Optical observations of close binary systems with a compact component - Thesis

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Optical observations of close binary systems with a compact component

Augusteijn, T.

Publication date

1994

Document Version

Final published version

Link to publication

Citation for published version (APA):

Augusteijn, T. (1994). Optical observations of close binary systems with a compact

component.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)

and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open

content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please

let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material

inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter

to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You

will be contacted as soon as possible.

Opticall observations of close binary

systemss with a compact component

CD D CO O O O

sz sz

0.55

-0 -0

Pi i

lnJI I

II 1 il

. . .. if lili, , , 1

6 5 5 00 6 5 7 5

Wavelengthh (A)

Thomass Augusteijn

Opticall observations of close binary systems with

aa compact component

Optischee waarnemingen van nauwe dubbelsterren

mett een compacte component

(mett een samenvatting in het Nederlands)

Academischh Proefschrift

terr verkrijging van de graad van doctor aan de Universiteit vann Amsterdam, op gezag van de Rector Magnificus prof. dr. P.. W. M. de Meijer, ten overstaan van een door het college van dekanenn ingestelde commissie in het openbaar te verdedigen in dee Aula der Universiteit (Oude Lutherse Kerk, ingang Singel 411,, hoek Spui), op dinsdag 29 november 1994 te 11.30 uur

door r

Thomass Augusteijn

geborenn te Amsterdam

promotiecommissie: :

promotor:: prof. dr. J. van Paradijs

overigee leden: prof. dr. E. P. J. van den Heuvel

prof.. dr. K. Home

prof.. dr. M. van der Klis

dr.. G.J. Savonije

dr.. H. C. Spruit

prof.. dr. P.S. The

prof.. dr. F.W. Verbunt

CoverCover illustration: grey-scale plot of the Ha emission line in the spectrum of the dwarf

novaa V485 Centauri as a function of phase at the 59 min orbital period (cf. Chapter 9).

Contents s

11 Introduction and summary 5 1.11 Close binaries containing a compact object 6

1.1.11 Orbital period distribution 6

1.1.22 Mass transfer 7 1.1.33 Angular-momentum losses 8

1.1.44 The period gap 9 1.1.55 The minimum period 10 1.22 Low-mads X-ray binaries 10

1.2.11 Classification H 1.2.22 Formation and evolution 12

1.2.33 Summary of results 13 1.33 Cataclysmic variables 14

1.3.11 Classification 14 1.3.22 Formation and evolution 15

1.3.33 Summary of results 16

11 L o w - m a s s X - r a y binaries 21 22 The optical counterpart of the Z Bource G X 3 4 9 + 2 23

2.11 Introduction 23 2.22 Observations and Analysis 23

2.33 Discussion 24 2.44 Conclusion 26 33 Coordinated X-ray and Optical observations of Sco X - l 29

3.11 Introduction 29 3.22 Observations 30 3.33 Results 31 3.44 Historical Walraven data 33

3.55 Conclusions 35 44 Phase-resolved spectroscopy of the atoll sources 1 6 3 6 - 5 3 6 / V 8 0 1 Ara and

1 7 3 5 - 4 4 / V 9 2 66 Sco 39 4.11 Introduction 39 4.22 Photometry 40 4.2.11 1636-536/V801 Ara 40 4.2.22 1735-444/V926 Sco 42 4.33 Spectroscopy 43 4.3.11 Average spectrum 43 1 1

2 2 Contents Contents

4.3.22 Radial-velocity variations 44

4.44 Discussion 4g 4.4.11 Emission from the bulge 48

4.4.22 A comparison of the two sources 49

I II C a t a c l y s m i c Variables 53 55 Spin-up of the white dwarf in the intermediate polar B G C M Ï / 3 A 0 7 2 9 + 1 0 3 55

5.11 Introduction 55 5.22 Observations and Reduction 56

5.33 The White Dwarf Spin Ephemeris 57 5.3.11 Deriving the ephemeris 57 5.3.22 Comparison with previous work 58

5.44 The Orbital Ephemeris 61

5.55 Discussion 62 5.5.11 Accretion models 63

5.5.22 Comparison with the observed values of P and fi 64

5.66 Conclusions 66 66 Periodicities in the optical brightness variations of the intermediate polar T V

Columbaee gg 6.11 Introduction 69

6.22 Observations and reduction 71

6.33 Results 71 6.3.11 The orbital period 71

6.3.22 The 5.2 hr photometric period 74 6.3.33 The 4 day photometric period 79

6.44 The 1911 sec X-ray period 80

6.55 Discussion 82 6.5.11 Long term brightness changes 82

6.5.22 Variations at the 4 day period 83

6.5.33 The outbursts 87

6.66 Summary 88 77 T i m e resolved spectroscopy of the dwarf nova V Y Aquarii in superoutburst

andd quiescence 91 7.11 Introduction 91 7.22 Observations 92 7.33 The outburst spectra 92

