0.96 (see Fig. 2.11),
here it occurs at phase ∼0.1–0.2. This indicates that the period used is not accurate
enough, which is compatible with the error quoted in the McSwain et al. (2001)
ephemeris (possible improvement of ephemeris is discussed in Casares et al. 2002).
In order to obtain the rotational broadening, we have followed the technique
applied to A0620-00 by Marsh et al. (1994) and described in their paper. Essentially,
we subtract different broadened versions of a O7V((f)) template (HD 168075) from
the Doppler corrected sum of our LS 5039 spectra and perform a χ 2 test on the
residuals. The O7V spectrum was broadened by convolution with the rotational
profile of Gray (1976) which assumes a linearized limb darkening coefficient ɛ. We
have taken ɛ=0.20 which is appropriate for the stellar parameters of our star (Teff
35000 K, log g = 4) in the B-band. We performed the analysis independently for
the two groups of spectra at different resolutions, 83 and 30 km s −1 , and we find

2.6. The X-ray counterpart: RX J1826.2−1450 65
that the minimum χ 2 is achieved for the spectrum broadened by σbroad = 105 ± 1
and 126±12 km s −1 , respectively. Since the template has a rotational broadening of
σHD 168075 = 79 ± 3 km s −1 (Penny 1996), we need to sum this quadratically to σbroad
in order to get the true rotational broadening of LS 5039. Therefore, we obtain
131 ± 3 and 149 ± 12 km s −1 for the two resolutions. These numbers are in excellent
agreement with the value V sin i = 131 ± 6 km s −1 obtained McSwain et al. (2001).
2.6 The X-ray counterpart: RX J1826.2−1450
The X-ray binary name that appears in the Liu et al. (2000) catalog of HMXBs,
which always comes from X-ray satellites, is RX J1826.2−1450, after the ROSAT All
Sky Suvey (RASS). However, in the definitive version of the ROSAT Bright Source
Catalog (RBSC) it is listed as 1RXS J182615.1−145034. In this section we present
the long-term X-ray lightcurve obtained by ASM/RXTE, pointed observations by
PCA/RXTE and pointed observations by BeppoSAX.
2.6.1 The ASM/RXTE data
As we have mentioned in Sect. 2.4.3, LS 5039 was included at our request in the
catalog of the All Sky Monitor (ASM), onboard the Rossi X-ray Timing Explorer
(RXTE), in order to follow its behavior in the X-ray domain. An analysis of the
first ∼ 33 months of ASM/RXTE data was presented in Ribó et al. (1999). Here
we report ASM/RXTE 19 observations spanning almost six and a half years (MJD
50136.78–52501.41), from 1996 February to 2002 August. The dataset contains more
than 1800 daily flux measurements in the energy range 1.5–12 keV. Each data point
represents the one-day average of the fitted source fluxes from a number (typically 5–
10) of individual ASM dwells, of ∼90 s each (see Levine et al. 1996 for more details).
The one-day average light curve is shown in Fig. 2.14. The big gap between Modified
Julian Date ∼ 50400 and ∼ 50500 corresponds to the passage of the Sun close to
the source during the first year of observations. This gap repeats the following years
(near MJD 50800, 51150, 51900 and 52250). That is the reason why we have a
dataset of only 1800 one-day average points spanning nearly ∼ 2400 days.
19 Quick-look results provided by the ASM/RXTE team.

66 Chapter 2. Multiwavelength approach to LS 5039
Most of time the source is at the threshold of ASM detectability. Nevertheless, we
have searched for possible periodicities in the range from 2 to 1000 d. The methods
employed were the Phase Dispersion Minimization (PDM) (Stellingwerf 1978) and
the CLEAN algorithm (Roberts et al. 1987). Our approach here is essentially the
same as in Paredes et al. (1997) when analyzing the periodic behavior in the X-ray
lightcurve of LS I +61 303.
ASM units [count/s]
6
4
2
0
6
4
2
0
ASM units [count/s] 50100 50150 50200 50250 50300 50350 50400 50450 50500 50550 50600
6
4
2
0
ASM units [count/s] 50600 50650 50700 50750 50800 50850 50900 50950 51000 51050 51100
6
4
2
0
ASM units [count/s] 51100 51150 51200 51250 51300 51350 51400 51450 51500 51550 51600
6
4
2
0
52100 52150 52200 52250 52300 52350 52400 52450 52500 52550 52600
Modified Julian Date [JD−2400000.5]
ASM units [count/s] 51600 51650 51700 51750 51800 51850 51900 51950 52000 52050 52100
Figure 2.14: ASM/RXTE one-day average lightcurve of RX J1826.2−1450 / LS 5039
in the 1.5–12 keV energy band.
After applying both the PDM and CLEAN methods to the ASM data, a period
of ∼ 52 d is found. This periodicity corresponds to the detection of some kind of
active events that appear rather evident at first glance in Fig. 2.14. Nevertheless,

2.6. The X-ray counterpart: RX J1826.2−1450 67
a careful inspection of the data reveals a suspicious detail. All those active events
take place when the data by dwell coverage is rather poor (less than 5 dwells per
day), thus reducing the statistical significance of the corresponding one-day average.
For some instrumental reason, the ASM coverage becomes poorer than normal every
∼ 52 d or so and, in the case of a weak X-ray source like RX J1826.2−1450, this can
affect somehow the period analysis. Therefore, the ∼ 52 d period is very likely to
be an instrumental artifact. Indeed, after removing all daily points resulting from
less than 5 dwells (∼ 26% of total), the timing analysis reveals no significant period
in the range from 2 to 1000 d when comparing the results from both periodograms.
We point out that we have not attempted to fold the ∼ 6.5 years of ASM/RXTE
data using the ephemeris by McSwain et al. (2001), since the orbital period is not
accurate enough (as we have seen in Sect. 2.5.4), and any small deviation from the
true period would completely modify the obtained folded lightcurve due to the span
of the data.
2.6.2 PCA/RXTE observations
We conducted 20 detailed X-ray observations with the Proportional Counter Array
(PCA) onboard RXTE 21 on 1998 February 8 and 16 (MJD 50852 and MJD 50860).
The total on-source integration time was 20 ks. The PCA is sensitive to X-rays in
the energy range 2–60 keV and comprises five identical co-aligned gas-filled propor-
tional counter units (PCUs), providing a total collecting area of ∼ 6500 cm 2 , an
energy resolution of

68 Chapter 2. Multiwavelength approach to LS 5039
Table 2.8: Log of the PCA/RXTE observation of RX J1826.2−1450.
Date Start time Stop time MJD Flux (a) Luminosity (a,b)
[TT] (c) [TT] (c) [erg s −1 cm −2 ] [erg s −1 ]
08/02/98 01:05:33 04:18:14 50852.1 5.6 × 10 −11 5.6 × 10 34
08/02/98 20:07:34 00:57:14 50852.9 4.6 × 10 −11 4.6 × 10 34
16/02/98 18:42:25 20:35:14 50860.8 4.6 × 10 −11 4.6 × 10 34
(a) In the energy range 3–30 keV. (b) For an assumed distance of 2.9 kpc.
(c) Terrestrial Time = International Atomic Time + 32.184 s.
30
25
20
15
0
25
5000 10000 15000
20
15
10
25
20
15
10
70000 75000 80000
755000 760000 765000
Figure 2.15: 3–30 keV lightcurve of RX J1826.2−1450 / LS 5039 covering the entire
PCA/RXTE observation. Time 0 is JD 2450852.547 and the bin size is 180 s.

2.6. The X-ray counterpart: RX J1826.2−1450 69
Figure 2.16: Characteristic power spectrum of RX J1826.2−1450 / LS 5039. The dashed
line represents the 95% confidence detection limit.
Timing
The PCA observations were used to study the time variability on various timescales.
Continuous stretches of clean data were selected from the light curve of the entire
observation. To reduce the variance of the noise powers, these intervals were divided
up into segments of 8192 bins each, with a bin size of 10 ms. Then the power density
spectra for each segment were calculated and the results averaged together. Fig. 2.16
shows the characteristic power spectrum in the frequency range 0.01–50 Hz. The
dashed line represents the 95% confidence detection limit (van der Klis 1989). As it
can be seen no power exceeds this value; the distribution of powers is flat at a level
of 2, consistent with Poissonian counting statistics. Following van der Klis (1989),
we can set a 95% upper limit of 60% on the rms of a pulsed source signal in the
range 0.01–50 Hz. This relatively high limit is a consequence of the faintness of the
source. On longer timescales, longer intervals (∼ 3200 s) were considered but no
evidence for pulsations was found either.
Likewise, we folded the lightcurve onto a set of trial periods with the FTOOLS
software package (a technique very similar to the PDM) and looked for a peak in
the χ 2 versus period diagram. None of the peaks found were statistically significant
enough. Thus, we conclude that no coherent periodicities were detected in the range
∼ 0.02 to ∼ 2000 s.

70 Chapter 2. Multiwavelength approach to LS 5039
The mean X-ray intensity in the energy range 3–30 keV shows a slight decreasing
trend with 19.7 ± 0.2 count s −1 at the beginning of the observation (upper panel of
Fig. 2.15) compared to 16.6 ± 0.1 count s −1 and 16.2 ± 0.2 count s −1 for the middle
and bottom panels of Fig. 2.15, respectively. The fractional rms of the 3–30 keV
lightcurve corresponding to the entire observation is 9%.
The fact that no X-ray pulsations have been found in RX J1826.2−1450 is con-
sistent with the proposed idea that radio emission and X-ray pulsations from X-ray
binaries seem to be statistically anti-correlated (Fender et al. 1997, Fender & Hendry
2000), i.e., no X-ray pulsar has ever shown significant radio emission.
We have not computed the orbital phase of each one of the PCA/RXTE obser-
vations according to the ephemeris by McSwain et al. (2001), since the ∼ 1000 d
difference between their T value and our observations is too high to obtain reli-
able results due to the lack of accuracy for the orbital period (as we have seen in
Sect. 2.5.4).
Spectrum
Since the lightcurve of the entire observation does not show sharp features, i.e.,
there is no significant spectral change throughout the observation, we obtained one
average PCA energy spectrum from the complete observation, that can be seen in
Fig. 2.17.
Acceptable fits of the X-ray continuum were obtained with an unabsorbed power-
law model, giving a reduced χ 2 =1.14 for 56 degrees of freedom (dof). A multicolor
disk model, as expected from an optically thick accretion disk (Mitsuda et al. 1984)
plus a power-law gave a reduced χ2 ν =1.11 for 53 dof. Bremsstrahlung and two blackbody
component models did not fit the data. Although the addition of a blackbody
component to the power-law formally produces an acceptable fit, the value of the
blackbody normalization was very low, with the error bar close to zero. In fact, an
F-test shows that the inclusion of a blackbody component is not significant. The
most salient feature that appears in the spectrum of RX J1826.2−1450 is a strong
iron line at ∼ 6.6 keV (Fig. 2.17). A Gaussian fit to this feature gives a line centered
at 6.62±0.04 keV, with an equivalent width (EW ) of 0.75±0.06 keV and a F W HM
of 0.9 ± 0.2 keV. The best-fit results are given in Table 2.9.

2.6. The X-ray counterpart: RX J1826.2−1450 71
Figure 2.17: PCA spectrum of RX J1826.2−1450 / LS 5039. The continuous line
represents the best-fit power-law model plus a Gaussian component for the iron line.
When this component is omitted the line shows up clearly in the residuals.
Table 2.9: Spectral fit results for the power-law model plus a Gaussian component for
the iron line in RX J1826.2−1450 / LS 5039. Uncertainties are given at 90% confidence
for one parameter of interest. The spectrum was fitted in the energy range 3–30 keV.
Parameters Fitted values
NH [×10 21 cm −2 ] 2 +1
−2
Γ 1.95 ± 0.02
Eline(Fe) [keV] 6.62 ± 0.04
EWline(Fe) [keV] 0.75 ± 0.06
F W HMline(Fe) [keV] 0.9 ± 0.2
χ 2 r (dof) 1.14 (56)

72 Chapter 2. Multiwavelength approach to LS 5039
In HMXBs the presence of an emission line around 6.4 keV is commonly seen,
with a variety of strengths. It is ascribed to fluorescent reprocessing by neutral or
mildly ionized iron in relatively cold circumstellar material (Fe Kα line). A higher
energy implies the presence of highly ionized iron (He-like ∼ FeXXV) or a different
origin than Kα emission. One possibility would be radiative recombinations of H-
like iron followed by cascade processes in a relatively hot corona. Unfortunately, the
PCA energy resolution prevents from distinguishing between these possible models.
On the other hand, the high EWline(Fe) value indicates that a large amount of
circumstellar matter is present in the system. Since LS 5039 lies at low galactic
latitude, we cannot rule out that part of the emission feature comes from the Galactic
ridge (Yamauchi & Koyama 1993). Nevertheless, this contribution cannot account
for the total flux of the line. In addition, Wang et al. (2002) have recently shown
that the ∼ 6.7 keV emission near the Galactic Center region arises from discrete
sources rather than from the diffuse continuum, giving support to the idea that the
∼ 6.7 keV line comes from LS 5039.
A hydrogen column density of ∼ 2 +1
−2 × 10 21 cm −2 is found from the fit. This
value is, however, not very well constrained. In fact, it is consistent with zero. The
difficulty in constraining the hydrogen column density from our X-ray data can be
attributed to the fact that the interstellar gas mainly absorbs X-ray photons with
energies lower than 2–3 keV, i.e., outside of the energy range considered here. We
note that from the λ 4430 and λ 6284 interstellar bands Motch et al. (1997) found
E(B − V ) = 0.8 ± 0.2. Using the relation NH = 5.3 × 10 21 cm −2 E(B − V ) (Predehl
& Schmitt 1995), we obtain that NH ∼ (4 ± 1) × 10 21 cm −2 , which is consistent,
within the errors, with the X-ray observation value. However, if we consider the
obtained photometric values for E(B − V ) quoted in Table 2.6 ( 1.2 ± 0.1), we
obtain NH (6.4 ± 0.5) × 10 21 cm −2 , which seems clearly inconsistent with the
results after the PCA/RXTE observations. In any case, the lack of sensitivity of
the detector below 3 keV may explain why we have detected such low values of NH.
Possible nature of the compact object based on the X-ray data
The non detection of X-ray pulses does not necessarily mean that RX J1826.2−1450
/ LS 5039 contains a black hole, since it could contain a low magnetic field neutron
star. Both types of objects have been found in REXBs.

2.6. The X-ray counterpart: RX J1826.2−1450 73
On the other hand, the multicolor disk model, although formally fitting the data,
does not help either. First, given the low luminosity (LX < 10 35 erg s −1 in the energy
range 2–10 keV) the system would be in the so-called low/hard state. We would
not expect then to detect a strong soft component. Second, the fit provides an
unrealistic value of the disk internal radius of Rin cos 1/2 (θ) ∼ 0.3 km.
Unfortunately, the source is too faint at energies above 30 keV to be detected
with the HEXTE/RXTE instrument. Thus, we cannot confirm from the present
data whether the hard tail that usually characterizes the energy spectrum of black
holes at high energies is indeed present.
2.6.3 BeppoSAX observations
With the aim to confirm and improve the obtained results after the PCA/RXTE
observations, we observed 22 RX J1826.2−1450 / LS 5039 with the BeppoSAX X-
ray satellite during ∼ 80 ksec on 2000 October 8 (MJD 51825.29–51826.21). Un-
fortunately, while the flux detected during the PCA/RXTE observations was of
∼ 5 × 10 −11 erg s −1 cm −2 , it was only of ∼ 5 × 10 −12 erg s −1 cm −2 during the
BeppoSAX observations, i.e., one order of magnitude lower. Hence, we could not
confirm the detection of the ∼ 6.7 keV iron line due to lack of signal-to-noise ra-
tio. However, we could improve the value of the hydrogen column density, NH =
1.0 ± 0.3 × 10 22 cm −2 , and obtain a new value of the photon index, Γ = 1.8 ± 0.2 (in
the energy range 3–10 keV).
Using the relation by Predehl & Schmitt (1995) the NH value implies E(B−V ) =
1.9±0.6, nearly compatible with the E(B−V ) 1.2±0.1 values derived from optical
photometry (see Table 2.6) and much higher than the E(B − V ) 0.4 +0.2
−0.4 value
derived from the PCA/RXTE observations, which was poorly constrained. On the
other hand, the value of Γ is compatible with the one found in the PCA/RXTE
observations, and typical of black holes in the low/hard state, although we note
that McSwain & Gies (2002) have recently suggested a neutron star nature for the
compact object in LS 5039.
The BeppoSAX observations took place in JD 2451825.79–2451826.71, hence
very close to the T =JD 2451822.12 value used in the McSwain et al. (2001)
22 Reig, P., Ribó, M., Paredes, J. M., & Martí, J. 2002, in preparation.

74 Chapter 2. Multiwavelength approach to LS 5039
ephemeris. Therefore, the lack of accuracy in the orbital period is not important in
this case in order to compute the orbital phase, which ranges from 0.89 to 0.11. As
discussed in Sect. 2.5.4, the time of an X-ray eclipse if one occurs is near phase 0.06
according to these ephemeris, and its maximum duration, for a star of 40 M⊙ and
an inclination angle of 90 ◦ , is expected to be of ∼ 0.2 d, or ∼ 0.05 phase interval
(McSwain, private communication). Hence, we observed the source just around the
time of a possible X-ray eclipse, but there is no signature of such an event in the
BeppoSAX data.
The non-detection of an X-ray eclipse is consistent with the proposed inclination
of the system of ∼ 30 ◦ (McSwain & Gies 2002). We also note that McSwain &
Gies (2002), based on the broad residual emission in the Hα profile, obtained a
higher mass loss rate of the optical star in 1998 than in 2000, which may explain
the difference in the detected flux between the PCA/RXTE observations (carried
out in year 1998) and the BeppoSAX observations (carried out in year 2000).
Therefore, we conclude that the source was intrinsically faint during these obser-
vations, and that there seem to be no X-ray eclipses in RX J1826.2−1450 / LS 5039,
in good agreement with the proposed inclination of this binary system.
2.7 The γ-ray counterpart: 3EG J1824−1514
The highly energetic processes that take place near compact objects in X-ray binaries
are observable from radio to hard X-rays (Mirabel & Rodríguez 1999) and possibly
beyond. Therefore, soon after the discovery of the microquasar nature of LS 5039, we
inspected the most recent high-energy catalogs in order to find possible counterparts
to LS 5039. While there was no counterpart in the COMPTEL (0.75–30 MeV)
source catalog (Schönfelder et al. 2000), there was a very interesting unidentified
high-energy γ-ray source in the Third EGRET Catalog (E > 100 MeV) (Hartman
et al. 1999), namely 3EG J1824−1514.
In Fig. 2.18 we show the location map of 3EG J1824−1514, together with the
X-ray sources from both the ROSAT All Sky Bright Source Catalog (RBSC) (Voges
et al. 1999) and Faint Source Catalog (RFSC) (Voges et al. 2000), as well as the
radio sources brighter than 20 mJy (for clarity) from the NRAO VLA Sky Survey
(NVSS) (Condon et al. 1998). As can be seen, there is only a RBSC source, namely

2.7. The γ-ray counterpart: 3EG J1824−1514 75
Galactic Latitude
0.0
-1.0
-2.0
ROSAT Bright
ROSAT Faint
NVSS (>20 mJy)
17.0 16.0
Galactic Longitude
15.0
Figure 2.18: Location map of 3EG J1824−1514. The contours represent, from inside
to outside, the 50%, 68%, 95%, and 99% statistical probability that a γ-ray source lies
within the given contour. The open big circles are ROSAT bright sources, the small
open circles are the ROSAT faint sources and the filled circles are radio sources from the
NVSS survey. The only source with X-ray and radio emission (filled circle inside an open
big circle, l = 16.88 ◦ and b = −1.29 ◦ ), and well inside the 95% contour, is LS 5039.

76 Chapter 2. Multiwavelength approach to LS 5039
RX J1826.2−1450 / LS 5039, within the 3EG J1824−1514 contours, and well inside
the 95% confidence contour. Moreover, this is the only X-ray source, bright or faint,
that also displays radio emission (addition of NVSS sources below 20 mJy does not
change this discussion). Such a good position agreement between an EGRET source
and a radio jet X-ray binary strongly suggests that both objects may be the same.
Thus, this microquasar system is likely associated with an EGRET source.
As can be seen in Fig. 2.19, the γ-ray emission observed from 3EG J1824−1514
reveals a rather persistent flux of > 100 MeV photons for the last decade. This
persistent behavior is also seen at radio wavelengths in LS 5039, thus giving support
to the proposed association.
s −1
]
photon cm −2
γ−ray flux [10 −8
80
60
40
20
EGRET (>100 MeV)
VLA (1.45 GHz)
GBI (2.25 GHz)
0
1988 1990 1992 1994 1996 1998 2000
0
2002
Year
Figure 2.19: Radio and γ-ray lightcurves of LS 5039 and 3EG J1824−1514. The radio
flux densities are the VLA ones at 20 cm wavelength (1.45 GHz) quoted in Table 2.1
and the GBI ones presented in Sect. 2.4.3. The error bars for the GBI data (±4 mJy)
are not shown for clarity, whereas those of the VLA are usually smaller than the symbol
size. The γ-ray fluxes correspond to the four viewing periods of EGRET.
80
60
40
20
Radio flux density [mJy]

2.7. The γ-ray counterpart: 3EG J1824−1514 77
Figure 2.20: EGRET spectrum of 3EG J1824−1514 (from Hartman et al. 1999).
The average γ-ray flux for all EGRET viewing periods in Fig. 2.19 is Φγ =
(35.2±6.5)×10 −8 photon cm −2 s −1 , with photon spectral index p = 2.18±0.18, where
Φγ ∝ E −p
γ
(see Fig. 2.20 where α = p). The corresponding integrated luminosity
at a distance of 2.9 kpc (Ribó et al. 2002) amounts to Lγ(> 100 MeV) ∼ 3.6 ×
10 35 erg s −1 , compared to an X-ray luminosity of LX(3–30 keV) ∼ 5 × 10 34 erg s −1
(Ribó et al. (1999) with a distance of 2.9 kpc, see also Table 2.8) .
We note that 3EG J1824−1514 is persistent but variable. In fact, its index of
γ-ray variability as defined in Torres et al. (2001) is I = 3. This value probably
rules out a pulsar or a SNR nature for the EGRET source. Moreover, the EGRET
photon index p = 2.18 ± 0.18 of LS 5039 is steeper than the p < 2 values usually
found for pulsars (Merck et al. 1996). This gives further support to the association.