7.3.11 Brightness variations 92 7.3.22 Radial velocity variations from Gaussian fits 95

7.3.33 Radial velocity variations from cross correlations 95

7.3.44 Variations of the system velocity 97 7.3.55 The emission component in the broad absorption lines 98

7.44 The quiescence spectra 98 7.4.11 Absorption components 98

7.4.22 Emission lines 99 7.55 The mass ratio and inclination 104

ContentsContents 3

88 A 59m photometric period in the dwarf nova V485 Cen 111

8.11 Introduction I l l 8.22 Observations and Analysis I l l

8.2.11 Photometry I l l 8.2.22 Spectroscopy 114

8.33 Discussion 115 8.3.11 A 1 hour rotation period of the white dwarf? 115

8.3.22 A 1 hour orbital period? 116 99 V485 Centauri: a dwarf nova with a 59m orbital period 119

9.11 Introduction 119 9.22 Photometry 120 9-33 Spectroscopy 123

9.3.11 Observations and data reduction 123 9-3.22 Radial-velocity variations 124 9.3.33 The spectral distribution 130

9.44 System parameters 132

9.55 Discussion 133 9.5.11 Evolutionary considerations 133

9.5.22 The mass transfer rate 134 9.5.33 The secondary 135 100 Outline of a comparative study of disk and halo cataclysmic variables 139

10.11 Introduction 139 10.22 Halo and disk CVs 140 10.33 Discussion of the selection criteria 140

10.44 Our sample 151 10.55 High Galactic Latitude CVs revisited 151

10.66 Large-amplitude dwarf novae 153

Samenvattingg 155 Dankwoordd 159

1 1

Introductionn and summary

Mostt stars in t h e Galaxy are p a r t of a binary or a multiple system (Abt 1983). Generally, the differentt components are far a p a r t , a n d by a n d large they evolve like single s t a r s . However, m a n yy binary stars - a n d often not t h e least interesting ones - are sufficiently close together t h a t t h ee evolution of its constituent components is influenced by t h e presence of a neighbour. In the studyy of such binary systems one often uses t h e Roche a p p r o x i m a t i o n , i.e., it is assumed t h a t the g r a v i t a t i o n a ll fields of the two stars are like those of two point masses, t h a t t h e orbit is circular, a n dd t h a t t h e stars r o t a t e synchronously w i t h t h e orbit (co-rotation). Since stars tend t o be centrallyy concentrated, this a p p r o x i m a t i o n can generally be expected to be good. In this Roche a p p r o x i m a t i o nn the shape of a star is given by a (Roche) equipotential surface (see Fig. 1.1). T h ee Roche lobe is t h e critical equipotential surface which passes t h r o u g h t h e first Lagrangean pointt i i a n d encompasses b o t h s t a r s . Deep inside t h e Roche lobe the equipotential surfaces aree practically spherical, b u t close t o t h e Roche lobe they are strongly distorted a n d almost pear-shaped.. If a star fills its Roche lobe, m a t t e r can flow t h r o u g h Li towards t h e companion star. .

F i g u r ee 1.1. Sections in the orbital planee of Roche equipotential sur-faces,, for a binary system with mass ratioo q = Mi/M2 = 1/0.3 (Marinus

1994).. The Lagrangean points are indicated.. The plus sign indicates thee center of mass of the system

Binaryy systems in which b o t h stars are within t h e Roche lobe are called detached. If one of thee stars fills its p a r t of t h e Roche lobe t h e system is called semi-detached, a n d when b o t h p a r t s off t h e Roche lobe are filled t h e s y s t e m is called a contact binary. Detached systems include alll visual binaries, non-evolved spectroscopic and eclipsing binaries. These l a t t e r systems are especiallyy i m p o r t a n t as they provide t h e most accurate values of stellar p a r a m e t e r s such as

6 6

11 Introduction and summary theirr masses and radii (see, e.g., Popper 1980). Also binary pulsars, binary white dwarfs, pre-cataclysmicc variables and some of the Algol systems are detached systems. Contact systems includee W Ursa Majoris systems and 0 Lyrae systems, but also binary supergiants which can havee orbital periods of 1 year or longer. Semi-detached systems include most of the Algol systems,, £ Aurigae systems, symbiotic stars, cataclysmic variables and X-ray binaries. Optical studiess of the latter two types of systems are the subject of this thesis.

1.11 Close binaries containing a compact object

Ann especially interesting class of semi-detached binary systems is formed by the close binaries withh a compact component (a white dwarf, a neutron star, or a black hole). In these systems thee compact object is often referred to as the primary, and the Roche-lobe filling companion as thee secondary or donor star. Matter from the secondary slightly outside its Roche lobe near the innerr Lagrangean point (Lj) wiU flow to the compact star. Because of the angular momentum of thee overflowing matter it will not fall straight onto the compact object (Lubow and Shu 1975), butt will spread out to form a ring around the compact object. Due to viscous processes this' ringg will form an accretion disk, through which matter slowly spirals inward in nearly Keplerian orbits.. At the point of intersection of the accretion stream and the disk a shock is formed. This producess a bulge (or "hot spot") at the disk rim.