78 Chapter 2. Multiwavelength approach to LS 5039
The importance of the association between LS 5039 and 3EG J1824−1514 is not
only because of the discovery of high-energy γ-ray emission in a microquasar, but
mainly because of the large number of still unidentified EGRET sources. The Third
EGRET catalog contains a total of 271 sources, of which 103 are identified objects:
66 high-confidence AGN, 27 low-confidence AGN, 7 pulsars, 1 detection from the
Large Magellanic Cloud, 1 solar flare and the radiogalaxy Centaurus A. Among the
remaining 168 unidentified sources (62%), 72 of them have |b| < 10 ◦ . Therefore, this
association opens the possibility that some of the high-energy unidentified γ-ray
sources at low galactic latitudes could be silent microquasars to be discovered in the
near future.
+180
Active Galactic Nuclei
Unidentified EGRET Sources
3EG J1824-1514 / LS 5039
Third EGRET Catalog
E > 100 MeV
+90
-90
Pulsars
LMC
Solar FLare
Figure 2.21: The Third EGRET source Catalog shown in galactic coordinates. The
size of the symbol represents the highest intensity seen for this source by EGRET. The
white cross inside a black circle (l = 16.88 ◦ and b = −1.29 ◦ ) indicates the γ-ray source
3EG J1824−1514, to be associated with LS 5039. Adapted from Hartman et al. (1999).
-180

2.8. A proposed scenario to explain the multiwavelength behavior 79
2.8 A proposed scenario to explain the multiwave-
length behavior
2.8.1 Spectral energy distribution
In Fig. 2.22 we show the observed spectral energy distribution of LS 5039 ranging
from radio wavelengths to γ-rays. We have quoted the domains within the electro-
magnetic spectrum and the different names of LS 5039 according to each wavelength
range.
log (ν F ν [erg s −1 cm −2 ])
−7
−9
−11
−13
−15
−17
NVSS J182614−145054
LS 5039
RADIO NIR/OPTIC
RX J1826.2−1450
3EG J1824−1514
X−RAY GAMMA−RAY
10 15 20 25
log (ν [Hz])
Figure 2.22: Observed spectral energy distribution of LS 5039. The radio data corre-
spond to the fit in Eq. 2.1, the near infrared (NIR) and optical data are those quoted in
Table 2.5, the X-ray data comes from the fit to the PCA/RXTE data shown in Fig. 2.17,
and the γ-ray data comes from the fit to the spectrum shown in Fig. 2.20.

80 Chapter 2. Multiwavelength approach to LS 5039
As we have already discussed, the radio emission originates as synchrotron ra-
diation in a jet-like flow in the presence of a magnetic field. The near infrared and
optical light arises from the high mass companion star. The X-rays probably arise
in the accretion disk around the compact object. Finally, the γ-rays are probably
produced by inverse Compton scattering of the ultraviolet photons of the companion
star by the relativistic electrons present in the jet flow, as we discuss below.
2.8.2 A model based on the γ-ray/radio emission
The movement of matter in the system is proposed to be as follows. The early type
O6.5V((f)) star LS 5039 has a strong wind that accounts for a mass loss rate of
∼ 10 −6 M⊙ yr −1 , according to McSwain & Gies (2002). While leaving the star,
part of this matter will be captured by the intense gravitational field of the nearby
compact object (probably a neutron star). The matter will then fall into the compact
object and form a disk around it, which by viscous heating will emit in X-rays. Part
of the X-ray emission can also be produced in a corona above the disk. A fraction
of the matter forming the disk will be ejected perpendicular to it and in opposite
senses, probably as a result of the Blandford & Payne (1982) mechanism, as seems to
be the case in the quasar M87 (Junor et al. 1999). While flowing away into opposite
jets, the relativistic electrons (matter) will be unavoidably exposed to a huge output
of ultraviolet (UV) photons from the hot optical star. As a consequence, the UV
photons will experience inverse Compton scattering, and γ-rays will be naturally
produced, as detected by EGRET. Later on, the relativistic electrons will produce
synchrotron radio emission. This scenario is schematically reproduced in Fig. 2.23.
It is interesting to use the presence of 5 GHz synchrotron emitting particles far
away from the compact object, ∼ 1000 AU as seen in Sect. 2.4.5, in order to constrain
the jet physical parameters. The starting point of the following calculations will be
the scenario for the proposed EGRET emission of LS 5039 due to IC scattering. We
recall that the energy shift in the IC process may be as high as Eγ ∼ 4γ 2 e EUV, where
the energies of the γ-ray and the stellar UV photon are related through the squared
Lorentz factor of the relativistic electron.
We will adopt here a very simple model of an expanding jet. Cylindrical coor-
dinates, z and r measured parallel and perpendicular to the jet axis, are the best
choice in our case. The jet is assumed to form at a distance z0 from the compact

Figure 2.23: Scenario where the multiwavelength emission originates in LS 5039.
Synchrotron
Radiation
Inverse Compton
Scattering
X-ray
L 3-30 keV ~ 5×10 34 erg/s
v jet ≥ 0.15c
e -
e -
e -
e -
e -
e -
γ-ray, E > 100 MeV, L γ ~ 4×10 35 erg/s
γ e ~ 10 3
Radio, L 0.1-100 GHz ~ 1×10 31 erg/s
UV, E ~ 10 eV
O6.5V((f))
L opt ~ 1×10 39 erg/s
2.8. A proposed scenario to explain the multiwavelength behavior 81

82 Chapter 2. Multiwavelength approach to LS 5039
object and flows with a velocity v = β c. The lateral expansion of the jet is parame-
terized as r = r0(z/z0) ɛ , where r0 is the initial jet radius. A freely expanding conical
jet would correspond to ɛ = 1, while for slowed lateral expansion we would have
ɛ < 1 (see e.g. Hjellming & Johnston, 1988). The expansion velocity perpendicular
to the jet axis is thus v⊥ = dr/dt = ɛvr/z. Concerning the magnetic field, we will
only consider the component perpendicular to the jet axis, because it has the slowest
decay if conservation of the magnetic flux is assumed. The magnetic field along the
jet will be thus parameterized as B = B0(r/r0) −1 = B0(z/z0) −ɛ .
A relativistic electron injected at the base of the jet will decrease its energy
mainly through adiabatic expansion, IC and synchrotron losses according to:
dE
dt
v⊥E
= −2
3 r − αICUradE 2 − αSB 2 E 2 . (2.8)
The factor 2/3 in the adiabatic expansion term comes from the lateral expansion of
the jet. The constants αIC = 3.97 × 10 −2 and αS = 2.37 × 10 −3 (c.g.s. units) are the
coefficients of the terms accounting for IC and synchrotron losses, respectively. The
radiation energy density Urad is assumed to be dominated by stellar UV photons from
the O6.5V star, whose luminosity above 10 eV amounts to L∗ 5 × 10 38 erg s −1
and Urad = L∗/[4πc(a 2 + z 2 )]. With the recently determined orbital parameters
(McSwain et al. 2001, McSwain & Gies 2002), the likely semimajor axis of the
orbit is a = 2.6 × 10 12 cm. Adopting these parameters, the radiation energy density
close to the compact object is so high that electrons with energies of 10 −3 –10 −2 erg
(γe ∼ 10 3 –10 4 ) will naturally produce γ-ray photons with energies of 100–1000 MeV,
as detected by EGRET. Moreover, integration of Eq. 2.8 taking only into account
the IC losses gives:
E(t) =
E0
1 + αICL∗E0
4πcav tan−1
, (2.9)
vt
a
which reveals that the energy of the electrons decays very fast to values of ∼ 5 ×
10 −4 erg, regardless of their initial energy (the expression above is asymptotic).
If we consider the IC contribution in Eq. 2.8 to be comparable to adiabatic losses
at the base of the jet (later on, the adiabatic expansion term will soon become
dominant due to its slower decay as z −1 ), at the base of the jet we can request the
following:
αIC
L∗
4πc(a 2 + z 2 0) E2 0
2ɛ vE0
3 z0
. (2.10)

2.8. A proposed scenario to explain the multiwavelength behavior 83
We can use the values quoted above, together with v = 0.2c and an energy of
E0 = 5 × 10 −4 erg for the electrons injected at the base of the jet, to solve for z0 in
this quadratic equation. The obtained results are 0.35, 0.85 and 1.30 AU for ɛ = 1,
1/2 and 1/3, respectively. The synchrotron term can be shown to be negligible a
posteriori and has not been considered here.
Hereafter, we will proceed taking into account only the adiabatic expansion term,
because it is dominant compared to the IC one, and rewrite Eq. 2.8 as:
dE
dt
−2ɛ
3
vE
z
. (2.11)
This can be easily integrated to give E E0(z/z0) −2ɛ/3 . As the energy decays, the
synchrotron emission of relativistic electrons will shift towards lower frequencies. In
particular, the synchrotron frequency expected at a distance z is given by:
νc = 6.27 × 10 18 BE 2 6.27 × 10 18 B0E 2 0 (z/z0) −7ɛ/3 . (2.12)
The electrons still producing 6 cm (νc = 5 × 10 9 Hz) synchrotron emission at about
z = 1000 AU (the size of the MERLIN jets) are probably those originally injected
with energies of E0 5 × 10 −4 erg. Then, the magnetic field B0 at the base of the
jet can be roughly estimated as:
B0 8.0 × 10 −10 E −2
0
z
z0
7ɛ/3
. (2.13)
This provides values of 3.6 × 10 5 , 12 and 0.6 G for ɛ = 1, 1/2 and 1/3, respectively.
If magnetic flux is conserved, the corresponding field at the end of the MERLIN
jet would be B1000 AU = 128, 0.4 and 0.06 G. For comparison, the only independent
estimates of the magnetic field in the LS 5039 jets come from simple equipartition
arguments. The results are merely indicative of the average magnetic field in the
solid angle of the sky covered by the jets. Using the total flux density and size of the
MERLIN jets given in Table 2.3 we derive a field value of > 8×10 −3 G, although this
is only a lower limit because the jet width is unresolved. The VLBA image presented
in Sect. 2.4.4, allowed us to estimate an approximate magnetic field of ∼ 0.2 G for
radio jets reaching up to a few mas. It is clear that a field of such intensity is more
consistent with the ɛ < 1 results than with the too high magnetic fields implied
by a simple conical jet (ɛ = 1). Therefore, a slowly expanding jet, and hence well
collimated, is favored to power the LS 5039 radio emission up to ∼ 1000 AU by
the electrons that previously contributed to the γ-ray emission, without in situ
acceleration being required. This is also in agreement with our upper limit of ≤ 6 ◦
for the jet half opening angle derived from the EVN map.

84 Chapter 2. Multiwavelength approach to LS 5039
2.8.3 Energetic considerations
The central engine in LS 5039 must be supplying at least ˙ Ee ∼ Lγ ∼ 3.6×10 35 erg s −1
in the form of relativistic electrons to power the detected γ-rays (see Sect. 2.7). The
electron energy distribution is expected to be a power law kE 2α−1 dE, which can be
estimated as kE −2 dE according to the observed spectral index α −0.5. Assuming
electron energies in the range mec 2 ≤ E ≤ γmaxmec 2 we have:
˙Ee = k
E 2α−1
EdE k
E −2
EdE = k
E −1 dE = k ln γmax . (2.14)
Therefore, if the proton mass is mp 1800 me = 1800E/c 2 for every relativistic
electron, the proton mass flow into the jets can be written as:
Mjet
˙ = mpk
E 2α−1 dE 1800E
c2 k
E −2 dE 1800
k . (2.15)
c2 From Eq. 2.14 we can obtain the value of the constant, k = ˙ Ee/ ln γmax, and rewrite
Eq. 2.15 as:
˙
Mjet 1800 ˙ Ee
c 2 ln γmax
. (2.16)
Using then ˙ Ee ∼ Lγ ∼ 3.6 × 10 35 erg s −1 and γmax ∼ 10 4 , we obtain:
˙
Mjet ∼ 7.8 × 10 16 g s −1 ∼ 1.2 × 10 −9 M⊙ yr −1 , (2.17)
which is weakly dependent on the maximum energy cutoff assumed (γmax). Con-
sidering a mass loss from the companion star of 10 −6 M⊙ yr −1 , as previously seen,
this tells us that approximately a fraction of 1/1000 of the mass that leaved the
companion is finally ejected forming relativistic jets perpendicular to the accretion
disk. The equivalent kinetic luminosity is given by Lk = (Γ − 1) ˙
Mjetc 2 , where Γ is
the Lorentz factor of the bulk motion of the jet. Hence, the kinetic luminosity highly
depends on the velocity of the flow, and for values of β in the range 0.15–0.4, and
hence Γ ∼ 1.01–1.1, we obtain Lk ∼ 10 36 –10 37 erg s −1 . This kinetic power is about
four to five orders of magnitude lower than that estimated for the strong ejections
of the superluminal microquasar GRS 1915+105 (Mirabel & Rodríguez 1994), but
interestingly much higher than the ∼ 5 × 10 34 erg s −1 X-ray luminosity of LS 5039.

2.8. A proposed scenario to explain the multiwavelength behavior 85
2.8.4 Future prospects
We carried out two runs of global VLBI observations of LS 5039 on 2000 June 2 and
7, spanning a total of 14 hours each. Preliminary reduction of these data indicates
the presence of the jets in a similar position angle as previously found with the
VLBA, EVN and MERLIN. These data will be analyzed soon, and we expect to
measure proper motions of individual components within the jets, that will help to
constrain the jet parameters and improve our proposed model.
On the other hand, the high eccentricity of LS 5039 provides both, a variable
accretion rate along the orbit implying a variable rate of electrons injected into the
jet, and a variable radiation energy density close to the compact object. Hence, we
predict variability in the radio and γ-ray luminosities correlated with the orbital
period, which may hopefully be detected by INTEGRAL and/or GLAST missions
at high energies, and with already scheduled VLA observations at low energies. On
the other hand, if the star exhibits intrinsic variations in the UV photon flux, a
variability in γ-rays would be seen correlated to it.
Since LS 5039 seems to have persistent radio jets, it would be interesting to study
the possibility of detecting Doppler-shifted lines. As the jet flow has an estimated
velocity of ∼ 0.2c, the amount of Doppler-shift is expected to be comparable to that
found in SS 433 (Kotani et al. 1996; Marshall et al. 2002). Hence, spectroscopic
X-ray observations with XMM or Chandra could be very useful to constrain the
physics of the relativistic flow and to study the matter content within the jet.
Finally, as pointed out by Levinson & Waxman (2001) and by Distefano et al.
(2002), if the jets are hadronic we would expect the formation of TeV neutrinos,
that could be detected in the future with detectors of km 2 -scale effective area.
Overall, LS 5039 is a very good target to be studied with new instruments in
order to gain knowledge on accretion/ejection processes and jet physics.

86 Chapter 2. Multiwavelength approach to LS 5039
2.9 Conclusions
1. Most of the known microquasars have been discovered only after undergoing
a noticeable outburst, that has triggered the detection by a battery of satel-
lites and ground-based observatories. In contrast, we have discovered a new
microquasar, namely LS 5039, after careful examination of modern archive
databases and follow-up interferometric radio observations with different res-
olution (with the VLA, VLBA, EVN and MERLIN).
2. We have found that LS 5039 displays persistent radio emission in the form of
relativistic jets, with a high brightness temperature of ∼ 10 8 K and an optically
thin spectral index of α −0.5. These results are indicative of non-thermal
synchrotron radiation mechanism.
3. We have obtained the following jet parameters: β 0.15, θ 80 ◦ , a size up to
∼ 1000 AU with a position angle of 125–150 ◦ and a half opening angle smaller
than 6 ◦ . The asymmetry in flux density and distance to the core between
the approaching and receding jet seems to be persistent along time. Small
differences in the position angle of the jets, observed at different epochs and
with different resolutions, suggest bending and/or precession of the jets.
4. The radio emission is moderately variable but no signatures of variability cor-
related with the orbital period, expected because of the LS 5039 eccentric
orbit, have been found. This is probably because of the lack of sensitivity of
the long-term radio observations available up to now (GBI).
5. We have shown that LS 5039 displays slight variability at optical wavelengths,
and moderate variability at near infrared wavelengths. Based on the new
spectral type classification, O6.5V((f)), and recent calibrations of early type
stars, we have estimated a distance of 2.9 ± 0.3 kpc to LS 5039.
6. We have presented new radial velocity measurements (obtained with the INT)
that confirm the 4.1 d orbital period of this binary system.
7. We have analyzed six and a half years of X-ray monitoring (ASM/RXTE) of
LS 5039, and found no evidences of periodic variability. In particular, we have
not found the orbital period of the system, expected because of variable accre-
tion rate along an eccentric orbit, probably because the source is marginally
detected in the analyzed dataset.

2.9. Conclusions 87
8. We have carried out pointed X-ray observations (PCA/RXTE and BeppoSAX)
that reveal a spectrum significantly hard up to 30 keV, with a strong iron line
around 6.6 keV. No pulsations or quasi periodic oscillations have been found.
Based on our X-ray data, we have estimated a hydrogen column density of
NH = 1.0 ± 0.3 × 10 22 cm −2 . We have found X-ray variability of an order of
magnitude on timescales of years, that can be related to the variable mass loss
rate of the companion star. There seem to be no X-ray eclipses in LS 5039, in
good agreement with the low proposed inclination (∼ 30 ◦ ).
9. We have proposed an association between LS 5039 and one of the ∼ 170
unidentified EGRET sources, which seems plausible due to the persistent and
moderately variable emission of the source at both radio and γ-ray wave-
lengths. If confirmed, this would be the first association between a microquasar
and an EGRET source, and we suggest that other unidentified high-energy
γ-ray sources could be silent microquasars waiting to be discovered.
10. We have suggested a possible scenario to explain the multiwavelength behavior
of LS 5039, and proposed a model based on the γ-ray/radio emission. This
model suggests a slowly expanding and well collimated radio jet.
11. Finally, we have made some energetic considerations and suggested future
observations that could help to better constrain the proposed model.
12. Overall, we think that a careful examination of modern archive databases may
reveal a previously unnoticed population of microquasars like LS 5039. If this
is correct, the microquasar phenomenon may not be as rare as it seems.

88 Chapter 2. Multiwavelength approach to LS 5039

Bibliography
Blandford, R. D., & Königl, A. 1979, ApJ, 232, 34
Blandford, R. D., & Payne, D. G. 1982, MNRAS, 199, 883
Casares, J., Ribó, M., Paredes, J. M., & Martí, J. 2002, in preparation
Clark, J. S., Reig, P., Goodwin, S. P., et al. 2001, A&A, 376, 476
Condon, J. J., Cotton, W. D., Greisen, E. W., et al. 1998, AJ, 115, 1693
Deutschman, W. A., Davis, R. J., & Schild, R. E. 1976, ApJS, 30, 97
Distefano, C., Guetta, D., Waxman, E., & Levinson, A. 2002, ApJ, 575, 378
Drilling, J. S. 1975, AJ, 80, 128
Drilling, J. S. 1991, ApJS, 76, 1033
Fender, R. P., & Hendry, M. A. 2000, MNRAS, 317, 1
Fender, R. P., Roche, P., Pooley, G. G., et al. 1997, in Proc. of the 2nd INTEGRAL
Workshop ‘The Transparent Universe’, ESA SP-382, 303
Gray, D. F. 1976, The Observations and Analysis of Stellar Photospheres, Wiley-
Interscience, New York
Hartman, R. C., Bertsch, D. L., Bloom, S. D., et al. 1999, ApJS, 123, 79
Helfand, D. J., Zoonematkermani, S., Becker, R. H., et al. 1980, ApJ, 80, 211
Hjellming, R. M., & Han, X. 1995, Radio properties of X-ray binaries, in X-ray Bi-
naries, ed. W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel (Cambridge
Univ. Press, Cambridge), 308
89

90 BIBLIOGRAPHY
Hjellming, R. M., & Johnston, K. J. 1988, ApJ, 328, 600
Horne, K. 1986, PASP, 98, 609
Johnson, H. L. 1966, ARA&A, 4, 193
Junor, W., Biretta, J. A., & Livio, M. 1999, Nature, 401, 891
Kilkenny, D. 1993, South African Astronomical Observatory Circular, 15, 53
Kilkenny, D., & Whittet, D. C. B. 1993, South African Astronomical Observatory
Circular, 15, 40
Kilkenny, D., van Wyk, F., Marang, F., Spencer Jones, J. H. & Whittet, D. C. B.
1993, South African Astronomical Observatory Circular, 15, 65
Kotani, T., Kawai, N., Matsuoka, M., & Brinkmann, W. 1996, PASJ, 48, 619
Kraus, A. 1997, PhD Thesis, Friedrich-Wilhelms-Universität Bonn, Germany
Lahulla, J. F., & Hilton, J. 1992, A&AS, 94, 265
Landolt, A.U. 1992, AJ, 104, 340
Ledden, J. E., Broderick, J. J., Condon, J. J., et al. 1980, AJ, 85, 780
Lejeune, T., & Schaerer, D. 2001, A&A, 366, 538
Levine, A. M., Bradt, H., Cui, W., et al. 1996, ApJ, 469, L33
Levinson, A., & Waxman, E. 2001, Phys. Rev. Lett., 87, 171101
Liu, Q. Z., van Paradijs, J., & van den Heuvel, E. P. J. 2000, A&AS, 147, 25
Marsh, T. R., Robinson, E. L., & Wood, J. H. 1994, MNRAS, 266, 137
Marshall, H. L., Canizares, C. R., & Schulz, N. S. 2002, ApJ, 564, 941
Martí, J., Paredes, J. M., & Ribó, M. 1998, A&A, 338, L71
Massi, M., Paredes, J. M., Estalella, R., & Felli, M. 1993, A&A, 269, 249
McSwain, M. V., & Gies, D. R. 2002, ApJ, 568, L27
McSwain, M. V., Gies, D. R., Riddle, R. L., Wang, Z., & Wingert, D. W. 2001,
ApJ, 558, L43

BIBLIOGRAPHY 91
Merck, M., Bertsch, D. L., Dingus, B. L., et al. 1996, A&AS, 120, 465
Mirabel, I. F., & Rodríguez, L. F. 1994, Nature, 371, 46
Mirabel, I. F., & Rodríguez, L. F. 1999, ARA&A, 37, 409
Mitsuda, K., Inoue, H., Koyama, K., et al. 1984, PASJ, 36, 741
Monet, D. 1996, BAAS, 188, 5404
Monet, D. G., Bird, A., Canzian, B., et al. 1999, USNO-A2.0 CD-ROM (U.S. Naval
Observatory, Washington DC)
Motch, C., Haberl, F., Dennerl, K., Pakull, M., & Janot-Pacheco, E. 1997, A&A,
323, 853 (M97)
Pacholczyk, A. G. 1970, Radio Astrophysics, Freeman, San Francisco, CA
Paredes, J. M., Martí, J., Peracaula, M., & Ribó, M. 1997, A&A, 320, L25
Paredes, J. M., Martí, J., Ribó, M., & Massi, M. 2000, Science, 288, 2340
Paredes, J. M., Ribó, M., Ros, E., Martí, J., & Massi, M. 2002, A&A, 393, L99
Peng, B., Kraus, A., Krichbaum, T. P., & Witzel, A. 2000, A&A, 145, 1
Penny, L. R. 1996, ApJ, 463, 737
Pooley, G. G., Fender, R. P., & Brocksopp, C. 1999, MNRAS, 302, L1
Predehl, P., & Schmitt, J. H. M. M. 1995, A&A, 293, 889
Rees, M. J. 1966, Nature, 211, 468
Reig, P., Ribó, M., Paredes, J. M., & Martí, J. 2002, in preparation
Ribó, M., Reig, P., Martí, J., & Paredes, J. M. 1999, A&A, 347, 518
Ribó, M., Paredes, J. M., Romero, G. E., et al. 2002, A&A, 384, 954
Roberts, D. H., Lehár, J., & Dreher, J. W. 1987, AJ, 93, 968
Schaerer, D., de Koter, A., Schmutz, W., & Maeder, A. 1996, A&A, 312, 475

92 BIBLIOGRAPHY
Schmidt-Kaler, Th. 1982, Landolt-Börnstein: Numerical Data and Functional Re-
lationships in Science and Technology, New Series ”Group 6 Astronomy and As-
trophysics”, Volume 2
Schönfelder, V., Bennett, K., Blom, J. J., et al. 2000, A&AS, 143, 145
Spencer Jones, J. H., Whittet, D. C. B., & He, L. 1993, South African Astronomical
Observatory Circular, 15, 73
Stellingwerf, R. F. 1978, ApJ, 224, 953
Stephenson, C. B., & Sanduleak, N. 1971, Publ. Warner & Swasey Obs., 1, 1
Taylor, A. R., Young, G., Peracaula, M., Kenny, H. T., & Gregory, P. C. 1996,
A&A, 305, 817
Torres, D. F., Romero, G. E., Combi, J. A., et al. 2001, A&A, 370, 468
Vacca, W. D., Garmany, C. D., & Shull, J. M. 1996, ApJ, 460, 914
van der Klis, M. 1989, in Timing Neutron stars, ed. H. Ögelman & E. P. J. van den
Heuvel (Dordrecht: Kluwer), 27
Voges, W., Boller, Th., Dennerl, K., et al. 1996, in Proc. of the Conference Rönt-
genstrahlung from the Universe, MPE Report, 263, 637
Voges, W., Aschenbach, B., Boller, Th., et al. 1999, A&A, 349, 389
Voges, W., Aschenbach, B., Boller, Th., et al. 2000, IAU Circ., 7432, 3
Wang, Q. D., Gotthelf, E. V., Lang, C. C. 2002, Nature, 415, 148
Yamauchi, S., & Koyama, K. 1993, ApJ, 404, 620

Chapter 3
LS 5039 as a runaway microquasar
The formation of the compact object in an X-ray binary necessarily requires a super-
nova explosion that does not disrupt the system. This explosive event is expected
to considerably change the kinematic properties of the binary star. The change
may be rather extreme (i.e., kick velocities approaching ∼ 10 3 km s −1 ) for highly
asymmetric supernovae, as it has been proposed for the microquasar Circinus X-1
(Tauris et al. 1999). This mechanism may also be responsible for ejecting X-ray
binaries into the halo of the Galaxy. The fast moving X-ray nova XTE J1118+480 is
a possible example (Mirabel et al. 2001), although it could have been ejected from
a globular cluster in the past. On the other hand, the microquasars GRO J1655−40
and Cygnus X-1, also display runaway velocities (Shahbaz et al. 1999; Kaper et al.
1999). With independence of their origin in the galactic plane or in the halo, the
existence of runaway microquasars is a new and important issue that deserves an
in-depth study.
In this chapter 1 we focus on the kinematic properties of LS 5039 and its sur-
roundings. In Sect. 3.1 we estimate the proper motions of the system, using optical
and radio data. In Sect. 3.2 we discuss the space velocity of LS 5039, whereas in
Sect. 3.3 we analyze the trajectory of the binary system in the past. In Sect. 3.4 we
study the possible association of this microquasar with a supernova remnant, and
in Sect. 3.5 we analyze their H i surroundings. Finally, we make a global discussion
in Sect. 3.6 and summarize our main conclusions in Sect. 3.7.
1 Published in Ribó, M., Paredes, J. M., Romero, E. G., Benaglia, P., Martí, J., Fors, O., &
García-Sánchez, J. 2002, A&A, 384, 954.
93

94 Chapter 3. LS 5039 as a runaway microquasar
3.1 Positions and proper motions
3.1.1 Optical positions
LS 5039 (V 11.2, B 12.2) is not included in most accurate astrometric catalogs,
like Hipparcos and Tycho-1 (which are complete up to V ∼ 9.0 and V ∼ 10.0,
respectively). Although it appears in Tycho-2, its position has a large uncertainty
when compared to the average error within its magnitude range (Høg et al. 2000).
As a result, the catalogued proper motions lack of the required precision. Moreover,
other astrometric catalogs which were not included in Tycho-2 estimate of the proper
motions, have been released since then. In view of all these facts, we decided to carry
out a thorough study of all available catalogs. First, we did a search using the query
forms within the VizieR database (Ochsenbein et al. 2000) at the Centre de Données
astronomiques de Strasbourg (CDS), and afterwards we looked for new information
on several public web sites. The results of this search are listed in Table 3.1, with
all coordinates in ICRS, and commented below.
The oldest astrometric information for LS 5039 is dated around 1905 and comes
from the Astrographic Catalog 2000 (AC 2000, Urban et al. 1998). However, a
re-reduction of this catalog was performed recently, using an improved reference
catalog that allowed a better handling of the systematic errors (S. E. Urban, private
communication). This new catalog will soon be released as AC 2000.2. Since changes
were most significant in the faint stars of the southern AC 2000 zones, the position
of LS 5039 suffered a noticeable shift. Actually, two positions corresponding to
different epochs are given in AC 2000.2.
The following available astrometry, dated around 1950, comes from the scanning
and reduction of the Palomar Observatory Sky Survey plates, carried out by Monet
et al. (1999), which is part of the USNO-A2.0 catalog.
We inspected the Second Cape Photographic Catalog (CPC2, Zacharias et al.
1999), a Southern Hemisphere astrometric catalog, containing observations between
1962 and 1973. Unfortunately, LS 5039 is not listed in it, because the limiting
magnitude of this catalog was V 10.5.
The Twin Astrographic Catalog version 2 (TAC 2.0) provides positional informa-

3.1. Positions and proper motions 95
Table 3.1: Compilation of optical and radio positions, with associated errors, of LS 5039.
Wavelength Epoch α (ICRS) σα cos δ δ (ICRS) σδ Catalog Ref.
domain [h, m, s] [mas] [ ◦ , ′ , ′′ ] [mas] name
Optical 1905.45 18 26 15.0194 255 −14 50 53.300 247 AC 2000.2 1
1907.43 18 26 15.0177 255 −14 50 53.075 247 AC 2000.2 1
1951.577 18 26 15.034 250 −14 50 53.59 250 USNO-A2.0 2
1979.484 18 26 15.0557 64 −14 50 54.075 47 TAC 2.0 3
1986.653 18 26 15.054 300 −14 50 54.29 300 GSC 1.2 4
1991.75 18 26 15.0427 149 −14 50 54.229 120 Tycho-2 5
2000.289 18 26 15.0563 13 −14 50 54.277 13 UCAC1 6
Radio 1998.24 18 26 15.056 10 −14 50 54.24 10 VLA obs. 7
2000.42 18 26 15.0566 4 −14 50 54.261 6 VLBA obs. 8
1. Urban, private communication. 2. Monet et al. 1999. 3. Zacharias & Zacharias 1999. 4. Morrison
et al. 2001. 5. Høg et al. 2000. 6. Zacharias et al. 2000. 7. Martí et al. 1998. 8. Ribó et al. 2002.
tion around 1980. Apart from the internal error of σα cos δ = 62 mas and σδ = 45 mas
associated to the LS 5039 position, an external error of 15 mas has been taken into
account following an estimate carried out by Zacharias & Zacharias (1999).
The last photographic catalog used is the Guide Star Catalog version 1.2 (GSC 1.2,
Morrison et al. 2001). This is a re-reduction of the GSC (Lasker at al. 1990), after
removing plate-based systematic distortions that are a function of magnitude and
radial distance from the plate center. In addition, by using the Starlink library
slalib, we have transformed the catalogued position from FK5 to ICRS.
As mentioned above, LS 5039 appears as well in the Tycho-2 catalog (Høg et al.
2000). The average internal error in position for a V 11.2 star is around 50 mas,
while the errors in α cos δ and δ for LS 5039 are 93 and 120 mas, respectively. This
fact tells us that the position estimate is not as good as one would expect. On the
other hand, the scatter-based errors from the 4 positions present in the Tycho-2
data are 149 and 60 mas, respectively. Hence, we have considered the higher error
estimate in each coordinate: σα cos δ = 149 mas and σδ = 120 mas.
A second epoch of the GSC was performed during the nineties. However, the
positional errors in the current version of this catalog, GSC 2.2.01, are intended
as indicators for operational use only, and cannot be used for scientific studies.

96 Chapter 3. LS 5039 as a runaway microquasar
Hence, we have considered that this catalog does not provide accurate astrometric
information to be used for estimating proper motions.
Finally, we have used the first release of the US Naval Observatory CCD Astro-
graph Catalog (UCAC1, Zacharias et al. 2000). As can be seen in Table 3.1, this is
by far the best optical astrometric catalog available.
3.1.2 Radio positions
Independent astrometric information can be obtained from radio observations of
LS 5039. The first available astrometric information comes from the NVSS by
Condon et al. (1998). However, the position error in this catalog, around 1 ′′ ,
is too large to be of any use for our purposes. Martí et al. (1998) carried out
VLA 2 -A configuration observations which provided accurate (10 mas) astrometry for
LS 5039. Finally, phase-referencing VLBA 2 observations conducted by Ribó et al.
(2002) provide the last radio position, with the best accuracy among all available
data. Details of the astrometry for the last two cases are listed in Table 3.1. The
relatively large location error of few mas in the VLBA data is due to confusion
produced by the high level of Galactic electron scattering in this region (for both
the reference source and LS 5039).
We must state that no correction for parallax has been applied to the obtained
positions, since at an estimated distance of 2.9 ± 0.3 kpc, this would translate into
less than half a mas, which is always much smaller than the available uncertainties.
3.1.3 Proper motions
It is clear from the astrometric data that both, the optical and radio sources, are
almost in the same position of the sky. However, in order to show that this is not
a chance coincidence and that both emissions originate in the same object, we have
computed independent proper motions for the optical and radio data.
2 The VLA and the VLBA are operated by the National Radio Astronomy Observatory (NRAO).
The NRAO is a facility of the National Science Foundation operated under cooperative agreement
by Associated Universities, Inc.

3.1. Positions and proper motions 97
Table 3.2: Proper motions estimates for LS 5039.
Data set µα cos δ µδ
(mas yr −1 ) (mas yr −1 )
Optical 4.7 ± 1.3 −11.0 ± 0.8
Radio 4.0 ± 4.9 −9.6 ± 5.3
Optical+Radio 4.7 ± 1.1 −10.6 ± 1.0
Although the position uncertainties from the AC 2000.2 and USNO-A2.0 cat-
alogs are relatively large, the epoch span achieved justifies their inclusion in the
estimate of the proper motions. Hence, from the optical data, and taking into
account the astrometric uncertainties, we obtain the following results: µα cos δ =
4.7 ± 1.3 mas yr −1 , µδ = −11.0 ± 0.8 mas yr −1 , where the errors come directly
from the least squares fit. The two accurate radio positions give proper motions of:
µα cos δ = 4.0 ± 4.9 mas yr −1 , µδ = −9.6 ± 5.3 mas yr −1 . Although this last result
has a large error because only two points are available, we can say that both, the
optical and radio sources, have very similar proper motions. Therefore, based only
on astrometric data, we are able to confirm that both, the optical and the radio
emission, originate in the same object. Hence, we will use all the data listed in
Table 3.1 to compute accurate proper motions for LS 5039, which happen to be:
µα cos δ = 4.7 ± 1.1 mas yr −1 , µδ = −10.6 ± 1.0 mas yr −1 . All these results are
summarized in Table 3.2. We can now transform the proper motions into galactic
coordinates and obtain µl = −7.2 ± 1.0 mas yr −1 and µb = −9.1 ± 1.0 mas yr −1 .
It is clear from these results that there is a noticeable motion perpendicular to the
galactic plane and moving away from it.
According to our least squares fits, and defining t = yr − 2000.0, the predicted
ICRS values for α and δ near t = 0 are:
α = 18 h 26 m 15.0565 s +
δ = −14 ◦ 50 ′ 54.260 ′′ −
4.7 t/ cos δ
, (3.1)
mas
10.6 t
, (3.2)
mas
being 9 + 1.2 t 2 / cos 2 δ and √ 9 + t 2 the errors in mas in α and δ, respectively.
Comparing the fits with the positions in Table 3.1 and the proper motions in Ta-
ble 3.2 we can say that, approximately, the offsets are determined by the VLBA data

98 Chapter 3. LS 5039 as a runaway microquasar
∆α cos δ (mas)
200
0
−200
−400
−600
−800
1900 1920 1940 1960
Time (yr)
1980 2000 2020
∆δ (mas)
1400
1200
1000
800
600
400
200
0
−200
−400
1900 1920 1940 1960
Time (yr)
1980 2000 2020
Figure 3.1: Offsets in Right Ascension (left) and Declination (right), relatives to the
fitted ones for year 2000.0, versus time for all the positions listed in Table 3.1. Open
circles represent the optical positions, and filled circles the radio ones. The solid lines
represent the least squares fits to the whole data sets.
point because of the small error in position, while the proper motions are obtained
by the optical points due to the huge time span.
In Fig. 3.1 we have plotted the offsets in α cos δ and δ from the fitted values for
year 2000.0 versus time, for the data listed in Table 3.1, together with the respective
fits to the data. Notice that in both plots the UCAC1 and VLBA data are almost
superimposed.
3.2 Ejection from the galactic plane
If we assume a distance of 2.9 ± 0.3 kpc to LS 5039, as seen in Sect. 2.5.3, and a
systemic Vr = 4.6 ± 0.5 km s −1 (McSwain et al. 2001), the proper motions estimates
translate into (U = 40 ± 5, V = −82 ± 16, W = −118 ± 19) km s −1 in the Local
Standard of Rest (LSR) defined by (U⊙ = 9, V⊙ = 12, W⊙ = 7) km s −1 . This gives
a total systemic velocity of vsys = (U 2 + V 2 + W 2 ) 1/2 = 149 ± 18 km s −1 . Using a
value of 215 km s −1 for the galactic rotation at 5.8 kpc from the galactic center (Fich
et al. 1989) we can transform the galactic plane space velocities (U and V ) from the
LSR into the LS 5039 regional standard of rest (RSR). Applying this transformation
we find U = 51 ± 6 and V = −71 ± 16 km s −1 , and vsys = 147 ± 17 km s −1 . All
these results are summarized in Table 3.3.

3.3. The past trajectory of LS 5039 99
Table 3.3: Space velocity estimates, in km s −1 , for LS 5039 related to the LSR and to
its RSR.
Frame U V W vsys
LSR 40 ± 5 −82 ± 16 −118 ± 19 149 ± 18
RSR 51 ± 6 −71 ± 16 −118 ± 19 147 ± 17
Taking into account a cosmic dispersion for early-type stars of (σU, σV , σW )
(7, 8, 4) km s −1 (Torra et al. 2000) we can conclude that this source is escaping
from its own RSR and has a large velocity component perpendicular to the galactic
plane. Since LS 5039 is a HMXB containing an early-type star, it seems very un-
likely to be a high speed halo object crossing the galactic plane. Hence, the most
plausible explanation for the observed velocity is that the binary system obtained
an acceleration during the supernova event that created the compact object in this
microquasar.
3.3 The past trajectory of LS 5039
An interesting thing to do once a position and space velocity estimates are available,
is to compute the trajectory of LS 5039 in the past, and then to look for OB
associations and SN Remnants (SNRs) in its path, in order to establish possible
relationships. For this purpose, we computed the galactic orbital motion of this
system under the gravitational field of the Galaxy. The galactic mass model by
Dauphole & Colin (1995) was adopted for the integration, which was performed
using a Runge-Kutta fourth-order integrator using time steps of 1000 years.
It is helpful to search for OB associations in the past trajectory of runaway X-ray
binaries, as done by Ankay et al. (2001) with HD 153919/4U 1700−37, because if
evidences are found that the runaway system originated in an OB association, it is
possible to estimate the age of the binary system after the SN explosion. According
to the computed trajectory in the past for LS 5039, there are two OB associations
in or close to its path in the plane of the sky: Sct OB3 and Ser OB2 (Melnik &
Efremov 1997). Unfortunately, the distance to Sct OB3 is ∼ 1.5 kpc (Melnik &
Efremov 1997), and the distance to Ser OB2 is 1.9 ± 0.3 kpc (Forbes 2000). Hence,

100 Chapter 3. LS 5039 as a runaway microquasar
the two OB associations found in the path of LS 5039 are too close to us to be
related to it.
Since we have not been able to fix a limit on the integration time in the past
thanks to a possible relationship with an OB association, we will use the vertical
distribution of early-type stars in the Galaxy for this purpose. O-type stars are
typically located at distances within 45 ± 20 pc from the galactic midplane (Reed
2000), i.e., from Z −65 to Z +65 pc. According to the galactic latitude of
LS 5039 and using a distance of 2.9 ± 0.3 kpc, its actual height is Z = −65 ± 7 pc.
Since it is escaping from the galactic plane, we can compute the trajectory backward
in time up to when it had a height of Z = +65 pc, which was ∼ 1.1 Myr ago. Hence,
the SN explosion probably took place in the last ∼ 1.1 Myr.
Thus, for a possible association of LS 5039 with a SNR, we will focus our attention
on the galactic trajectory of the binary system for the last ∼ 1 Myr. In addition,
the likelihood of detecting a SNR decreases with time, so it is not expected that the
radio emission of the SNR can be observed much beyond the time interval considered
above (e.g., Shklovskii 1968).
A search in a catalog of galactic supernova remnants (Green 2000) reveals the
presence of several SNRs near the path of LS 5039 on the plane of the sky, though
the center of none of them is crossed by its trajectory. Only three of them, namely
SNR G016.7+00.1, SNR G016.8−01.1, and SNR G018.8+00.3, are within 1 ◦ of the
LS 5039 path. The distance to SNR G016.7+00.1 was estimated to be at ∼ 14 kpc
by Reynoso & Mangum (2000), making impossible an association with LS 5039.
Dubner et al. (1999) concluded that SNR G018.8+00.3 is located at 1.9 ± 0.5 kpc,
and has an age of ∼ 16 000 yr. Although the distance is not very different from
the one to LS 5039, the age of this SNR is by far too short to be associated with
the microquasar. Hence, the only remaining candidate is SNR G016.8−01.1, which
deserves a detailed study.
3.4 SNR G016.8−01.1
A wide field radio map of the surrounding of LS 5039 is shown in Fig. 3.2, with
the microquasar (contours) and the nearby SNR G016.8−01.1 (grey scale), together
with the position of the H ii-region RCW 164 (which is the stronger source in the

3.4. SNR G016.8−01.1 101
DECLINATION (J2000)
-14 30
35
40
45
50
55
-15 00
LS 5039
SNR G016.8-01.1
o
b = -0.8
RCW 164
18 26 30 15 00 25 45 30 15 00 24 45 30
RIGHT ASCENSION (J2000)
Figure 3.2: Wide field radio map of LS 5039, its nearby shell-like SNR G016.8−01.1
and the H ii-region RCW 164 (which is the stronger source in the field). The grey scale
emission is taken from the Parkes-MIT-NRAO tropical survey at the 6 cm wavelength
(Tasker et al. 1994). The overlaid contours correspond to the NVSS map of the region
at the 20 cm wavelength (Condon et al. 1998). The arrow marks the proper motion
sense (see text). The dashed line is the computed trajectory for the last 10 5 yr, with
the corresponding error boxes in position at that epoch and 5 × 10 4 yr ago. The dotted
line represents positions with a galactic latitude of −0.8 ◦ for reference purposes.

102 Chapter 3. LS 5039 as a runaway microquasar
field). The arrow in this figure marks the proper motion sense of LS 5039 as it would
be seen from the LSR (i.e., correction for the peculiar velocity of the Sun has been
applied), while the dashed line represents the trajectory in the past up to 10 5 yr,
with the corresponding error boxes in position at that epoch and 5 × 10 4 yr ago.
From this figure we can see that LS 5039 crosses the projection in the plane of the
sky of SNR G016.8−01.1 between ∼ 4 × 10 4 and ∼ 1.3 × 10 5 years ago.
If one of the two stars forming a close binary system experiences a SN explosion,
it may form a SNR and an X-ray binary. Since conservation of the linear momentum
after the SN explosion is required, if the SNR and the X-ray binary are related, we
expect them to be aligned with the proper motion of the latter, and any deviation
from this behavior should be explained by the projection in the plane of the sky of
the initial space velocity of the binary system prior to the SN event. This does not
seem to be the case here, because SNR G016.8−01.1 and LS 5039 are not well aligned
with the LS 5039 proper motion. Moreover, if we assume that the center of the shell-
like remnant is located at the minimum of radio emission in Fig. 3.2, approximately
located at α = 18 h 25 m 03 s , δ = −14 ◦ 37 ′ 30 ′′ (or l = 16.94 ◦ , b = −0.93 ◦ ), a
peculiar velocity in the plane of the sky for the original system of 70 km s −1
is obtained. This velocity is much higher than the typical ∼ 10 km s −1 found for
early-type stars. In order to reduce this velocity we can use in our favor the error in
the proper motion angle, and find that it turns out to be 50 km s −1 , which is still
too high. However, the center of the shell-like remnant could be in another position,
closer to the LS 5039 trajectory, and the derived peculiar velocity for the system
prior to the SN explosion could fit in the typical values of early-type stars. Hence,
from the kinematical point of view, we are not able to firmly discard a possible
association between both objects. Therefore, we have performed an in-depth study
of the remnant.
SNR G016.8−01.1 was discovered by Reich et al. (1986) using a combination
of survey data analysis and new multifrequency observations. The extended radio
source has a complex structure due to the superposition of the H ii-region RCW 164
(Rodgers et al. 1960), which is coincident with the peak of the continuum emission
(in black in Fig. 3.2). The diffuse radiation observed around the peak is strongly
polarized (Rodgers et al. 1960), indicating a synchrotron nature. Polarization de-
creases towards the position of the foreground H ii-region as a result of the thermal
contribution. Thermal and non-thermal components of the radiation cannot be
clearly separated, but the total flux from the SNR at 5 GHz seems to be ∼ 1 Jy

3.4. SNR G016.8−01.1 103
Normalized Intensity
1,4
1,2
1,0
0,8
0,6
0,4
0,2
0,0
-0,2
-0,4
-0,6
-100 -80 -60 -40 -20 0 20 40 60 80
Velocity (km/s)
Figure 3.3: H166α line observations of the H ii-region RCW 164. The solid line repre-
sents a Gaussian fit to the data, with its maximum being at v = 16.5 ± 0.8 km s −1 .
(J. A. Combi, private communication). The angular diameter of the remnant is
∼ 30 ′ , and its distance remains unknown.
3.4.1 A lower limit of the distance
In order to determine a lower limit of the distance to SNR G016.8−01.1, we car-
ried out H166α recombination line (1424.734 MHz) observations of the foreground
H ii-region on 2000 November 17. We used a 30-m radiotelescope at the Instituto
Argentino de Radioastronomía (IAR), Villa Elisa. The receiver is a helium-cooled
HEMT amplifier with a 1008-channel autocorrelator at the back end. The HPBW
at a wavelength of 21 cm is 30 ′ and the temperature of the system on the cold sky
during the observations was about 35 K. The H166α line was detected at a veloc-
ity of 16.5 ± 0.8 km s −1 (see Fig. 3.3) after 1.5 hours of integration time, with a
signal-to-noise ratio of ∼ 4. At a location of l ≈ 16.8 ◦ , b ≈ −1.1 ◦ , standard galactic
rotation models (Fich et al. 1989) indicate that the observed velocity corresponds
to a distance d ∼ 1.8 kpc. Consequently, SNR G016.8−01.1 should be farther than
∼ 2 kpc, as suggested by its relatively small angular size.

104 Chapter 3. LS 5039 as a runaway microquasar
3.4.2 Particle density estimates
If we adopt for SNR G016.8−01.1 the same distance to LS 5039, we can make
some estimates of the parameters that would characterize the SNR in case of being
associated with the X-ray binary. Using a size of 30 ′ and a distance of 2.9 kpc, the
inferred radius would be R = 12.7 pc and it would be, thus, still in the adiabatic
expansion phase. Using the standard Sedov (1959) solutions, we can express the
particle density of the ambient medium as
n = 4.44 × 10 −8 t 2 E51 R −5
1 cm −3 , (3.3)
where t is the age of the remnant in years, E51 is the original energy release of the
SN explosion in units of 10 51 erg and R1 is the remnant radius in units of 10 pc.
Since the binary system is bound after the SN explosion, its total mass has to be
larger than the mass of the SNR (van den Heuvel 1978), which, therefore, should
be moving at least with the same space velocity in the opposite direction. Hence,
the maximum age of the remnant if it were related to LS 5039, would be around
5 × 10 4 yr. If we use this age and assume that E51 = 0.4 (Spitzer 1998), we find
an ambient density of n ∼ 13.5 cm −3 . Considering the uncertainty in the distance,
we can say that 8 < n < 23 cm −3 . Although such number densities would not be
very unusual towards this direction of the inner part of the Galaxy, we have made
H i observations in order to estimate the typical column densities in the region of
interest.
The H i observations were performed during 2000 November 29 and December
1, with the already mentioned IAR telescope. The H i line was observed in a hybrid
total-power mode with a sampling on a 0.25 ◦ -lattice around the position of SNR
G016.8−01.1, covering a total of 4 ◦ × 4 ◦ . The velocity resolution obtained was
1.05 km s −1 , with a total coverage of ±450 km s −1 . The rms level in brightness
temperature was ∼ 0.1 K. Those channel maps in the velocity range from 10 to
50 km s −1 (corresponding to distances between 1.5 and 4.5 kpc) were inspected
looking for clouds or a local minimum that could be associated to the SNR.
The maps show a strong gradient of brightness temperature towards the galactic
plane, but no clear evidence for a structure that could be associated with the SNR.
In Fig. 3.4, we show the integrated column density map for the velocity interval
between 26–34 km s −1 , where the contour labels are in units of 10 19 cm −2 . With
the column densities observed in the region where SNR G016.8−01.1 is located

3.4. SNR G016.8−01.1 105
Galactic latitude
1.0
0.0
-1.0
-2.0
-3.0
-19.0 | -18.0 | -17.0 | -16.0 | -15.0 |
Galactic longitude
Figure 3.4: Neutral hydrogen column density distribution towards SNR G016.8−01.1, in-
tegrated over the velocity interval 26–34 km s −1 . Contour levels are in units of 10 19 cm −2 .
The circle marks the position of SNR G016.8−01.1, while the star represents LS 5039.
(∼ 4 × 10 20 cm −2 ), the ambient density should be ∼ 5 cm −3 . This is nearly a factor
of 3 lower than the estimated density from the expected size of the SNR if it is at
2.9 kpc. Taking into account the uncertainty in the distance, the ambient density
determined from the remnant’s size (8 < n < 23 cm −3 ) would be slightly larger than
that estimated from the H i observations. It could be that there is more material
under the form of other molecular species, or that the original SN energy release
could have been different from what we have assumed, or, finally, that the distance
to the SNR is not 2.9 kpc and it is not related to LS 5039. The case, consequently,
remains unsolved.

106 Chapter 3. LS 5039 as a runaway microquasar
3.5 The H i surroundings of LS 5039
Notwithstanding the absence of structures that can be clearly associated with the
SNR in the H i distribution, the channel maps in the interval 10–46 km s −1 (see
Fig. 3.5) reveal the existence of a large, semi-open cavity which is very similar to
bubbles blown out by early-type stars in regions with steep density gradients (e.g.,
Benaglia & Cappa 1999). This phenomenon has also been observed around another
HMXB (HD 153919) by Benaglia & Cappa (1999). The ambient material is thought
to be swept up by the strong wind of the massive early-type star in the binary system
creating a local minimum around the star. Density gradients towards the galactic
plane in the original matter distribution can result in a preferred escape direction for
the wind, yielding open structures. Even in some cases the star appears displaced
from the actual minimum in the H i distribution.
The bubble detected around LS 5039 is centered at a systemic velocity of vsys
33 km s −1 , with an expansion velocity vexp 12 km s −1 . The galactic rotation model
by Fich et al. (1989) indicates that it is located at a distance of 3.2 kpc, quite
consistent with the distance to the microquasar. The total mass removed in order
to create the cavity is ∼ 1.5 × 10 4 M⊙, implying a kinetic energy ∼ 2 × 10 49 erg s −1 ,
which can be considered as a rough estimate of the energy deposited by the wind
of the star into the ISM. Such a value is in accordance with previous estimates for
similar Of star wind blown bubbles (e.g., Benaglia & Cappa 1999). With a radius of
∼ 60 pc, the dynamical age of the bubble is ∼ 3×10 6 yr, which should be considered
as an upper limit for its actual age since it ignores possible contributions from the
SN explosion and the wind of the progenitor of the compact star.
We have also searched for other contributors to the cavity, among stars earlier
than B2 that were projected over the cavity, and whose distances were in agree-
ment with that of LS 5039. It seems unlikely that the Wolf Rayet star WR 115
(WN6+OB?) could contribute in some way, because its distance is 2 kpc, and is
located over the shell of the cavity (l = 16.98 ◦ , b = −1.03 ◦ ). The other luminous
stars in the field, LS 5005, LS 5017, LS 5022, LS 5047, LS 5048 and HD 169673 can
sum up at most 20% of the energy required to create such a minimum in the H i
distribution. The O((f)) star in LS 5039, then, seems to be the main agent forming
the bubble.
However, the actual space velocity indicates that the system has been close to the

3.5. The H i surroundings of LS 5039 107
Galactic latitude
1.0
0.0
-1.0
-2.0
-3.0
1.0
0.0
-1.0
-2.0
-3.0
1.0
0.0
-1.0
-2.0
-3.0
-19.0 -18.0 -17.0 -16.0
v = 12 km/s v = 16 km/s v = 20 km/s
v = 24 km/s v = 28 km/s v = 32 km/s
v = 36 km/s v = 40 km/s v = 44 km/s
-19.0 -18.0 -17.0 -16.0
Galactic longitude
-19.0 -18.0 -17.0 -16.0 -15.0
| | | | | | | | | | | | |
Figure 3.5: Neutral hydrogen channel maps for the velocity interval 10–46 km s −1 ,
towards LS 5039. Each map covers 4 km s −1 , the central velocity is given in each label.
A local minimum and an open low density cavity can be clearly seen close to the HMXB.
The cavity is open towards the direction opposite to the strongest mass concentrations
near the galactic plane, as expected from a wind-blown structure.

108 Chapter 3. LS 5039 as a runaway microquasar
cavity only during the last ∼ 2 × 10 5 yr, which is much shorter than the upper limit
found for the dynamical age of the bubble (∼ 3 × 10 6 yr). Nevertheless, the cavity
could be originated by LS 5039 if the pre-SN system was close to the actual position,
in such a way that the progenitor of the compact object could have also contributed
to the formation of the bubble. In such a case, the X-ray binary would be only about
∼ 2 × 10 5 yr old, and we would expect to find the radio SNR originated in the SN
event with a current surface brightness of Σ408 ∼ 3×10 −22 W m −2 Hz −1 sr −1 (Caswell
& Lerche 1979). Filtering techniques of the diffuse galactic emission (e.g., Combi
et al. 1998) allow to detect very low brightness SNRs with Σ408 ∼ 10 −22 W m −2
Hz −1 sr −1 , but nothing has been found in this region except for SNR G016.8−01.1,
whose surface brightness at 408 MHz would be Σ408 ∼ 6 × 10 −22 W m −2 Hz −1 sr −1
for a typical spectral index α = −0.5. This level of flux is expected at this latitude
for a younger SNR, say with t ∼ 10 5 yr, but in such a case the particle density of
the medium around SNR G016.8−01.1 should be higher than the estimates given in
Sect. 5.2, with values of n ∼ 50 cm −3 , in order to confine the SNR to its inferred size
at d ∼ 2.9 kpc. These densities are 1 order of magnitude higher than those derived
from the H i observations. However, a clear rejection of SNR G016.8−01.1 cannot be
established on this basis alone, due to the flux contamination from the H ii region,
which could be responsible for errors as large as 100 % in the flux density estimate
of the source. Consequently, a picture where the H i bubble has been created by
both the stellar winds from the binary plus the SN explosion remains as an open
possibility.
3.6 Discussion
There are two ways to accelerate a binary system by a supernova explosion: ejec-
tion of material from the binary in a symmetric SN or an additional velocity kick,
produced by asymmetries in the SN itself. In Nelemans et al. (1999), the authors
conclude that there is no need of additional velocity kicks in systems like Cygnus X-1
or GRO J1655−40, both containing black hole candidates as the compact objects,
to explain their observed large runaway velocities. On the other hand, from the
study of Be/X-ray binaries containing neutron stars, van den Heuvel et al. (2000)
conclude that, in order to explain the observed large eccentricities and low space
velocities of these systems, a kick in an asymmetric SN explosion scenario is needed.
Moreover, the microquasar Circinus X-1, containing a neutron star, seems to have

3.6. Discussion 109
experienced a highly asymmetric SN explosion (Tauris et al. 1999). Unfortunately,
the nature of the compact object in LS 5039 still remains unknown. However, we
can try to study it in the symmetric SN scenario.
Let MX and MO be the masses of the compact object and the optical companion,
respectively. Taking into account the stellar evolution tracks by Schaerer et al.
(1996), we can assume that MO 40 M⊙. Considering that it must rotate slower
than the breakup speed, and using the mass function f(m) = 0.00103±0.00020 M⊙
by McSwain et al. (2001), an upper limit of MX < 9 M⊙ is obtained. The minimum
mass for the compact object would be that of a neutron star of the Chandrasekar
mass, MX > 1.4 M⊙. Hence, a total mass of ∼ 41–49 M⊙ remains in the system.
Since the actual eccentricity of the system is e = 0.41 ± 0.05, and has a short
orbital period of P = 4.117 ± 0.011 days, both parameters are expected to de-
crease with time due to tidal forces which act to re-circularize the orbit. Hence,
the eccentricity just after the SN explosion was epost−SN > 0.41. Using the equation
∆ M = epost−SN (MX + MO) and MX = 1.4 M⊙ we find that at least 17 M⊙ were
lost in the SN explosion in the symmetric case. If we adopt MX = 9 M⊙ the min-
imum mass loss is 20 M⊙. Moreover, in order to have the system bound after the
explosion, i.e., epost−SN < 1, the mass loss during the SN event must be less than
the remaining mass in the system, i.e., < 49 M⊙. Hence, the values for the reduced
mass of the system, defined as µ = 1/(1 + epost−SN), range from 0.5 to 0.7. Using
Eq. (5) of Nelemans et al. (1999) with the actual values of P and e we find that
the re-circularized period will be Pre−circ 3 days. Since the period of the initial,
pre-SN, binary system is given by Pi = µ 2 Pre−circ, the orbital period before the SN
explosion was in the range 0.8 to 1.5 days. Using Kepler’s third law and assuming
the latter value to allow the maximum separation possible between the progenitor
stars, initial total masses between 60–100 M⊙ lead to semimajor axis in the range
22–26 R⊙. This means that the two stars did not evolve separately, but probably
through a common envelope phase. Hence, a detailed evolution of the system should
be carried out in order to try to compute the initial mass of the progenitor star of
the compact object, the maximum age of the system before the SN event, and the
mass loss during the SN explosion itself. This kind of study is beyond the scope of
this work.
Nevertheless, we can try to explain the measured vsys in the context of the
symmetric SN explosion scenario using the following equation from Nelemans et al.

110 Chapter 3. LS 5039 as a runaway microquasar
(1999):
vsys
km s −1
1
−
∆ M MO Pre−circ
= 213
M⊙ M⊙ day
3 MX + MO
M⊙
− 5
3
. (3.4)
Assuming that MO = 40 M⊙ and Pre−circ = 3 days we can find two minimum values
of vsys depending on the adopted mass for the compact object. For MX = 1.4 M⊙,
and hence ∆ M > 17 M⊙, we find vsys 200 km s −1 . Similarly, for MX = 9 M⊙,
and hence ∆ M > 20 M⊙, we find vsys 180 km s −1 . Our measured recoil velocity
is vsys 150 ± 20 km s −1 , which would be perfectly reached in the SN explosion
scenario, as we have just seen. In fact, the measured value is a little bit lower than
the computed ones, which could indicate that part of the eccentricity of the system
was reached thanks to a kick in an asymmetric SN explosion. However, we must
be cautious with the use of the value for vsys obtained after Eq. 3.4, because some
effects have been ignored, as pointed out by Nelemans et al. (1999), and also the
errors in e and Pre−circ could reduce the computed values. Hence, we can say that
both, the high eccentricity measured by McSwain et al. (2001), and the high space
velocity reported here, are perfectly compatible, within errors, with a symmetric
SN explosion scenario. In other words, the actual eccentricity of the system is
naturally explained with the mass loss necessary to explain the measured space
(recoil) velocity of LS 5039. We note that McSwain & Gies (2002) have recently
arrived to very similar conclusions using our proper motions.
Another common mechanism referred to produce runaway stars is based on dy-
namical ejection from young open clusters. Numerical simulations (Kiseleva et al.
1998) show that it can explain some moderately high-velocity stars observed in the
Galactic disk, but only about 1% of the stars would be ejected with velocities higher
than 30 km s −1 . Hence, it seems difficult to explain the high observed velocity of
LS 5039 as a result of such a mechanism.
It is interesting to note that, contrary to other HMXB with low orbital peri-
ods, LS 5039 is still in the circularization process. Using the actual value of P , a
typical value of R = 10.3 R⊙ for the radius of the O6.5V star, and different for-
malisms (Claret et al. 1995; Claret & Cunha 1997) we obtain typical circularization
timescales of the order of (0.5 – 5) × 10 5 yr. These values are compatible with the
system being still in the circularization process, since its formation just in the galac-
tic midplane would take 5 × 10 5 yr to bring it to the actual position according to
the integration of its trajectory.

3.6. Discussion 111
At this point we might ask what is the point of studying runaway microquasars.
In the case of LS 5039, if the SN explosion had taken place recently, we could expect
the system to survive for the following ∼ 5 Myr, because the O6.5 star is still in
the main sequence. Hence, according to the computed trajectory, it would reach
a height of Z −600 pc, which translates into b −12 ◦ . On the other hand,
LS 5039 could be associated with the high energy γ-ray source 3EG J1824−1514,
as suggested by Paredes et al. (2000). Hence, if this association is correct, we could
be able to detect γ-ray microquasars up to values of |b| > 10 ◦ , although practically
all confirmed microquasars lie within |b| < 5 ◦ . In particular, runaway microquasars
could be connected with some of the unidentified faint, variable, and soft γ-ray
EGRET sources above/below the galactic plane, as suggested by Romero (2001)
and Mirabel et al. (2001).
Hence, it would be interesting to study the proper motions and radial velocities
of as many microquasars as possible. Unfortunately, this kind of information is not
easy to obtain for a wide sample of systems. Proper motions can be obtained either
from optical data or either from radio data. However, most optical counterparts
of these sources are faint objects not present in old astrometric catalogs. Hence, it
is mandatory to acquire new positions. In this context, it is better to do it in the
radio domain because the uncertainties are always smaller than those obtained in
the optical. On the other hand, it is necessary to carry out optical spectroscopy to
obtain the radial velocity curve of these sources and use the correct value for the
radial velocity of the system.
An alternative approach could be the search for signatures of bow shocks in the
microquasar vicinity. However, recent studies by Huthoff & Kaper (2002) of runaway
OB stars, reveal that the success of this method is highly dependant on the density
of the ISM around the runaway object. In particular, in only one object, namely
Vela X-1, a bow shock has been detected. Finally, one could search for signatures
of explosive events in the ISM. Sensitive spectral line observations in the radio are
probably the best tool for this purpose, as done in GRO J1655−40 by Combi et al.
(2001).
In any case, the study of runaway microquasars such as LS 5039 is likely to
contribute significantly to different areas of modern high-energy Astrophysics with
independence of the observing technique.

112 Chapter 3. LS 5039 as a runaway microquasar
3.7 Conclusions
After an in-depth study of the proper motions and surroundings of LS 5039 our
main conclusions are:
1. Positions at optical and radio wavelengths have been used to compute inde-
pendent optical and radio proper motions, which are perfectly compatible.
Therefore, based only on astrometric data we are able to confirm that both,
the optical and the radio emission, originate in the same object.
2. From the combined optical and radio positions we have computed an accurate
proper motion for LS 5039. This, together with the new estimate of 2.9 ±
0.3 kpc for the distance (Sect. 2.5.3) and the radial velocity of the system
(McSwain et al. 2001), allows us to compute a space velocity of (U = 51,
V = −71, W = −118) km s −1 in its Regional Standard of Rest (RSR). This
result implies that LS 5039 is a runaway microquasar with vsys 150 km s −1 ,
escaping from its own RSR with a large velocity component perpendicular to
the galactic plane. This is probably the result of the SN event that created
the compact object in this binary system.
3. We have computed the past trajectory of LS 5039. Two OB associations have
been found close to its path in the plane of the sky. However, they are too close
to us to be related to the microquasar. On the other hand, we have also found
three SNRs near the path of LS 5039. After discarding two of them based on
distance arguments, we have focused our attention on SNR G016.8−01.1. A
study of this source could not clearly confirm nor reject the association due to
the large uncertainties in the estimated radio flux density of the SNR. This fact
perhaps justifies future, high sensitivity searches of low-brightness remnants
in this region.
4. We have found a semi-open H i cavity close to the LS 5039 position. Although
the O((f)) star in this microquasar seems to be the main agent forming the
bubble, a contribution from the progenitor of the compact object cannot be
ruled out.
5. We have been able to explain both, the high space velocity and the high
eccentricity observed, in a symmetric SN explosion scenario with a mass loss
of ∆ M ∼ 17 M⊙.

3.7. Conclusions 113
6. Finally, the high space velocity and possible lifetime of this microquasar indi-
cate that it could reach a galactic latitude of b = −12 ◦ . Therefore, if the pro-
posed association between LS 5039 and the EGRET source 3EG J1824−1514 is
correct, we could be able to detect γ-ray microquasars up to values of |b| 10 ◦ .
In particular, runaway microquasars could be connected with some of the
unidentified faint, variable, and soft γ-ray EGRET sources above/below the
galactic plane.

114 Chapter 3. LS 5039 as a runaway microquasar

Bibliography
Ankay, A., Kaper, L., de Bruijne, et al. 2001, A&A, 370, 170
Benaglia, P., & Cappa, C. E. 1999, A&A, 346, 979
Caswell, J. L., & Lerche, I. 1979, MNRAS, 187, 201
Claret, A., & Cunha, N. C. S. 1997, A&A, 318, 187
Claret, A., Giménez, A., & Cunha, N. C. S. 1995, A&A, 299, 724
Combi, J. A., Romero, G. E., & Benaglia, P. 1998, A&A, 333, L91
Combi, J. A., Romero, G. E., Benaglia, P., & Mirabel, I. F. 2001, A&A, 370, L5
Condon, J. J., Cotton, W. D., Greisen, E. W., et al. 1998, AJ, 115, 1693
Dauphole, B., & Colin, J. 1995, A&A, 300, 117
Dubner, G., Giacani, E., Reynoso, E., et al. 1999, AJ, 118, 930
Fich, M., Blitz, L., & Stark, A. A. 1989, ApJ, 342, 272
Forbes, D. 2000, AJ, 120, 2594
Green, D. A. 2000, A Catalog of Galactic Supernova Remnants, Mullard Radio
Astronomy Observatory, Cambridge, England, UK (available on the World Wide
Web at http://www.mrao.cam.ac.uk/surveys/snrs/)
Høg, E., Fabricius, C., Makarov, V. V., et al. 2000, A&A, 357, 367
Huthoff, F., & Kaper, L. 2002, A&A, 383, 999
115

116 BIBLIOGRAPHY
Kaper, L., Camerón, A., & Barziv, O. 1999, in Proc. of the IAU Symp. 193: Wolf-
Rayet phenomena in massive stars and starburst galaxies, ed. K. A. van der Hucht,
G. Koenigsberger, and P. R. J. Eenens, ASP Conference Series, 193, p. 316
Kiseleva, L. G., Colin, J., Dauphole, B., & Eggleton, P. 1998, MNRAS, 301, 759
Lasker, B. M., Sturch, C. R., McLean, B. J., et al. 1990, AJ, 99, 2019
Martí, J., Paredes, J. M., & Ribó, M. 1998, A&A, 338, L71
McSwain, M. V., & Gies, D. R. 2002, ApJ, 568, L27
McSwain, M. V., Gies, D. R., Riddle, R. L., Wang, Z., & Wingert, D. W. 2001,
ApJ, 558, L43
Melnik, A. M., & Efremov, Y. N. 1997, VizieR Online Data Catalog, 902, 110013
Mirabel, I. F., Dhawan, V., Mignani, R. P., Rodrigues, I., & Guglielmetti, F. 2001,
Nature, 413, 139
Monet, D. G., Bird, A., Canzian, B., et al. 1999, USNO-A2.0 CD-ROM (U.S. Naval
Observatory, Washington DC)
Morrison, J. E., Röser, S., McLean, B., Bucciarelli, B., & Lasker, B. 2001, AJ, 121,
1752
Nelemans, G., Tauris, T. M., & van den Heuvel, E. P. J. 1999, A&A, 352, L87
Ochsenbein, F., Bauer, P., & Marcout, J. 2000, A&AS, 143, 23
Paredes, J. M., Martí, J., Ribó, M., & Massi, M. 2000, Science, 288, 2340
Reed, B. C. 2000, AJ, 120, 314
Reich, W., Fuerst, E., Reich, P., Sofue, Y., & Handa, T. 1986, A&A, 155, 185
Reynoso, E. M., & Mangum, J. G. 2000, ApJ, 545, 874
Ribó, M., Paredes, J. M., Martí, J., & Massi, M. 2002, in preparation
Rodgers, A. W., Campbell, C. T., & Whiteoak, J. B. 1960, MNRAS, 121, 103
Romero, G. E. 2001, in Proc. of The Nature of Unidentified Galactic High-energy
Gamma-ray Sources, ed. A. Carramiñana, O. Reimer, and D. J. Thompson, ASSL
Series of Kluwer Academic Publishers 51, p. 65

BIBLIOGRAPHY 117
Schaerer, D., de Koter, A., Schmutz, W., & Maeder, A. 1996, A&A, 312, 475
Sedov, L. I. 1959, Similarity and Dimensional Methods in Mechanics, John Wiley
and Sons, New York
Shahbaz, T., van der Hooft, F., Casares, J., Charles, P. A., & van Paradijs, J. 1999,
MNRAS, 306, 89
Shklovskii, I. S. 1968, Supernovae, John Wiley and Sons, New York
Spitzer, L. 1998, Physical Processes in the Interstellar Medium, John Wiley and
Sons, New York
Tasker, N. J., Condon, J. J., Wright, A. E., & Griffith, M. R. 1994, AJ, 107, 2115
Tauris, T. M., Fender, R. P., van den Heuvel, E. P. J., Johnston, H. M., & Wu, K.
1999, MNRAS, 310, 1165
Torra, J., Fernández, D., & Figueras, F. 2000, A&A, 359, 82
Urban, S. E., Corbin, T. E., Wycoff, G. L., et al. 1998, AJ, 115, 1212
van den Heuvel, E. P. J. 1978, in Physics and Astrophysics of Neutron Stars and
Black Holes, ed. R. Giacconi and R. Ruffini (Amsterdam: North Holland), 828
van den Heuvel, E. P. J., Portegies Zwart, S. F., Bhattacharya, D., & Kaper, L.
2000, A&A, 364, 563
Zacharias, N., & Zacharias, M. I. 1999, AJ, 118, 2503
Zacharias, N., Zacharias, M. I., & de Vegt, C. 1999, AJ, 117, 2895
Zacharias, N., Urban, S. E., Zacharias, M. I., et al. 2000, AJ, 120, 2131

118 BIBLIOGRAPHY

Part II
A search for new microquasars
119

Chapter 4
The cross-identification method
4.1 Introduction
As discussed in Sect. 1.3, the reduced population of known microquasars gave us a
strong motivation to start a long-term project focused on the search for new micro-
quasars in the Galaxy. In this context, X-ray, optical and radio catalogs provide a
fundamental tool for the search of new sources with known multiwavelength behav-
ior. Therefore, we have used the best available X-ray and radio catalogs to search
for new REXB candidates on the basis of positional coincidence, together with very
restrictive selection criteria. As a result we have obtained a sample of 13 radio
emitting X-ray sources, which is expected to contain some galactic X-ray binaries.
4.2 Cross-identification between RBSC and NVSS
catalogs
In order to find new microquasar candidates we have to search for new REXBs. The
first step, hence, has consisted of a positional cross-identification 1 of X-ray and radio
sources using the most complete available catalogs at both wavelengths, namely the
ROSAT all sky Bright Source Catalog (RBSC) and the NRAO VLA Sky Survey
(NVSS), which are briefly introduced now.
1 Published in Paredes, J. M., Ribó, M., & Martí, J. 2002, A&A, 394, 193.
121

122 Chapter 4. The cross-identification method
In the X-ray domain, the RBSC (Voges et al. 1999) contains in its current
version a total of 18 806 sources in the energy band 0.1–2.4 keV, and is derived
from the ROSAT All Sky Survey (Voges et al. 1996). Four energy bands in the
following ranges of keV are provided: A=0.1–0.4, B=0.5–2.0, C=0.5–0.9, D=0.9–
2.0. From these bands, two hardness ratios are computed: HR1 = (B −A)/(B +A),
HR2 = (D − C)/(D + C). The 1σ positional uncertainties are typically in the range
10–20 ′′ .
In the radio domain, the NVSS (Condon et al. 1998) catalog covers the sky
north of δ = −40 ◦ (82% of the celestial sphere) at a frequency of 1.4 GHz (20 cm
wavelength) using the VLA configurations D and DnC. It contains over 1.8 × 10 6
sources stronger than its 2.5 mJy completeness limit. The rms positional uncer-
tainties are less than 1 ′′ for sources stronger than 15 mJy and 7 ′′ for the faintest
detectable sources.
Since our aim is to maximize the probability of retaining only galactic REXB
systems by cross-identifying the RBSC and the NVSS catalogs, we have adopted the
following selection criteria:
1. Sources with absolute galactic latitude < 5 ◦ have been selected from the RBSC
catalog. Since the NVSS has a limit of δ > −40 ◦ , only the RBSC sources
above this declination have been allowed to continue in the selection process.
Therefore, approximately 75% of the whole |b| < 5 ◦ area is covered (l 347–
259 ◦ and α 17.2–8.6 h ).
2. Among the remaining RBSC sources, those containing screening flags about
nearby sources contaminating measurements or problems with position deter-
minations (Voges et al. 1999) have been rejected.
3. From statistical RBSC identification studies, Motch et al. (1998) concluded
that X-ray binaries are essentially recognizable from their hard X-ray spectra.
Therefore, in an attempt to avoid AGNs and cataclysmic variables, we have
only selected the sources with HR1 + σ(HR1) higher than 0.9. Although
this criterion would exclude microquasars in an outburst, we note that this
situation only lasts a short time of their life, and the probability of discarding
one of them is very low. A total of 241 RBSC sources remain in our sample
at this stage.

4.2. Cross-identification between RBSC and NVSS catalogs 123
4. The X-ray and radio positions must agree within errors. For this purpose, we
have selected NVSS sources within the 2σ (95% probability) error boxes of
the RBSC sources. We have also used the constraint of a maximum offset of
40 ′′ between the RBSC and the NVSS positions, since, for higher distances,
no reliable identification is expected (Voges et al. 1999).
5. No extended radio source has been allowed to continue in the selection process,
since any REXB is expected to appear compact at the NVSS resolution 2 .
The sources selected with the RBSC/NVSS cross-identification, a total of 35,
were then filtered with complementary optical information using the following cri-
teria:
1. We have inspected the SIMBAD database and the NASA/IPAC Extragalactic
Database (NED) for each source, and if it is listed as an extragalactic object,
we have obviously rejected it from the sample.
2. We have also inspected the Digitized Sky Survey, DSS1 and DSS2-red images 3 ,
and looked for possible optical counterparts in position agreement with the
NVSS sources. If an optical object is present and displays extended emission,
i.e., with a galaxy-like appearance, it has also been removed from the sample.
At the end of this process the resulting sample contained 16 sources. Among
them, we found the well known REXB LS I +61 303, which displays a one-sided
jet at milliarcsecond scales (Massi et al. 2001). We also recovered the well known
microquasars LS 5039, SS 433 and Cyg X-3. We show in Fig. 4.1 the distribution
of known microquasars in galactic coordinates, together with the area of the sky
covered by our search and the recovered sources clearly marked. It is interesting to
note that all these sources are HMXB, and none of the known microquasars with
low mass companions were found by this technique. This can be easily explained
by the fact that most of LMXB are transients, and the remaining persistent ones
are too faint to be present in the NVSS. Moreover, all but one HMXB persistent
microquasars in our explored range of galactic latitudes, namely LS 5039, SS 433
2 The sources resolved only in one axis and with an angular size smaller than the other axis were
allowed to continue on the process.
3 http://archive.eso.org/dss/dss/

124 Chapter 4. The cross-identification method
and Cyg X-3, have been recovered by our cross-identification method. The remain-
ing one, namely Cyg X-1, although marginally present in the corresponding NVSS
image, does not appear in the NVSS catalog. This was probably due to the source
low flux density and diffuse background around it, which prevented its automatic
detection by the source search software.
Galactic latitude [deg]
90
60
30
0
−30
−60
XTE J1118+480
LS I +61 303
Microquasars in the Galaxy
Cyg X−1
GRS 1758−258
GRS 1915+105
SS 433
Cyg X−3
LS 5039
Sco X−1
V4641 Sgr
XTE J1748−288
GRO J1655−40
Cir X−1
GX 339−4
1E1740.7−2942
XTE J1550−564
LMXB
LMXB (rel. jets ?)
HMXB
HMXB (rel. jets ?)
−90
180 120 60 0 300 240 180
Galactic longitude [deg]
Figure 4.1: Distribution of known microquasars in galactic coordinates. Filled circles
represent those sources where relativistic jets have been imaged, while open circles are
used for those where hints of relativistic jets have been seen or are clearly suspected.
The solid lines delimit the area of the sky covered by our search, where the recovered
sources are marked in boldface.
After removing the previously known sources of our sample, we ended with a
total of 12 new unidentified REXB candidates. Among them, 7 had offsets between
the X-ray and radio positions within the 1σ RBSC position error, and belong to the
hereafter Group 1 sample. The other 5 sources had offsets between 1–2σ and form

4.2. Cross-identification between RBSC and NVSS catalogs 125
the Group 2 sample.
When we started this project in 1998 another source fullfiled all the selection
criteria. Now we have performed the cross-identification process again, in order to
present here an updated version of it, and we have found that this source, namely
1RXS J072418.3−071508, was identified by Perlman et al. (1998) as a quasar
(WGA J0724.3−0715 in their paper and listed as PMN J0724−0715 in the NED
database). Although it is clear that it is not any more a REXB candidate, we have
preferred to include it in the list and present the observational results obtained up
to now in the following chapters. Since the corresponding offset in position for this
object is less than 1σ, it is included in Group 1.
All the sources belonging to Group 1 (a total of 8), Group 2 (a total of 5)
and the already known REXBs are listed in Table 4.1, where Cols. 1 to 4 contain
the RBSC names (which also provide the positions), the 1σ errors in position, the
count rates, and HR1. In Col. 5 we show the offsets between the RBSC and the
NVSS positions. In Cols. 6 to 8 we list the NVSS names (which provide a limited
position information), the 1σ errors in position and the NVSS flux densities (at
20 cm wavelength). Finally, galactic coordinates for all sources are listed in Cols. 9
and 10. The horizontal line divides the two groups of sources and the already known
REXBs.
Here we will focus on the Group 1 sources, while Group 2 sources will be eventu-
ally studied in the future. In order to have a look at the targets, we show in Fig. 4.2,
for each source of Group 1, the NVSS radio contours overlaid on 6 ′ ×6 ′ optical DSS1
images. To mark the RBSC positions we have plotted as open crosses the 3σ errors
in position, allowing a better display than if we had plotted the 1σ errors.
As can be seen in Table 4.1, the NVSS errors are much smaller than the RBSC
ones. Hence, the radio positions are more accurate than the X-ray ones to look
for optical counterparts. An inspection of the DSS1 images in Fig. 4.2 reveals that
some candidates have an optical counterpart approximately in the middle of the
radio contours, while others do not. This is understandable because of the low
galactic latitude of these objects, which implies a high degree of extinction along
the line of sight. On the other hand, crowded fields near the galactic plane also
prevent us to be sure about the counterparts detected. Therefore, the next step of
the process is to obtain accurate radio positions and then search for possible optical
counterparts, which is presented in the next chapter.

126 Chapter 4. The cross-identification method
Table 4.1: Selected sources from the RBSC/NVSS cross-identification. Cols. 1 to 4
contain the RBSC name (which also provides the position), the 1σ error in position,
the count rate, and the hardness ratio 1 for each source. In Col. 5 we show the offset
between the RBSC and the NVSS positions, while in Cols. 6 to 8 we list the NVSS name
(constructed with truncated coordinates), the 1σ error in position and the NVSS flux
density for each source. Galactic coordinates for all sources are listed in Cols. 9 and
10. The horizontal lines divide Group 1 (top), Group 2 (middle), and the already known
REXB sources (bottom). The source 1RXS J072418.3−071508 is a quasar (see text).
RBSC NVSS Gal. coord.
1RXS name Pos. err. Count rate HR1 Off. NVSS name Pos. err. Flux density l b
[ ′′ ] [10 −2 s −1 ] [ ′′ ] [ ′′ ] [mJy] [ ◦ ] [ ◦ ]
J001442.2+580201 9 8.5 ± 1.4 1.00 ± 0.12 3 J001441+580202 3 7.1 ± 0.5 118.07 −4.49
J013106.4+612035 7 25.0 ± 2.4 0.90 ± 0.05 6 J013107+612033 1 19.1 ± 0.7 127.67 −1.16
J042201.0+485610 14 5.1 ± 1.1 1.00 ± 0.24 4 J042200+485607 7 2.3 ± 0.4 154.41 −0.63
J062148.1+174736 8 8.8 ± 1.6 1.00 ± 0.13 4 J062147+174734 1 12.2 ± 0.5 193.78 +1.72
J072259.5−073131 8 17.4 ± 2.1 0.94 ± 0.05 6 J072259−073135 1 84.1 ± 3.1 223.24 +3.52
J072418.3−071508 23 5.2 ± 1.2 1.00 ± 0.13 19 J072417−071519 1 330.4 ± 9.9 223.15 +3.93
(Quasar)
J181119.4−275939 14 14.0 ± 3.1 0.87 ± 0.14 13 J181120−275946 1 27.9 ± 1.0 3.64 −4.43
J185002.8−075833 13 6.1 ± 1.6 1.00 ± 0.24 10 J185002−075842 1 49.9 ± 1.6 25.66 −3.32
J050339.8+451715 11 6.1 ± 1.3 0.85 ± 0.17 16 J050339+451658 1 34.3 ± 1.1 161.81 +2.31
J080451.8−274924 11 7.0 ± 1.8 1.00 ± 0.15 14 J080451−274911 1 821 ± 25 245.80 +2.01
J082404.3−302033 8 17.5 ± 2.4 0.96 ± 0.06 11 J082403−302038 1 85.0 ± 2.6 250.23 +4.11
J190333.1+104355 14 10.8 ± 1.7 1.00 ± 0.04 16 J190333+104409 7 4.7 ± 0.6 43.86 +2.22
J205644.3+494011 16 7.7 ± 1.0 1.00 ± 0.06 17 J205642+494005 1 167.3 ± 5.0 89.32 +2.76
J024033.5+611358 13 5.4 ± 1.1 1.00 ± 0.17 19 J024031+611345 1 42.2 ± 1.3 135.68 +1.09
(LS I +61 303)
J182615.1−145034 11 6.5 ± 1.6 1.00 ± 0.16 21 J182614−145054 1 23.4 ± 0.9 16.88 −1.29
(LS 5039)
J191149.7+045857 8 52.6 ± 3.7 0.96 ± 0.03 3 J191149+045858 1 867 ± 26 39.69 −2.24
(SS 433)
J203226.2+405725 7 284.9 ± 6.5 0.98 ± 0.00 7 J203225+405728 1 87.3 ± 3.2 79.84 +0.69
(Cyg X-3)

4.2. Cross-identification between RBSC and NVSS catalogs 127
DECLINATION (J2000)
1RXS J001442.2+580201
58 05
04
03
02
01
00
57 59
00 15 00 14 55 50 45 40 35 30 25 20
RIGHT ASCENSION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
1RXS J042201.0+485610
48 59
58
57
56
55
54
-07 29
30
31
32
33
34
-27 57
04 22 15 10 05 00 21 55 50 45
RIGHT ASCENSION (J2000)
1RXS J072259.5-073131
07 23 10 05 00 22 55 50
RIGHT ASCENSION (J2000)
58
59
-28 00
01
02
1RXS J181119.4-275939
18 11 30 25 20 15 10
RIGHT ASCENSION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
61 23
22
21
20
19
18
1RXS J013106.4+612035
01 31 30 25 20 15 10 05 00 30 55 50 45
RIGHT ASCENSION (J2000)
17 50
49
48
47
46
45
1RXS J062148.1+174736
06 22 00 21 55 50 45 40
RIGHT ASCENSION (J2000)
-07 13
14
15
16
17
18
-07 56
57
58
59
-08 00
01
1RXS J072418.3-071508
07 24 25 20 15 10 05
RIGHT ASCENSION (J2000)
1RXS J185002.8-075833
18 50 15 10 05 00 49 55
RIGHT ASCENSION (J2000)
Figure 4.2: Optical, radio and X-ray composition of Group 1 sources. The NVSS radio
contours are overlaid on the 6 ′ × 6 ′ optical DSS1 images, whereas the open crosses
denote the 3σ RBSC astrometric errors.

128 Chapter 4. The cross-identification method
4.3 Conclusions
We have presented a cross-identification method to search for REXBs, which are
potential microquasar sources. The obtained results give confidence to the proposed
method, since the output list of objects included all but one known persistent HMXB
microquasars within |b| < 5 ◦ . The unidentified sources in the list have been divided
in two groups, depending on the offset between the RBSC and NVSS positions.
The Group 1 source list contains 8 objects with offsets smaller than 1σ, while the
Group 2 source list contains 5 objects with offsets between 1 and 2σ.
Finally, since some microquasars could have been ejected from the galactic plane
(Ribó et al. 2002), we are planning to extend this project up to higher galactic
latitudes (5 ◦ < |b| < 10 ◦ ).

Bibliography
Condon, J. J., Cotton, W. D., Greisen, E. W., et al. 1998, AJ, 115, 1693
Massi, M., Ribó, M., Paredes, J. M., Peracaula, M., & Estalella, R. 2001, A&A,
376, 217
Motch, C., Guillout, P., Haberl, F., et al. 1998, A&AS, 132, 341
Perlman, E. S., Padovani, P., Giommi, P., et al. 1998, AJ, 115, 1253
Ribó, M., Paredes, J. M., Romero, G. E., et al. 2002, A&A, 384, 954
Voges, W., Boller, Th., Dennerl, K., et al. 1996, in Proc. of the Conference Rönt-
genstrahlung from the Universe, MPE Report, 263, 637
Voges, W., Aschenbach, B., Boller, Th., et al. 1999, A&A, 349, 389
129

130 BIBLIOGRAPHY

Chapter 5
Radio and optical observations
5.1 Introduction
After the cross-identification of catalogs presented in the previous chapter, the next
logical step of the process 1 is to obtain accurate radio positions, that will be ex-
plained in Sect. 5.2, and then search for possible optical counterparts, that will be
presented in Sect. 5.3. A discussion on the obtained results is carried out in Sect. 5.4,
while we state our conclusions in Sect. 5.5.
5.2 Radio observations
The main goal of the radio observations was to obtain accurate sub-arcsecond po-
sitions, but we also wished to monitor the variability of radio flux and spectrum
of our targets. On the other hand, it was also important to investigate the source
structure beyond the NVSS resolution. For REXBs, most of the source flux den-
sity is expected to be concentrated in a compact core plus possible arcsecond or
sub-arcsecond extended features. By observing each target with the VLA in A con-
figuration, we were able to verify to what extend our sources were indeed compact,
and to look for possible elongations or jets.
To this end, multi-frequency and multi-epoch observations of the Group 1 sources
1 Published in Paredes, J. M., Ribó, M., & Martí, J. 2002, A&A, 394, 193.
131

132 Chapter 5. Radio and optical observations
were carried out with the NRAO 2 VLA in A configuration. We did not observe
1RXS J181119.4−275939 and 1RXS J185002.8−075833, the last two sources of
Group 1 in Table 4.1, because they had very different right ascensions, compared to
the other sources, and were not visible during the scheduled VLA observing time.
Three VLA A configuration observing sessions were carried out on 1999 July 9,
22 and 30 at 3.6 cm (8.4 GHz) and 6 cm (5 GHz) wavelengths (frequencies). A
typical observation consisted of 14 minutes at 3.6 cm and 4 minutes at 6 cm on each
target, preceded and followed by a 2 minute observation of a phase calibrator. Dur-
ing the last observing session, July 30, we included two additional 20 cm (1.4 GHz)
measurements of 1RXS J072259.5−073131 and 1RXS J072418.3−071508. The am-
plitude calibrator used in all cases was 3C 48, while we used the phase calibrator
J0102+584 for the first two sources, J0359+509 and J0603+177 for the third and
fourth sources, respectively, and finally J0730−116 for the last two sources. The
data were edited and calibrated using the aips software package of NRAO.
5.2.1 Results
All sources were detected at all frequencies, allowing us to obtain accurate positions
and flux density measurements. To achieve the first goal, we made use of the phase-
reference technique to calibrate the data at 3.6 cm wavelength, the ones with the
highest angular resolution. Then we concatenated all these data for each source,
with the task dbcon within aips. Finally, we performed uniform weighted maps
from which positions were obtained after fitting elliptical Gaussians using jmfit
within aips. A realistic estimate of the position error is about 0.01 ′′ .
In order to obtain good quality maps, once accurate positions were known, we
carried out a similar process, but using phase self-calibration with the previously
obtained positions for each individual snapshot at 3.6 cm wavelength. This was
not possible for 1RXS J042201.0+485610 because the flux density was too low and
prevented self-calibration. The resulting uniform weighted maps, together with the
optical images that will be presented in the following section, are shown in Fig. 5.1.
The flux densities of all sources were measured as follows. First of all we self-
2 The National Radio Astronomy Observatory is a facility of the National Science Foundation
operated under cooperative agreement by Associated Universities, Inc.

5.2. Radio observations 133
calibrated all snapshots at all wavelengths (again with the exception of the source
1RXS J042201.0+485610). Then we performed natural weighted maps with im-
proved dynamic range, from which flux densities were measured using again elliptical
Gaussian fits.
The obtained results for all observed sources are summarized in Table 5.1. In
Col. 1 we list the RBSC object names, while in Cols. 2 and 3 we show the radio posi-
tions derived from our phase-referenced uniform weighted maps of all concatenated
3.6 cm wavelength observations. In Col. 4 we list the observing dates and, in Cols. 5
and 6, the corresponding flux densities at both observing wavelengths, measured on
the self-calibrated natural weighted maps. The flux density errors quoted are the
rms noise of the maps in mJy beam −1 . In Col. 7 we show the computed spectral
index α (defined in such a way that Sν ∝ ν +α , where Sν is the flux density at a
given frequency ν) of each source for all epochs. The radio spectra corresponding
to the values listed in Table 5.1 are plotted in Fig. 5.2 for all sources, including the
20 cm wavelength flux densities measured on July 30 for the last two sources. We
have also plotted, for all sources, the non-simultaneous 20 cm measurements from
the NVSS survey obtained in VLA D configuration.
Inspection of all obtained maps (with phase-reference or self-calibration tech-
niques, with uniform or natural weights, of individual snapshots or after concate-
nating all of them, and at all frequencies), reveals that all sources are point-like
except 1RXS J072259.5−073131, which presents extended structure towards the
east, marginally present in the 3.6 and 6 cm maps, and visible as a one-sided jet in
the 30 July snapshot observation at 20 cm wavelength (see Fig. 5.3).

DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
134 Chapter 5. Radio and optical observations
1RXS J001442.2+580201 I band
58 03 00
02 45
30
15
00
01 45
30
15
58 02 02.0
00 14 48 46 44 42 40 38 36
RIGHT ASCENSION (J2000)
01.5
01.0
00.5
VLA 3.6cm
00 14 42.25 42.20 42.15 42.10 42.05 42.00
RIGHT ASCENSION (J2000)
17 48 30
15
00
47 45
30
15
00
46 45
17 47 36.0
35.5
35.0
34.5
1RXS J062148.1+174736 I band
06 21 51 50 49 48 47 46 45 44
RIGHT ASCENSION (J2000)
VLA 3.6cm
06 21 47.82 47.80 47.78 47.76 47.74 47.72 47.70 47.68
RIGHT ASCENSION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
1RXS J013106.4+612035 I band
61 21 30
15
00
20 45
30
15
00
19 45
61 20 34.0
01 31 14 12 10 08 06 04 02 00
RIGHT ASCENSION (J2000)
33.5
33.0
32.5
-07 30 45
VLA 3.6cm
01 31 07.35 07.30 07.25 07.20 07.15 07.10
RIGHT ASCENSION (J2000)
31 00
15
30
45
32 00
15
30
-07 31 34.0
1RXS J072259.5-073131 I band
07 23 03 02 01 00 22 59 58 57 56
RIGHT ASCENSION (J2000)
34.5
35.0
35.5
VLA 3.6cm
07 22 59.74 59.72 59.70 59.68 59.66 59.64 59.62
RIGHT ASCENSION (J2000)
(See figure caption on next page)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
DECLINATION (J2000)
48 57 00
56 45
30
15
00
55 45
30
15
1RXS J042201.0+485610 I band
04 22 06 04 02 00 21 58 56
RIGHT ASCENSION (J2000)
48 56 04.5
04.0
03.5
03.0
-07 14 30
45
15 00
15
30
45
16 00
15
VLA 3.6cm
04 22 00.60 00.55 00.50 00.45
RIGHT ASCENSION (J2000)
1RXS J072418.3-071508 I band
07 24 21 20 19 18 17 16 15 14
RIGHT ASCENSION (J2000)
-07 15 19.5
20.0
20.5
21.0
21.5
22.0
22.5
23.0
SE
NW
VLA 3.6cm
07 24 17.45 17.40 17.35 17.30 17.25 17.20
RIGHT ASCENSION (J2000)

5.3. Optical observations 135
Figure 5.1: Optical images and radio maps, grouped in pairs, of the six sources listed in
Table 5.1. The 2 ′ × 2 ′ optical images were obtained through the I Johnson filter with
the CAHA 2.2 m telescope on 1999 December 11, and circles have been used to clearly
mark the optical counterparts. The radio maps, computed using uniform weights, are
the result of concatenating, for each source, all VLA 1999 July observations at 3.6 cm
wavelength, while the crosses indicate the optical positions with the 1σ 0.3 ′′ astrometric
errors. The radio maps are 2 ′′ × 2 ′′ wide for all sources except for the last one, where a
4 ′′ × 4 ′′ map has been plotted to allow the inclusion of the close SE optical object. The
synthesized beam is plotted in the bottom left corner of each map. The radio contours
start at a signal-to-noise ratio of 6 for all sources except for the last one, where a value
of 15 has been used, in order to avoid reproducing cleaning effects at low signal-to-noise
ratios in all cases.
5.3 Optical observations
Having in mind the goal of identifying the optical counterparts, CCD observations
were made at Calar Alto (Almería, Spain) with the 1.5 m telescope of the Observa-
torio Astronómico Nacional (OAN), between 24 and 30 November 1998. We used
the Ritchey-Chrétien focus and a Tektronics TK1024AB chip that provides a scale
factor of 0.4 ′′ per pixel and a 6.9 ′ × 6.9 ′ field of view. Deep CCD images were ob-
tained for all our candidates, with the exception of 1RXS J181119.4−275939 and
1RXS J185002.8−075833 that were not visible at the epoch of our observations. The
images were taken with the V , R and I Johnson filters, and the exposure times were
typically of 600–1200 seconds. Although the images were useful for photometric
measurements, their quality did not allow us to obtain accurate positions.
Complementary CCD observations were conducted on 1999 December 11 also at
Calar Alto, but using in this case the 2.2 m telescope of the Centro Astronómico
Hispano-Alemán (CAHA). Images were obtained using the Ritchey-Chrétien focus
and CAFOS, leading a scale factor of 0.53 ′′ per pixel and a 8.8 ′ × 8.8 ′ field of view,
through the I Johnson filter and with exposure times between 180 and 600 seconds.
The good quality of these images allowed us to perform accurate astrometric mea-
surements, although no precise photometric information could be obtained since the
images could not be photometrically calibrated.
The observations were reduced using standard procedures within IRAF.

136 Chapter 5. Radio and optical observations
Table 5.1: Radio positions, flux densities and spectral indices obtained after the
VLA A configuration observations for all sources belonging to Group 1 except
1RXS J181119.4−275939 and 1RXS J185002.8−075833. The coordinates are those
obtained from the uniform weighted maps of all concatenated 3.6 cm wavelength obser-
vations. The three rows of flux densities and spectral index for each source correspond
to the three observing runs, i.e., July 9, 22, and 30, respectively. The flux densities at
3.6 and 6 cm have been obtained from natural weighted self-calibrated maps for each
source, with the exception of 1RXS J042201.0+485610, where self-calibration was not
possible.
1RXS name α (2000) δ (2000) Flux density [mJy] Spectral index
[h, m, s] [ ◦ , ′ , ′′ ] S3.6 cm S6 cm α3.6−6 cm
J001442.2+580201 00 14 42.1262 +58 02 01.219 6.94 ± 0.04 7.90 ± 0.07 −0.24 ± 0.02
± 0.0013 ± 0.010 6.36 ± 0.07 7.05 ± 0.09 −0.19 ± 0.03
5.98 ± 0.06 6.47 ± 0.07 −0.14 ± 0.03
J013106.4+612035 01 31 07.2267 +61 20 33.376 16.95 ± 0.04 17.95 ± 0.08 −0.10 ± 0.01
± 0.0014 ± 0.010 16.62 ± 0.07 16.49 ± 0.09 +0.01 ± 0.01
17.43 ± 0.07 17.95 ± 0.08 −0.05 ± 0.01
J042201.0+485610 04 22 00.5244 +48 56 03.634 0.70 ± 0.03 0.30 ± 0.04 +1.6 ± 0.3
± 0.0010 ± 0.010 0.74 ± 0.05 0.57 ± 0.05 +0.5 ± 0.2
0.71 ± 0.03 0.35 ± 0.04 +1.3 ± 0.2
J062148.1+174736 06 21 47.7522 +17 47 35.078 9.63 ± 0.04 8.37 ± 0.07 +0.26 ± 0.02
± 0.0007 ± 0.010 9.53 ± 0.07 9.87 ± 0.07 −0.06 ± 0.02
11.45 ± 0.05 10.40 ± 0.07 +0.18 ± 0.01
J072259.5−073131 07 22 59.6809 −07 31 34.805 40.90 ± 0.05 47.67 ± 0.08 −0.280 ± 0.004
(a) ± 0.0007 ± 0.010 44.29 ± 0.07 44.5 ± 0.1 −0.009 ± 0.005
45.37 ± 0.08 54.29 ± 0.08 −0.329 ± 0.004
J072418.3−071508 07 24 17.2912 −07 15 20.339 438.5 ± 0.2 423.4 ± 0.3 +0.064 ± 0.002
(b) ± 0.0007 ± 0.010 455.75 ± 0.09 439.3 ± 0.2 +0.067 ± 0.001
428.5 ± 0.3 404.9 ± 0.3 +0.104 ± 0.002
(a) Also observed at 20 cm on 1999 July 30, with a flux density of 55.1 ± 0.4 mJy, and showing a
one-sided radio jet.
(b) Also observed at 20 cm on 1999 July 30, with a flux density of 248.3 ± 0.9 mJy.

5.3. Optical observations 137
Flux density [mJy]
9
8
7
6
20
18
16
3
2
1
0.5
0.3
14
12
10
8
100
80
60
40
500
400
300
200
1RXS J001442.2+580201
1RXS J013106.4+612035
1RXS J042201.0+485610
1RXS J062158.1+174736
1RXS J072259.5−073131
1RXS J072418.3−071508
1 2 5 8 10
Frequency [GHz]
Figure 5.2: Radio spectra of the six sources listed in Table 5.1 at different epochs. The
VLA data from July 9, 22 and 30 are denoted by filled circles, open circles and open
triangles, respectively. The non-simultaneous NVSS data, at 20 cm wavelength, are
plotted as filled squares. Both axes are in logarithmic scale. Error bars not visible are
smaller than the symbol’s size.

138 Chapter 5. Radio and optical observations
DECLINATION (J2000)
-07 31 28
30
32
34
36
38
40
42
44
1RXS J072259.5-073131 VLA 20cm
07 23 00.2 00.0 22 59.8 59.6 59.4 59.2
RIGHT ASCENSION (J2000)
Figure 5.3: Natural weighted map of the self-calibrated 20 cm data of
1RXS J072259.5−073131 obtained on the July 30 snapshot observation. Contours cor-
respond to −3, 3, 4, 5, 7, 10, 15, 25, 40, 60, 85, 110 and 130 times 0.4 mJy beam −1 ,
the rms noise.
5.3.1 Results
Promising optical counterparts for all observed sources, i.e., the first six ones in Ta-
ble 4.1, were found in both observing runs. Thanks to the CAHA 2.2 m observations
we were able to obtain accurate positions. For this purpose, we performed a detailed
astrometric reduction of each image using field stars present in the USNO-A2.0 cat-
alog (Monet et al. 1999). First of all we selected point-like objects in the image, not
elliptical or binary objects due to crowded fields, and determined their positions (X,
Y ). We also rejected very faint objects to allow a significant signal-to-noise ratio.
On the other hand, very bright objects were also rejected, because the positions
given in the USNO-A2.0 catalog are from plates obtained around 1950, and nearby
(bright) stars could have experienced a shift in position since then due to relatively
high proper motions. Then we fitted an astrometric solution to our image using
the USNO-A2.0 (α, δ) and image (X, Y ) positions for the common stars in both

5.4. Discussion 139
the catalog and the image, and rejected spurious points above 3σ. We proceeded
iteratively until convergence was achieved. This process allowed us to obtain plate
solutions with an rms of ∼ 0.2 ′′ in each coordinate. This relatively high error prob-
ably arises from the non-zero proper motions of the field stars finally used in the
fit, between 36 and 155 depending on each image. We have to quadratically add to
this error the 1σ empirical uncertainty estimate of the USNO-A2.0 catalog relative
to the ICRF, which is also ∼ 0.2 ′′ in each coordinate (see Table 1 of Deutsch 1999).
Therefore, the final 1σ error of our obtained coordinates is estimated to be 0.3 ′′ .
We show the central 2 ′ × 2 ′ of the CAHA 2.2 m images in Fig. 5.1, with the
optical counterparts marked with a circle (which is not any X-ray or radio er-
ror box). All sources appeared basically as point-like objects, except the quasar
1RXS J072418.3−071508, which had a complex structure with two separate opti-
cal components, hereafter NW (northwest) and SE (southeast). The NW position
falls well within the NVSS error box and close to our obtained radio position. On
the contrary, the SE position is clearly out the NVSS error box, and cannot be
considered as a reliable counterpart to the radio source.
Photometry was obtained thanks to the OAN 1.5 m observations. The absolute
photometry is believed to be accurate only to ±0.1 magnitude in the best cases, be-
cause we could only determine the approximate photometric zero point by observing
a few standard stars from Landolt (1992).
The obtained results are presented in Table 5.2, where Col. 1 gives the RBSC
object name, Cols. 2 and 3 give the J2000.0 ICRS coordinates of the optical po-
sition with the 1σ error of 0.3 ′′ below each coordinate, Col. 4 gives the observing
dates and Cols. 5 to 7 give the Johnson V , R and I magnitudes. Magnitudes for
1RXS J072418.3−071508 correspond to the sum of both the NW and SE compo-
nents.
5.4 Discussion
The observations reported here provide accurate positions in the optical (0.3 ′′ ) and
specially in the radio (0.01 ′′ ) domains. The astrometric agreement for each source
can be seen in the radio maps of Fig. 5.1, and is numerically expressed in Table 5.3,
where we list the offsets in right ascension, in declination and the total offset between

140 Chapter 5. Radio and optical observations
Table 5.2: Optical astrometric and photometric results for all sources listed in Table 5.1.
Accurate positions were obtained from the I Johnson filter CAHA 2.2 m telescope
observations carried out on 1999 December 11. Magnitudes in V , R and I Johnson
filters were obtained in different dates of November 1998 using the OAN 1.5 m telescope.
Note that two positions are given for the source 1RXS J072418.3−071508.
1RXS name α (2000) δ (2000) Day Magnitude
[h, m, s] [ ◦ , ′ , ′′ ] Nov. 1998 V R I
J001442.2+580201 00 14 42.111 +58 02 01.32 25 - - > 19.0
± 0.038 ± 0.30 27 - - 20.1 ± 0.4
27 - - 19.7 ± 0.2
J013106.4+612035 01 31 07.267 +61 20 33.60 27 19.5 ± 0.1 18.8 ± 0.1 17.9 ± 0.1
± 0.042 ± 0.30
J042201.0+485610 04 22 00.533 +48 56 03.84 28 20.4 ± 0.2 18.8 ± 0.1 17.4 ± 0.1
± 0.030 ± 0.30 29 - - 17.5 ± 0.1
J062148.1+174736 06 21 47.748 +17 47 35.23 28 19.8 ± 0.1 18.6 ± 0.1 17.5 ± 0.1
± 0.021 ± 0.30 29 - - 17.6 ± 0.1
J072259.5−073131 07 22 59.689 −07 31 34.78 28 18.5 ± 0.1 - -
± 0.020 ± 0.30 29 18.3 ± 0.1 17.7 ± 0.1 16.8 ± 0.1
J072418.3−071508(NW) 07 24 17.299 −07 15 20.46 29 18.8 ± 0.1 18.0 ± 0.1 17.2 ± 0.1
± 0.020 ± 0.30
(SE) 07 24 17.387 −07 15 22.33
± 0.020 ± 0.30
the optical and radio positions. As can be seen, the offsets in each coordinate are
always smaller than the 1σ optical errors in position.
Although the agreements in position are certainly very promising, we have esti-
mated the probability of a random coincidence. For this purpose, we have obtained
the limiting I magnitude of our CAHA 2.2 m images and counted the number
of objects until this threshold. Then we have estimated the object-density of our
images by assigning to each object the area of the 1σ error box in position, i.e.,
(2 × 1σ) 2 = (0.6 ′′ ) 2 . Finally, we have divided the area occupied by the stars by
the area of the image, and obtained a naive estimate of the probability to find an
optical object, up to a given limiting I magnitude, close to the radio position by
less than 1σ, or 0.3 ′′ , in each coordinate. The limiting I magnitudes and obtained
probabilities of random coincidence are listed in the last two columns of Table 5.3.

5.4. Discussion 141
Table 5.3: Offsets from the radio to the optical positions in right ascension, in declination
and the total offset. The limiting I magnitude of our CAHA 2.2 m images and the
probability of a random coincidence are listed in the last two columns.
1RXS name αo − αr [ ′′ ] δo − δr [ ′′ ] o-r [ ′′ ] Limiting I mag. Probability [%]
J001442.2+580201 −0.12 0.10 0.16 22 0.3
J013106.4+612035 0.29 0.22 0.37 21 0.3
J042201.0+485610 0.08 0.21 0.22 20 0.2
J062148.1+174736 −0.06 0.15 0.16 20 0.2
J072259.5−073131 0.12 0.02 0.12 19 0.2
J072418.3−071508 0.12 −0.12 0.17 18 0.2
Since these probabilities are always smaller than 1%, and even much smaller if we
consider the particular offsets instead of the 1σ optical errors, we conclude that
probably all optical objects are the counterparts of the radio sources, and none of
them is a field star not related to the radio source.
On the other hand, an analysis of observed radio spectra of X-ray binaries can
be found in Fender (2001). It seems clear that negative spectral indices (−1 ≤ α ≤
−0.2) are detected when observing synchrotron emission from expanding plasmons,
which are the result of discrete ejections after an outburst. On the contrary, flat
or inverted radio spectra (0.0 ≤ α ≤ 0.6) are typical of the low/hard X-ray state
of black hole candidates, and are believed to arise in synchrotron emission from a
partially self-absorbed jet. Hence, a variety of spectral indices can be found in the
objects we are looking for. However, inverted spectra with α > 1 at high radio
frequencies (above 5 GHz) would probably rule out a REXB nature.
REXBs often display a variable flux, although extragalactic sources may vary as
well. Our observations, performed at one/two week interval, allow us to estimate the
degree of variability given by (Smax − Smin)/(Smax + Smin), being Smax and Smin the
maximum and the minimum flux density, respectively. As it is clear from Table 5.1
and Fig. 5.2, all sources in the sample show some degree of flux density variability at
centimeter wavelengths during the three runs of our monitoring. This variability is
always within 10%, except for 1RXS J042201.0+485610 at 6 cm, where a variability
up to ∼ 30% is detected, although it should be considered with caution due to the
low emission level.

142 Chapter 5. Radio and optical observations
The optical emission seen in microquasars arises from the non-degenerated star
of the system, i.e., the companion of the compact object. Hence, in microquasars
containing a high mass companion spectral types O or B are found, while in those
containing a low mass companion the spectral type is later than A (White et al.
1995). Therefore, as optical counterpart of a given candidate, we expect to find a
non-degenerated star with any spectral type, and with a luminosity class ranging
from the main sequence to supergiant. In fact, once photometric magnitudes of an
object through different filters and its distance are known, one can easily deduce
its spectral type assuming it is a star. However, we have no information about the
distance to our sources, and no correction for extinction can be applied to our data.
This is particularly important in our case, since our search has been carried out at
low galactic latitudes, where a high degree of extinction is expected, preventing any
spectral type classification from the photometric data alone.
5.4.1 Discussion on individual sources
1RXS J001442.2+580201. The radio counterpart, see Table 5.1, shows a mod-
erate degree of variability but always displaying a negative spectral index of ∼ −0.2,
suggesting an optically thin or partially self-absorbed synchrotron emitter. As can
be seen in Fig. 5.2, the non-simultaneous NVSS flux density measurement at 20 cm
wavelength is compatible with this behavior or even with a flattening of the spec-
trum at longer wavelengths, although this could be due to intrinsic variability. Non-
thermal synchrotron radiation remains therefore as the most plausible interpretation
for the radio emission of this source. The optical counterpart to the radio source
(Fig. 5.1) appears as a point-like source with I ∼ 20, which makes it the weakest
optical object of our sample. No R or V magnitudes could be obtained, although
a typical behavior for low latitude highly absorbed sources is an increase of ∼ 1
magnitude when changing from I to R and from R to V , as can be seen in the other
sources of Table 5.2. All the available information points towards a good REXB
candidate.
1RXS J013106.4+612035. In the radio domain it shows a slightly negative or
close to zero spectral index, reminiscent of optically thin or partially self-absorbed
jets in REXBs, in which the non-simultaneous NVSS flux measurement fits well, and
it displays a very low degree of variability of 2% at 3.6 cm and 4% at 6 cm. This

5.4. Discussion 143
object has also been observed within the Green Bank 5 GHz radio survey (Gregory
& Condon 1991; Gregory et al. 1996) with a measured flux density of 24 mJy. This
value is a little bit higher than the ones reported here, although this discrepancy
could be due to intrinsic variability and/or to the different angular scales sampled
by the interferometric VLA observations. In fact, VLA A configuration observations
carried out by Laurent-Muehleisen et al. (1997) on October 19, 1992, indicate a flux
density of 16 mJy at 5 GHz, in good agreement with our data. In the optical domain,
the counterpart to the radio source appears as a point-like object with a nearby field
object at ∼ 3 ′′ to the northeast, visible in Fig. 5.1, which is ∼ 1 magnitude fainter
in the I band. The optical magnitudes in Table 5.2 correspond to the sum of both
objects (unresolved in the OAN 1.5 m images), and are typical of low latitude highly
absorbed sources. Overall it seems a good REXB candidate.
1RXS J042201.0+485610. In the RBSC this source appears catalogued with
an X-ray extension of 17 ′′ , which could suggest an extragalactic origin of the source.
However, X-ray halos produced by the scattering of the original X-ray photons by
interstellar dust grains are seen in some galactic microquasars, like for example in a
Chandra observation of Cyg X-3 (Predehl et al. 2000). In fact, Cyg X-3 has an X-ray
extension of 63 ′′ within the RBSC, while SS 433 has one of 25 ′′ . Hence, the presence
of an X-ray extension does not rule out a REXB nature for this source. The radio
counterpart shows the lowest radio emission level of all sources in our sample, and
displays an inverted spectrum up to α ∼ +1.6 (see Table 5.1), difficult to account
for self-absorbed synchrotron radiation at such high frequencies. Nevertheless, as
it happens in some AGN, a highly inverted spectrum would be expected if the jet
is stopped in a dense medium on a compact scale. On the other hand, the non-
simultaneous NVSS flux density measurement seems to be incompatible with self-
absorbed synchrotron radiation. Although this could be due to intrinsic variability
of the source, this would require variations of an order of magnitude in flux. This
discrepancy can be explained if we assume a thermal origin for the radio emission and
if we attribute the drop in flux density in our VLA A configuration observations due
to over resolution when compared to the NVSS VLA D configuration observations.
The optical counterpart to the radio source has a nearby field object at ∼ 3 ′′ to
the west, marginally visible in Fig. 5.1, which is ∼ 2 magnitudes fainter in the I
band. The optical magnitudes in Table 5.2 correspond to the sum of both objects
(unresolved in the OAN 1.5 m images), although the nearby one is not expected
to contaminate appreciably, and are again typical of low latitude highly absorbed

144 Chapter 5. Radio and optical observations
sources. Although at first sight our target appears as a point-like source in Fig. 5.1,
if we perform Gaussian fits to field objects with similar magnitudes, we find our
target to have a Full Width at Half Maximum (FWHM) ∼ 15% greater, suggesting
that it is in fact an extended optical object. In summary, the high spectral index
and the extended optical emission indicate that this source is not a very promising
REXB candidate.
1RXS J062148.1+174736. In the radio domain, this source displays an inverted
or flat spectrum, which could account for a partially self-absorbed jet. On the
other hand, the moderate variability observed, around 10%, could explain the higher
non-simultaneous NVSS flux density measurement. In the optical domain, this
source appears as a point-like object with magnitudes typical of low latitude highly
absorbed sources. However, a careful inspection of the optical image, reveals that
the FWHM of a Gaussian fit to our target source is ∼ 30% greater than those of
other sources with similar magnitudes. This fact points to an extragalactic nature
for this source. It is worth to mention that this source is the only one in our
sample present in the second incremental release of the Two Micron All Sky Survey
(2MASS) (Cutri et al. 2000). Its position, obtained from images taken in October
1997, is there listed as α = 6 h 21 m 47.749 s ± 0.009 s , δ = +17 ◦ 47 ′ 35.07 ′′ ± 0.13 ′′ ,
in very good agreement with our obtained coordinates. Finally, we would like to
point out that Motch et al. (1998) list 3 stars within the X-ray error box as possible
counterparts to the X-ray source. Our proposed optical counterpart on the basis of
the radio position, would rule out any of these 3 stars as the optical counterpart of
this X-ray source. Overall, it looks like an extragalactic object because of its optical
extension.
1RXS J072259.5−073131. The radio counterpart presents an extended struc-
ture towards the east, marginally present in the 3.6 and 6 cm maps, and visible as
a one-sided jet in the 30 July 4 minute snapshot observation at 20 cm wavelength
(see Fig. 5.3). Although the one-sided morphology is suggestive of an extragalac-
tic nature for this source, specially if we take into account that we are mapping the
source at arcsecond scales, we cannot rule out a possible galactic nature on the basis
of the detected morphology. The source also shows a moderate degree of variability,
displaying most of time a negative spectral index but also close to 0 in the July 22
observation. As can be seen in Fig. 5.2, the non-simultaneous NVSS flux density

5.5. Conclusions 145
measurement at 20 cm wavelength could be compatible with a non-thermal optically
thin spectrum. However, simultaneous observations during July 30 show a flattening
of the spectrum at higher wavelengths, which could be due to self-absorption in this
energy range. The difference between the NVSS and our flux density measurement
at 20 cm wavelength could be due to intrinsic variability and/or to the different
VLA configurations used. Anyway, non-thermal synchrotron radiation remains as
the most plausible interpretation for the radio emission of this source, specially at
higher frequencies. The optical counterpart appears as a point-like object with the
magnitudes of an absorbed object at low galactic latitudes. These properties are in
good agreement with the expected ones for a microquasar, although the one-sided
radio jet at arcsecond scales is most common in extragalactic objects.
1RXS J072418.3−071508. This source has been recently (March 2002) classi-
fied as a quasar in the SIMBAD database, and it is not any more a REXB can-
didate. It is listed as PMN J0724−0715 in the NED database, and is the source
WGA J0724.3−0715 in Perlman et al. (1998), who reported a faint and quite broad
Hα emission line (rest-frame Wλ = 30.3 ˚A, FWHM=4000 km s −1 ), and classified it
as a flat spectrum radio quasar with z = 0.270. Nevertheless, we have reported here
our observational results for this source, since it was a candidate when we performed
the observations. In the radio it displays a low degree of variability of 3% at 3.6 cm
and 4% at 6 cm, with a flat or even inverted spectrum. In the optical it appears as
a complex double object, with a NW and a SE components, being the first one the
counterpart to the radio source. Magnitudes in V , R and I typical of absorbed objects
at low galactic latitudes were found. On the other hand, it appears catalogued
with an X-ray extension of 40 ′′ in the RBSC.
5.5 Conclusions
After the cross-identification of catalogs presented in Chapt. 4, we found 8 interesting
sources to be studied (Group 1 sources). Here we have reported observations
of all these sources except 1RXS J181119.4−275939 and 1RXS J185002.8−075833,
that were not visible during our observations. We have studied the remaining 6
radio sources of Group 1, and found optical counterparts to all of them. We have
obtained accurate positions at both radio and optical wavelengths, perfectly com-

146 Chapter 5. Radio and optical observations
patible between them. We also have obtained radio spectra and optical magnitudes
of the sources.
After a detailed analysis of the obtained data, we conclude that the sources
1RXS J001442.2+580201 and 1RXS J013106.4+612035 are good REXB candidates,
while 1RXS J042201.0+485610 is not so promising due to its highly inverted spec-
trum at high frequencies and its optical extended emission. A careful study of the
optical counterpart of 1RXS J062148.1+174736 reveals it is extended, which points
towards an extragalactic nature for this object. The situation is not clear in the case
of 1RXS J072259.5−073131, because the optical data agrees with a microquasar na-
ture, while the detected one-sided radio jet reaching arcsecond scales is most common
in extragalactic sources. The last studied source, 1RXS J072418.3−071508, is an
already identified quasar. Hence, two sources in our sample are very promising to
be new microquasars in the Galaxy.
The next logical step was to obtain optical spectroscopy of these sources, to
unveil their galactic or extragalactic nature. Several attempts have been carried out
in years 2000 and 2001, without success due to bad weather conditions, and a new
attempt is in progress for autumn 2002. On the other hand, VLBI observations
of the 6 sources studied here will be published soon (Ribó et al. 2002), and are
reported in the next chapter.

Bibliography
Cutri, R. M., Skrutskie, M. F., Van Dyk, S., et al. 2000, Explanatory Supple-
ment to the 2MASS, Second Incremental Data Release, Caltech (available at
http://www.ipac.caltech.edu/2mass/releases/second/doc/explsup.html)
Deutsch, E. W. 1999, AJ, 118, 1882
Fender, R. P. 2001, MNRAS, 322, 31
Gregory, P. C., & Condon, J. J. 1991, ApJS, 75, 1011
Gregory, P. C., Scott, W. K., Douglas, K., & Condon, J. J. 1996, ApJS, 103, 427
Landolt, A.U. 1992, AJ, 104, 340
Laurent-Muehleisen, S. A., Kollgaard, R. I., Ryan, P. J., et al. 1997, A&AS, 122,
235
Monet, D. G., Bird, A., Canzian, B., et al. 1999, USNO-A2.0 CD-ROM (U.S. Naval
Observatory, Washington DC)
Motch, C., Guillout, P., Haberl, F., et al. 1998, A&AS, 132, 341
Perlman, E. S., Padovani, P., Giommi, P., et al. 1998, AJ, 115, 1253
Predehl, P., Burwitz, V., Paerels, F., & Trümper, J. 2000, A&A, 357, L25
Ribó, M., Ros, E., Paredes, J. M., Massi, M., & Martí, J. 2002, A&A, 394, 983
White, N. E., Nagase, F., & Parmar, A. N. 1995, The properties of X-ray binaries,
in X-ray Binaries, ed. W. H. G. Lewin, J. van Paradijs, & E. P. J. van den Heuvel
(Cambridge Univ. Press, Cambridge), 1
147

148 BIBLIOGRAPHY

Chapter 6
EVN and MERLIN observations
6.1 Introduction
A sample of six radio emitting X-ray sources, among the obtained after the cross-
identification of catalogs reported in Chapt. 4, has been further studied in Chapt. 5
(see Paredes et al. 2002). Here we present VLBI observations 1 of these six sources,
aimed at revealing possible jet-like features at milliarcsecond scales. We describe the
observations and the data reduction in Sect. 6.2, present the results and a discussion
in Sect. 6.3, and state our conclusions in Sect. 6.4.
6.2 Observations and data reduction
We observed the six sources studied in Chapt. 5 simultaneously with the Multi-
Element Radio-Linked Interferometer Network (MERLIN) 2 and the European VLBI
Network (EVN) 3 on 2000 February 29th/March 1st (23:30–23:05 UT) at 5 GHz.
Single dish flux density measurements were carried out with the MPIfR 100 m
antenna in Effelsberg, Germany.
1Published in Ribó, M., Ros, E., Paredes, J. M., Massi, M., & Martí, J. 2002, A&A, 394, 983.
2MERLIN is operated as a National Facility by the University of Manchester at Jodrell Bank
Observatory on behalf of the UK Particle Physics & Astronomy Research Council.
3The European VLBI Network is a joint facility of European, Chinese and other radio astronomy
institutes funded by their national research councils.
149

150 Chapter 6. EVN and MERLIN observations
Table 6.1: Observing scheme for target sources and calibrators (EVN and MERLIN).
1XRS name Phase-reference calibrators Cycling scheme (a) Obs. Time
[s] [s]
J001442.2+580201 J0007+5706 (b) 300/150 7200
J013106.4+612035 J0147+5840 2×(300/150)/300/120 8000
4C 61.02 (c)
J042201.0+485610 (c) TXS 0422+496 300/150/300/120/300/150 7500
NRAO 150
J062148.1+174736 J0630+1738 (b) 300/150 6200
J072259.5−073131 1RXS J072418.3−071508 2×(270/180)/240/180 6100
J072418.3−071508 1RXS J072259.5−073131 2×(180/270)/180/240 4700
(a) target/calibrator/2nd calibrator.
(b) Used in the VLBI fringe fitting process to guide the target source fringe detection (see text).
(c) No fringes found to the radio source.
6.2.1 MERLIN
MERLIN is a connected radio interferometer across England, with baselines reaching
up to 217 km length. This array observed with 2-bit sampling at dual polarization
with two blocks of 16 channels, each channel of 1 MHz bandwidth. We analyzed
the left hand circular polarization data excluding one channel at both edges of the
band, yielding a final bandwidth of 14 MHz. The correlator integration time was of
4 s. Some antennas did not observe during short periods due to strong winds.
Since some of the target radio sources were faint, we scheduled the observations
introducing phase-reference calibrators with cycle times of around 7 min (compatible
with the expected coherence times). The observing scheme is described in Table 6.1,
where we quote the total on-source time for the target sources in the last column.
We also observed the fringe-finder DA 193 and the MERLIN flux density calibrator
3C 286.
The MERLIN data reduction was carried out at Jodrell Bank Observatory,
using standard procedures within the aips software package. We did not detect
1RXS J042201.0+485610. All other sources were detected, and accurate positions
for the target radio sources were obtained via phase-referencing. These positions,
presented in Table 6.2, were used later as a priori information for the VLBI cor-

6.2. Observations and data reduction 151
Table 6.2: MERLIN positions for the five detected target sources, obtained via phase-
referencing. The uncertainties quoted for the target sources were provided by the jmfit
task in aips (measuring in the phase-referenced image) and do not include systematic
errors. The positions of all phase-reference calibrators except 1RXS J072418.3−071508
have an accuracy better than 1 mas.
Target sources Phase-reference calibrators
1RXS name α (J2000.0) δ (J2000.0) Source name α (J2000.0) δ (J2000.0)
J001442.2+580201 00 h 14 m 42. s 12822 +58 ◦ 02 ′ 01. ′′ 2460 J0007+5706 (a) 00 h 07 m 48. s 47110 +57 ◦ 06 ′ 10. ′′ 4540
±0. s 00013 ±0. ′′ 0022
J013106.4+612035 01 h 31 m 07. s 23210 +61 ◦ 20 ′ 33. ′′ 3752 J0147+5840 (a) 01 h 47 m 46. s 54380 +58 ◦ 40 ′ 44. ′′ 9750
±0. s 00014 ±0. ′′ 0016
J062148.1+174736 06 h 21 m 47. s 75264 +17 ◦ 47 ′ 35. ′′ 0818 J0630+1738 (b) 06 h 30 m 07. s 25870 +17 ◦ 38 ′ 12. ′′ 9300
±0. s 00006 ±0. ′′ 0017
J072259.5−073131 07 h 22 m 59. s 68188 −07 ◦ 31 ′ 34. ′′ 8009 1RXS J072418.3 (c) 07 h 24 m 17. s 2912 −07 ◦ 15 ′ 20. ′′ 339
±0. s 00011 ±0. ′′ 0022 −071508
J072418.3−071508 (d) 07 h 24 m 17. s 2912 −07 ◦ 15 ′ 20. ′′ 339 J0730−116 07 h 30 m 19. s 1125 −11 ◦ 41 ′ 12. ′′ 601
±0. s 0007 ±0. ′′ 010
(a) Position from Patnaik et al. (1992).
(b) Position from Browne et al. (1998).
(c) This position was obtained from VLA observations (listed in the line below).
(d) The entries in this line correspond to VLA observations reported in Chapt. 5 (Table 5.1).
relation. The position given in Table 6.2 for the quasar 1RXS J072418.3−071508
was obtained from the VLA observations presented in Sect. 5.2. The position of
1RXS J072259.5−073131 is deduced from the phase-reference offset relative to the
source 1RXS J072418.3−071508 provided by MERLIN.
To image the sources, we averaged the data in frequency and exported them to
be processed into the difference mapping software difmap (Shepherd et al. 1994),
where we time-averaged the data in 32 s bins after careful editing.
6.2.2 EVN
The EVN observations were performed with the array described in Table 6.3, record-
ing in MkIV mode with 2-bit sampling at 256 Mbps with left hand circular polar-
ization. A bandwidth of 64 MHz was used, divided into 8 intermediate frequency
(IF) bands (v6cm-256-8-2-L mode). Some antennas did not observe during short
periods due to strong winds or snow.

152 Chapter 6. EVN and MERLIN observations
Table 6.3: EVN array used in the observations.
Antenna Code Location Diameter DPFU (a) Tsys (b)
[m] [K Jy −1 ] [K]
Effelsberg EB Germany 100 1.47 27
Jodrell Bank JB U.K. 25 0.11 40
Cambridge CM U.K. 32 0.21 38
Westerbork WB Netherlands 14×25 1.0 67
Medicina MC Italy 32 0.15 53
Noto NT Italy 32 0.16 47
Shanghai SH China 25 0.088 40
Toruń TR Poland 32 0.15 31
Onsala85 ON Sweden 25 0.043 45
(a) Degrees per flux unit. (b) Best value during the experiment.
The data were processed at the EVN MkIV correlator at the Joint Institute for
VLBI in Europe (JIVE), in Dwingeloo, The Netherlands. The correlator integration
time was of 4 s. A first post-processing analysis was also carried out at JIVE. The
data were processed using aips. A first a priori visibility amplitude calibration was
performed using antenna gains and system temperatures measured at each antenna.
The fringe fitting (fring) of the residual delays and fringe rates was performed
for all the radio sources. No fringes were found for 1RXS J042201.0+485610 and
4C 61.02. Fringes for many baselines were missing for 1RXS J001442.2+580201,
1RXS J062148.1+174736 and TXS 0422+496.
To improve the fringe detection on all baselines for 1RXS J001442.2+580201
and 1RXS J062148.1+174736 we used the delay, rate, and phase solutions from
their corresponding phase-reference calibrators (J0007+5706, 1 ◦ 18 ′ separation, and
J0630+1738, 1 ◦ 59 ′ separation, see Table 6.1) and interpolated them to the target
sources using the aips task clcal. We fringe-fitted the target sources again using
narrower search windows and obtained solutions for all baselines. A similar attempt
on 1RXS J042201.0+485610 (with respect to TXS 0422+486 and NRAO 150) was
unfruitful. Effelsberg was used as reference antenna throughout the aips data re-
duction process.
We then averaged the data in frequency and exported them to be imaged and
self-calibrated in difmap. The a priori visibility amplitude calibration was not suffi-

6.2. Observations and data reduction 153
cient to reliably image the weakest radio sources. We improved that by first imaging
in difmap the calibrator sources J0007+5706, J0147+5840, and J0630+1738, with
appropriate amplitude self-calibration. We deduced correction factors for each an-
tenna, these being consistent for the three radio sources within 2%. The factors
were: EB:0.98, JB:1.10, CM:0.78, WB:1.18, MC:0.87, NT:1.12, SH:1.03, TR:0.98
and ON:0.97. We corrected the amplitude calibration back in aips using the task
sncor (optype ’mula’) for all the radio sources and exported the data again into
difmap, where the final imaging was performed after editing and averaging of the
visibilities in 32 s blocks.
6.2.3 Combining EVN and MERLIN
The EVN and MERLIN arrays have one common baseline, between JB and CM,
which allows to combine both data sets and map them together. Combining data
sets from both arrays allows us to reach (u, v) resolution ranges from 0.04 Mλ (MK2-
Tabley) to 140 Mλ (NT-SH) at 5 GHz. We processed the data within aips in order
to combine both arrays. The B1950.0 (u, v) coordinates of the MERLIN data had
to be corrected to the ones of the EVN for the same reference system (J2000.0)
with uvfix. Then, the MERLIN data were self-calibrated with the EVN images
(see below), and the phase solutions were limited to the longest MERLIN baselines.
The MERLIN data were imaged and the peak-of-brightness of both data sets were
checked to be similar. As a next step, the EVN data were averaged in frequency
to correspond to the MERLIN data. The aips headers of both data sets were
modified conveniently to match together, and the weighting of both data sets was
also modified to be equal with wtmod. Finally, both data sets were concatenated
(using dbcon) and exported to difmap to be imaged with different data weighting
in (u, v) distance (tapering) after time averaging in 32 s bins.
6.2.4 Flux density measurements at the 100 m antenna in
Effelsberg
To complement the amplitude calibration and obtain additional information on the
radio sources, we interleaved cross-scans (in azimuth and elevation) with the 100 m
Effelsberg antenna to measure the radio source flux densities (A. Kraus, private

154 Chapter 6. EVN and MERLIN observations
Table 6.4: Flux densities and parameters used to produce the images in Figs. 6.1–6.5.
Single dish Array (a)
1RXS name SEB (taper FWHM) beam size P.A. Stot Speak Smin
[mJy] [Mλ] [mas] × [mas] [ ◦ ] [mJy] [mJy b. −1 ] [mJy b. −1 ]
J001442.2+580201 6.5 ± 0.5 (b) M 57 × 51 37 6.2 5.8 0.1
E+M (10) 7.7 × 7.2 −4 10.1 9.9 0.18
E 1.77 × 0.86 −20 11.5 7.0 0.15
J013106.4+612035 20.1 ± 0.6 M 71 × 39 −63 17.5 17.9 0.5
E+M (15) 7.8 × 6.2 −75 19.2 18.7 0.8
E 1.04 × 0.99 −1 17.6 11.6 0.4
J042201.0+485610 < 5 — — — — — —
J062148.1+174736 7.1 ± 0.7 (b) M 88 × 39 24 5.8 5.6 0.2
E+M (8) 13.1 × 11.1 −24 6.0 6.8 0.3
E 8.8 × 4.1 47 7.0 6.4 0.3
J072259.5−073131 67.9 ± 1.1 M 115 × 66 4 66.0 62.9 0.7
E+M (15) 9.5 × 7.3 −61 52.1 41.9 0.9
E 5.74 × 1.12 11 46.9 36.2 0.9
J072418.3−071508 282.2 ± 4.1 M 120 × 64 8 301.0 285.7 0.9
E+M (10) 11.7 × 9.8 −61 287.0 263.4 2.0
E 5.73 × 1.02 11 248.0 184.8 0.7
(a) M: MERLIN. E+M: EVN+MERLIN, FWHM of the tapering function (weighting of visibilities)
in parenthesis. E: EVN.
(b) Values with low SNR in the Gaussian fits.
communication). We fitted a Gaussian function to the flux-density response for
every cross-scan, and we averaged the different Gaussians. We linked the flux density
scale by observing primary calibrators such as 3C 286, 3C 48, or NGC 7027 (see e.g.
Kraus 1997; Peng et al. 2000). We list the single dish flux density measurements
in the second column of Table 6.4. The flux density values for the main VLBI
calibrators were of 7.5 ± 0.1 Jy for 3C 286, and 5.9 ± 0.2 Jy for DA 193.
6.3 Results and discussion
We present all the imaging results in Figs. 6.1–6.5, and the image parameters in
Table 6.4. The total flux density values for the different images diverge from each

6.3. Results and discussion 155
Figure 6.1: Images of 1RXS J001442.2+580201 using the different arrays and with the
parameters given in Table 6.4. The axes units are in mas. A (u, v)-tapering with a
FWHM at 10 Mλ has been used to perform the combined EVN+MERLIN image. In the
EVN image, N1, N2, S2 and S1 indicate the components discussed in the text.
other and from the single dish measurements, due to the amplitude self-calibration
process in all cases. Therefore, those values should be considered with care. The
minimum contours in the images are those listed as Smin in Table 6.4, while con-
secutive higher contours scale with 3 1/2 . Here follows a detailed discussion on each
source.
6.3.1 1RXS J001442.2+580201 and its two-sided jet
As can be seen in our images, shown in Fig. 6.1, this source appears point-like
at MERLIN resolution, partially resolved in the tapered EVN+MERLIN image
and clearly resolved at EVN scales. In this last case, it shows a two-sided jet-like
structure roughly in the north-south direction, with brighter components towards
the south. The trends are visible in the closure phases, giving us confidence that
the structure observed is not a consequence of sidelobes or imaging artifacts. In
the tapered EVN+MERLIN images, the structure extends up to 20–30 mas outside
of the core, and more clearly towards the north. This discrepancy could be due to
calibration problems.
Model fitting of the EVN visibilities with circular Gaussians provides a param-
eterization of the inner structure. Five components reproduce the visibilities. The
central one has 7.8 mJy, with a FWHM of 0.4 mas. Towards the north, one compo-
nent (N2) of 0.6 mJy at 3.6 mas (P.A. −10 ◦ , FWHM 0.7 mas) and another one (N1)
of 0.5 mJy at 8.6 mas (P.A. 2 ◦ , extended over 3 mas) are needed. Brighter compo-
N1
N2
S2
S1

156 Chapter 6. EVN and MERLIN observations
Table 6.5: Model fitted positions of the components in the 1RXS J001442.2+580201
radio jets and obtained jet parameters.
Comp. Distance P.A. βmin θmax
[mas] [ ◦ ] [ ◦ ]
N1 8.6 ± 0.3 2 0.23 ± 0.02 77 ± 1
N2 3.6 ± 0.2 −10 0.16 ± 0.02 81 ± 1
S2 5.0 ± 0.1 177 0.16 ± 0.02 81 ± 1
S1 13.7 ± 0.3 177 0.23 ± 0.02 77 ± 1
nents are present southwards, one (S2) of 1.1 mJy at 5.0 mas (P.A. 177 ◦ , FWHM of
0.8 mas) and the other one (S1) of 0.4 mJy at 13.7 mas (P.A. 177 ◦ , FWHM below
0.3 mas).
If we assume that components S1 and N1 correspond to a pair of plasma clouds
ejected at the same epoch near the compact object and perpendicularly to the
accretion disk, we can estimate some parameters of the jets by using the following
equation:
β cos θ = µa − µr
µa + µr
= da − dr
da + dr
, (6.1)
β being the velocity of the clouds in units of the speed of light, θ the angle between
the direction of motion of the ejecta and the line of sight and µa and µr the proper
motions of the approaching and receding components, respectively (see Sect. 1.2.1).
Although we do not know the epoch of ejection of the clouds, we can cancel the
time variable by using the relative distances to the core da and dr, as expressed in
Eq. 6.1. Since both variables, β and cos θ, take values between 0 and 1, it is clear
that knowing β cos θ allows us to compute a lower limit for the velocity (βmin) and
an upper limit for the angle (θmax). The same applies for the S2 and N2 components.
In Table 6.5 we list the positions of the components obtained from model fitting,
together with the derived values from βmin and θmax for each one of the pairs. The
slightly different results obtained using pair 1 or 2, could be due to the fact that
the position for the S1 component obtained with model fitting happens to be at
the lower part of this elongated component, hence increasing β cos θ, or to intrinsic
different velocities for each one of the pairs. Hereafter we will use β > 0.20 ± 0.02
and θ < 78 ± 1 ◦ .
A similar approach to obtain the jet parameters of the source can be performed

6.3. Results and discussion 157
thanks to the brightness asymmetry of the components using the following equation
(see Sect. 1.2.1):
β cos θ = (Sa/Sr) 1/(k−α) − 1
(Sa/Sr) 1/(k−α) + 1
, (6.2)
where Sa and Sr are the flux densities of the approaching and receding components,
respectively, k equals 2 for a continuous jet and 3 for discrete condensations, and α
is the spectral index of the emission (Sν ∝ ν +α ). However, the equation above is
only valid when the components are at the same distance from the core. If this is
not the case (i.e., dr < da), the ratio Sa/Sr will be lower than the one that should
be used in Eq. 6.2 (because the flux density decreases with increasing distance from
the core). In consequence, Eq. 6.2 only allows us to obtain a lower limit for β cos θ.
In order to use this approach we will consider k = 3, because the components
seem to be discrete condensations, and α = −0.20 ± 0.05, according to the overall
spectral index reported in Table 5.1, since we do not have spectral index information
of the components. The use of the flux densities obtained after model fitting gives
β cos θ > 0.09 ± 0.04 for the S2–N2 pair and β cos θ > −0.07 ± 0.08 for the S1–N1
pair. This last value is certainly surprising, although it can be explained by the
fact that the S1 flux density obtained after model fitting does not account for the
total flux of this plasma cloud. In fact, better estimates of the flux densities can be
obtained summing together the flux densities of the clean components obtained
within each one of the four plasma clouds. This yields to β cos θ > 0.13 ± 0.05 for
the S2–N2 pair and β cos θ > 0.17±0.02 for the S1–N1 pair, in good agreement with
the values computed using the distances from the components to the core, shown in
Table 6.5.
If we compare the VLA position reported in Table 5.1 with the MERLIN position
in Table 6.2 we can see that they are different. In fact, taking into account the errors,
the MERLIN−VLA position offsets can be expressed as: ∆α cos δ = 16 ± 10 mas
and ∆δ = 27 ± 10 mas. Assuming that the difference in position is due to intrinsic
proper motions and considering the time span between both observations, 224 days,
we obtain: µα cos δ = 26 ± 16 mas yr −1 and µδ = 44 ± 16 mas yr −1 . Although a
proper motion of the source would clearly indicate its microquasar nature, because
no proper motion would be detected in an extragalactic source, we must be cautious
with this result. First of all, the phase-reference sources where different in the
VLA and MERLIN observations, as well as the observing frequencies from which
positions were estimated. On the other hand, none of these results exceeds the 3σ

158 Chapter 6. EVN and MERLIN observations
Figure 6.2: Images of 1RXS J013106.4+612035 using the different arrays and with the
parameters given in Table 6.4. The axes units are in mas. A (u, v)-tapering with a
FWHM at 15 Mλ has been used to perform the combined EVN+MERLIN image.
value, preventing to state that a proper motion has been detected.
Summarizing, our results indicate that this source exhibits relativistic radio jets
with β > 0.20 and, therefore, together with the results reported in Chapt. 5, we
consider 1RXS J001442.2+580201 as a very promising microquasar candidate.
6.3.2 1RXS J013106.4+612035 and its one-sided jet
We show in Fig. 6.2 the images obtained after our observations. Although the source
appears compact at MERLIN and EVN+MERLIN scales, there is a weak one-sided
radio jet towards the northwest in the EVN image. Model fitting with circular
Gaussian components can reproduce the observed visibilities as follows: a central
15.4 mJy component with 0.64 mas FWHM, and a northwest 2.1 mJy component
with a FWHM of 0.74 mas, located at 1.8 mas in P.A. −73 ◦ . As can be seen, this
last component is located at a distance of ∼ 2 times the beam size from the core.
Using the fact that we do not detect a counter-jet, we can use Eq. 6.2 replacing
Sr with the 3σ level value. This, of course, will only provide a lower limit to β cos θ,
expressed as follows:
β cos θ > (Sa/3σ) 1/(k−α) − 1
(Sa/3σ) 1/(k−α) + 1
. (6.3)
Using Sa = 2.1 mJy, 3σ = 0.30 mJy (the 1σ value has been taken as the root mean
square noise in the image), α = −0.05 ± 0.05 (see Table 5.1), and k = 3 to be
consistent with the lowest limit, we obtain β cos θ > 0.31 ± 0.05 (β > 0.31 ± 0.05

6.3. Results and discussion 159
and θ < 72 ± 3). Hence, a lower limit of β ≥ 0.3 is obtained, pointing towards
relativistic radio jets as the origin of the elongated radio emission present in the
EVN image. Although the one-sided jet morphology at mas scales is found mostly
in extragalactic sources, it is also present in some galactic REXBs, like Cyg X-3
(Mioduszewski et al. 2001) or LS I +61 303 (Massi et al. 2001). Hence, we cannot
rule out a possible galactic nature on the basis of the detected morphology.
As done for the previous source, we can compare the VLA position with the
MERLIN one, and find that they differ in ∆α cos δ = 39 ± 10 mas and ∆δ =
−0.8 ± 10 mas. If the offsets are real, this would imply µα cos δ = 64 ± 16 mas yr −1
and µδ = −1 ± 16 mas yr −1 . Hence, it seems possible that we have detected a
proper motion in right ascension at a 4σ level. However, we must be cautious since,
as in the previous case, the phase-reference sources and observing frequencies were
different in each observation.
Overall, these results are indicative of relativistic radio jets, and hence, together
with the results reported in Chapt. 5, allow us to classify this source as a promising
microquasar candidate.
6.3.3 1RXS J042201.0+485610, a non-detected source
This radio source was marginally seen in the cross-scans carried out in Effelsberg. In
fact, this is compatible with the low flux density of 2.3±0.4 mJy at 1.4 GHz listed in
the NVSS. The source was not detected with MERLIN or with the EVN, mainly due
to problems with the phase-reference sources, as pointed out in Sect. 6.2.2. In fact,
as discussed in Chapt. 5, VLA A configuration observations showed a flux density of
0.4 mJy at 5 GHz and an inverted spectrum with a spectral index up to α = +1.6,
suggesting thermal radio-emission resolved at higher angular resolutions. However,
we cannot exclude the possibility of having a highly variable source.
6.3.4 1RXS J062148.1+174736, a compact source
As can be seen in our images, shown in Fig. 6.3, the radio source is compact on all
scales. In fact, model fitting of the EVN visibilities converges to a point-like radio
source (the FWHM of a circular Gaussian tends to zero). We must note that when

160 Chapter 6. EVN and MERLIN observations
Figure 6.3: Images of 1RXS J062148.1+174736 using the different arrays and with the
parameters given in Table 6.4. The axes units are in mas. A (u, v)-tapering with a
FWHM at 8 Mλ has been used to perform the combined EVN+MERLIN image.
this source was observed it was below the horizon in SH. Therefore, the obtained
beam size for the EVN image is larger than the ones obtained for the other sources,
as can be seen in Table 6.4, hence providing lower angular resolution than in the
other cases. A comparison between the VLA and MERLIN positions reveals that
they are perfectly compatible within the errors, suggesting an extragalactic nature
or a small proper motion if it turns out to be a galactic microquasar. Although the
compactness of the source is not indicative of a galactic or an extragalactic origin,
the extended optical counterpart, reported in Chapt. 5, suggests an extragalactic
origin for this source. In fact, the detected radio variability from 8 to 11 mJy within
a week reported in Chapt. 5, is compatible with the compactness in an extragalactic
object (see IDV phenomenon, Wagner & Witzel 1995).
6.3.5 1RXS J072259.5−073131 and its bent one-sided jet
The images obtained at different resolutions are plotted in Fig. 6.4. The MERLIN
image presents a compact structure with some elongation eastwards, while the EVN
and combined EVN+MERLIN images show a clear one-sided jet towards the east,
with a slight bent towards the south at larger core separations. The closure phases
clearly show that the source departs from symmetry, with preferred emission to the
east, both in the MERLIN and the EVN data sets. Two distinct components are
present in the EVN image, at 9 and 17 mas from the compact core (P.A. of 89 and
113 ◦ , respectively). Those components can be model fitted with elliptical Gaussians,
yielding flux densities of 3.7 and 2.8 mJy, respectively, for a core of 40.3 mJy.

6.3. Results and discussion 161
Figure 6.4: Images of 1RXS J072259.5−073131 using the different arrays and with the
parameters given in Table 6.4. The axes units are in mas. A (u, v)-tapering with a
FWHM at 15 Mλ has been used to perform the combined EVN+MERLIN image.
Using again the fact that we do not detect a counter-jet, we can use Eq. 6.3 with
Sa = 3.7 mJy (the closest component to the core in the EVN image), 3σ = 0.54 mJy
(as previously done, the 1σ value has been taken as the root mean square noise
in the image), α = −0.25 ± 0.2 (see Table 5.1), and k = 3 to be consistent with
the discrete nature of the components, to obtain β cos θ > 0.29 ± 0.05, and hence
β > 0.29 ± 0.05 and θ < 73 ± 3 ◦ . Therefore, these results point towards relativistic
radio jets as the origin of the elongated radio emission present in the images.
As pointed out for 1RXS J013106.4+612035, the one-sided jet morphology does
not rule out a microquasar nature for this object. However, the bending of the jet
at such small angular scales resembles the ones seen in blazars. In fact, as reported
in Chapt. 5, a one-sided arcsecond scale jet is also present in VLA A configuration
observations at 1.4 GHz, an unusual feature in the already known microquasars.
A comparison between the VLA and MERLIN positions reveals that they agree
within the errors, suggesting an extragalactic nature or a small proper motion if it
turns out to be a galactic microquasar.
Overall, these results are indicative of relativistic radio jets, although this source
shows characteristics more similar to blazars than to microquasars.

162 Chapter 6. EVN and MERLIN observations
Figure 6.5: Images of the already identified quasar 1RXS J072418.3−071508 using the
different arrays and with the parameters given in Table 6.4. The axes units are in mas.
A (u, v)-tapering with a FWHM at 10 Mλ has been used to perform the combined
EVN+MERLIN image.
6.3.6 1RXS J072418.3−071508, a quasar with a bent one-
sided jet
This radio source has recently been (March 2002) classified as a quasar in the SIM-
BAD database, and it is not any more a microquasar candidate. It is catalogued in
the Parkes-MIT-NRAO (Griffith et al. 1994) and the Texas Survey (Douglas et al.
1996). It is listed as PMN J0724−0715 in the NED database, and is the source
WGA J0724.3−0715 in Perlman et al. (1998), who reported a faint and quite broad
Hα emission line (rest-frame Wλ = 30.3 ˚A, FWHM=4000 km s −1 ), and classified
it as a Flat Spectrum Radio Quasar (FSRQ) with z = 0.270. Nevertheless, we
have reported here our observational results for this source, since it was a candidate
when we performed the observations. It presents (Fig. 6.5) a one-sided pc-scale jet
oriented towards the northeast, changing from a P.A. of ∼ 50◦ at 5 mas from the
core (EVN image) to 20◦ up to 200 mas, at MERLIN scales.
Model fitting of the EVN visibilities with circular Gaussians reveals a compact
core (0.7 mas FWHM) with 245.4 mJy, and two distinct components, one with
11.9 mJy at 6.2 mas (P.A. 46◦ , 1.9 mas FWHM, present in the right panel image of
Fig. 6.5) and another one with the size of the beam, a flux density of 1.7 mJy at a
distance of 29.6 mas in P.A. 27◦ (visible in the middle panel image of Fig. 6.5).
Using again the fact that we do not detect a counter-jet, we can use Eq. 6.3
with Sa = 11.9 mJy (the closest component to the core), 3σ = 0.45 mJy, α =

6.4. Conclusions 163
0.07 ± 0.02 (see Table 5.1), and k = 3 to be consistent with the discrete nature of
the components, to obtain β cos θ > 0.51±0.04. Hence, an upper limit of θ < 60±3 ◦
and a lower limit of β > 0.51 ± 0.04 is obtained, pointing towards relativistic radio
jets as the origin of the elongated radio emission present in the images of this already
identified quasar.
6.4 Conclusions
We have presented EVN and MERLIN observations of the six sources studied in
Chapt. 5 in a search for microquasar candidates at low galactic latitudes. The
first one, namely 1RXS J001442.2+580201, displays a two-sided radio jet, which
after analysis implies β > 0.20 ± 0.02 and θ < 78 ± 1 ◦ . 1RXS J013106.4+612035,
displays a one-sided radio jet, requiring β > 0.31 ± 0.05 and θ < 72 ± 3 ◦ . The third
one, namely 1RXS J042201.0+485610, was not detected due to its low flux density
and/or to phase-referencing problems. 1RXS J062148.1+174736 appeared compact
at all scales. The fifth one, namely 1RXS J072259.5−073131, displays a bent one-
sided radio jet, implying β > 0.29 ± 0.05 and θ < 73 ± 3 ◦ . Finally, the already
known quasar 1RXS J072418.3−071508 shows also a bent one-sided jet, requiring
β > 0.51 ± 0.04 and θ < 60 ± 3 ◦ .
We summarize our obtained results for these sources in the next chapter.

164 Chapter 6. EVN and MERLIN observations

Bibliography
Browne, I. W. A., Wilkinson, P. N., Patnaik, A. R., & Wrobel, J. M. 1998, MNRAS,
293, 257
Douglas, J. N., Bash, F. N, Bozyan, F. A, Torrence, G. W., & Wolfe, C. 1996, AJ,
111, 1945
Griffith, M. R., Wright, A. E., Burke, B. F., & Ekers, R. D. 1994, ApJS, 90, 179
Kraus, A. 1997, PhD Thesis, Friedrich-Wilhelms-Universität Bonn, Germany
Massi, M., Ribó, M., Paredes, J. M., Peracaula, M., & Estalella, R. 2001, A&A,
376, 217
Mioduszewski, A. J., Rupen, M. P., Hjellming, R. M., Pooley, G. G., & Waltman,
E. B. 2001, ApJ, 553, 766
Paredes, J. M., Ribó, M., & Martí, J. 2002, A&A, 394, 193
Patnaik, A. R., Browne, I. W. A., Wilkinson, P. N., & Wrobel, J. M. 1992, MNRAS,
254, 655
Peng, B., Kraus, A., Krichbaum, T. P., & Witzel, A. 2000, A&A, 145, 1
Perlman, E. S., Padovani, P., Giommi, P., et al. 1998, AJ, 115, 1253
Shepherd, M. C., Pearson, T. J., & Taylor, G. B. 1994, BAAS, 26, 987
Wagner, S. J., & Witzel, A. 1995, ARA&A, 33, 163
165

166 BIBLIOGRAPHY

Chapter 7
Summary
After a detailed analysis of our data, we show in Table 7.1 a summary of the results
obtained after the VLA and optical observations presented in Chapt. 5, and the
EVN+MERLIN observations reported in Chapt. 6.
As can be seen, 1RXS J001442.2+580201 and 1RXS J013106.4+612035 are
promising microquasar candidates. 1RXS J042201.0+485610 is probably of thermal
nature due to its highly inverted spectrum at high radio frequencies, and we also note
its extended appearance at optical wavelengths. 1RXS J062148.1+174736 is proba-
bly extragalactic due to the extended optical counterpart. 1RXS J072259.5−073131
displays a bent one-sided jet, which is a common property in blazars. Finally, the
source 1RXS J072418.3−071508 is an already identified quasar.
We note that none of the 6 sources studied here appears in the Third EGRET
Catalog of high-energy γ-ray sources.
We point out that 1RXS J072259.5−073131 is bright enough at radio wave-
lengths to attempt an H i absorption experiment, that could allow to determine if
this source is galactic or not. In any case, optical spectroscopic observations of the
first five sources are in progress for autumn 2002, to clearly unveil their galactic or
extragalactic nature.
Finally, the two sources belonging to Group 1 that have not been observed yet,
namely 1RXS J181119.4−275939 and 1RXS J185002.8−075833 will be eventually
studied in the future, together with Group 2 sources.
167

¡¢ ¢
¢
¡¢
£¤
¡¢
¤ ¢ ¢
¡¢
¡¢
¤
¤
¤ £
¡¢
¤
£¦
£¤ ¤ £
¥ ¥
§ ¢
§ ¢
¦
¦ ¥
¢
¤
¢
§
¢
¨
¤
©
©
¢
¨
¤
¢
¦
¤
£
¢
¥
¢
§
¨
©
©
¢
¨
¤
£¦
¥ ¢
£
¥ ¢
¦
©
¨
¢
§
¢
¢
¨
¨
¨
¤
©
©
¥
¦
¤
©
©
¢
¢
¨
¨
¤
¥
¢
Table 7.1: Summary of the obtained results after the VLA and optical observations
presented in Chapt. 5, and the EVN and MERLIN observations reported in Chapt. 6.
An asterisk indicates a non-expected behavior for microquasars.
168 Chapter 7. Summary

Part III
General conclusions
169

General conclusions 171
Since each chapter contains its own conclusions, here we will briefly list our
general conclusions after this work.
Part I:
1. We have discovered the microquasar LS 5039 after a careful examination of
modern archive databases and follow-up interferometric radio observations.
Therefore, we conclude that similar studies may reveal a previously unnoticed
population of silent microquasars like LS 5039. If this is correct, the micro-
quasar phenomenon may not be as rare as it seems.
2. We have performed an in-depth study of the LS 5039 multiwavelength be-
havior, covering from radio wavelengths to high-energy γ-rays. In doing so,
we have carried out observations aimed to study the flux, flux variability and
spectrum at radio, optical and X-ray wavelengths. In the radio domain we
have also obtained interferometric images in the continuum and single dish
spectral line maps. All these data, together with γ-ray observations from the
literature, have allowed us to propose a scenario and a model to explain the
observed phenomenology of LS 5039. Multiwavelength observations are clue
to understand accretion/ejection processes near compact objects.
3. We have proposed an association between LS 5039 and one of the ∼ 170
unidentified EGRET sources, which seems plausible due to the persistent and
moderately variable emission of the source at both radio and γ-ray wave-
lengths. If confirmed, this would be the first association between a microquasar
and an EGRET source, and we suggest that other unidentified high-energy γ-
ray sources could be silent microquasars waiting to be discovered.
4. We have discovered that LS 5039 is a runaway microquasar with vsys 150
km s −1 , escaping from its own regional standard of rest with a large velocity
component perpendicular to the galactic plane. We are able to explain both,
the high space velocity and the high eccentricity observed, in a symmetric SN
explosion scenario with a mass loss of ∆ M ∼ 17 M⊙.
5. The high space velocity and possible lifetime of this microquasar indicate that
it could reach a galactic latitude of b = −12 ◦ . Therefore, if the proposed asso-
ciation between LS 5039 and the EGRET source 3EG J1824−1514 is correct,
we could be able to detect γ-ray microquasars up to values of |b| 10 ◦ . In

172 General conclusions
particular, runaway microquasars could be connected with some of the uniden-
tified faint, variable, and soft γ-ray EGRET sources above/below the galactic
plane.
Part II:
1. We have presented a cross-identification method to search for REXBs, which
are potential microquasar sources. The obtained results give confidence to
the proposed method, since the output list of objects included all but one
known persistent HMXB microquasars within |b| < 5 ◦ . The obtained list
after removing previously known sources contains 13 new radio emitting X-
ray sources, of which 8 have been classified as high priority objects.
2. We have studied 6 of these 8 sources, and found optical counterparts to all of
them. We have obtained accurate positions at both radio and optical wave-
lengths, perfectly compatible between them. We also have obtained radio
spectra and optical magnitudes of the sources.
3. We have presented EVN and MERLIN observations of these six sources. Five
of the six objects have been detected and imaged, presenting different mor-
phologies: one source has a two-sided jet, three sources have one-sided jets,
and one source is compact.
4. After our observations we conclude that the sources 1RXS J001442.2+580201
and 1RXS J013106.4+612035 are promising microquasar candidates in the
Galaxy. 1RXS J042201.0+485610 is probably of thermal nature due to its
highly inverted spectrum at high radio frequencies, and we note its extended
appearance at optical wavelengths. 1RXS J062148.1+174736 is probably ex-
tragalactic due to the extended optical counterpart. 1RXS J072259.5−073131
displays a bent one-sided jet, which is a common property in blazars. Finally,
the source 1RXS J072418.3−071508 is an already identified quasar.
5. Work in progress. Optical spectroscopic observations of the first five sources
are in progress for autumn 2002, to clearly unveil their galactic or extragalactic
nature. On the other hand, the remaining sources in the list will be eventually
studied in the future.