Semi-detachedd binaries containing a neutron star or a black-hole are X-ray binaries. Because off the small radius of a neutron star, or a stellar-mass black-hole (R ~ 10 km), accretion of thee transferred matter will produce X rays. X-ray binaries can be divided into two groups (see, e.g.,, Van Paradijs 1983): high-mass X-ray binaries (HMXBs), with a massive (> 10 M0)

early-typee companion, and low-mass X-ray binaries (LMXBs) with a low mass (< 1 M0) late-type

companion.. Semi-detached close binaries consisting of a low-mass donor and a white dwarf are cataclysmicc variables (CVs). The radiation released by accretion of matter from the companion ontoo a white dwarf (R ~ 10 000 km), wiU be emitted mainly in the ultra-violet.

1.1.11 Orbital p e r i o d distribution

Thee only well-known property for a large number of LMXBs and CVs is their orbital period. Inn Fig. 1.2 we present their orbital period distributions, as derived from Ritter and Kolb (1994) andd Van Paradijs (1994). It can be seen that the range in periods covered by these two types off system are very similar, but that the distributions are different. Two striking features in the periodd distribution of CVs are a significant lack of systems in the period range ~ 2 - 3 hr (the "periodd gap"), and a sharpe cut-off in the distribution at ~80 min. There is a clear lack of LMXBss with periods below ~ 3 hr.

Thee orbital period does not only constrain the system dimensions, but it also tells us the meann density of the Roche-lobe filling secondary. Paczynski (1971) showed that the radius of a sphericall star having the same volume as that of the Roche lobe can be approximated by

RL2RL2 / I N 1 /3

~??

=

°-

462

\T+V

for

* ~

2

' U-D

wheree a is the orbital separation and q = MxjM2, with the subscripts 1 and 2 referring to the

primaryy and secondary, respectively. Using Kepler's third law

==

(GM

(l + q)P*\

1.11.1 Close binaries containing a compact object 7 7 40 0 30 0 N2 0 0 10 0 0 0 8 8 6 6 NN 4 2 2 0 0 _ ll 1 1 1 I 1

f f

J J

HS S

n n

7 7

F i g u r ee 1.2. The orbital period distributionn of CVs (top), as de-rivedd from Rittei and Kolb (1994), andd LMXBs (bottom), as derived fromm Van Paradijs (1994)

wheree P is t h e orbital period, one finds for t h e m e a n density of t h e Roche-lobe filling (R2 = RL)

secondary y P2 2 3 M2 2 4 i r ü f f H II , 3 7 2 - 5 / c m m (1.3) )

wheree P is in hours. If we know t h e mass-radius for the secondary, e.g. if it is a main-sequence star,, t h e n it follows from Eq. (1.3) t h a t the m a s s (and radius) of t h e Roche-lobe filling secondary iss defined by t h e period alone.

1 . 1 . 22 M a s s t r a n s f e r

Ass mentioned above t h e secondary, in an L M X B or a CV fills its Roche lobe, a n d transfers masss t o t h e compact primary. This can h a p p e n either because t h e secondary e x p a n d s , or t h e Rochee lobe shrinks. In systems which have long orbital periods (P > 10 hrs) t h e secondaries aree generally evolved a n d t h e m a s s transfer is governed by t h e expansion of the secondary as aa result of its nuclear evolution. In systems with smaller orbital periods t h e secondaries are t h o u g hh t o b e main-sequence like stars and t h e reason why mass transfer occurs at t h e observed r a t ee is m o r e complex.

T h ee orbital angular m o m e n t u m of t h e b i n a r y system is given by

M\M\ M\M\ M ii + M2

(1.4) )

wheree t h e subscripts 1 a n d 2 again refer t o t h e p r i m a r y a n d secondary, respectively. If t h e m a s s transferr is conservative, i.e., no m a s s is lost from t h e system a n d no angular m o m e n t u m is lost fromm t h e orbital m o t i o n , t h e n we have M\ = - M2. Differentiating Eq. (1.4) one derives

// M A M2 VV Mj M2

8 8 _{11 Introduction and summary}

andd substituting Eq. (1.1) one obtains

ÊLÊL = 9 ^ - 2 (-- ¥l\ **1 t\ fC\

RLRL Jorb U Mj M2 ' {l'b)

Iff we assume a mass-radius relation for the secondary of the form R2 oc M£ we also have

RR22 M2

Thee requirement that the secondary always fill its Roche lobe

ÊlÊl

^L

RR22 RL

inn the case of conservative mass transfer (i.e., Jorb= 0) implies the condition

M?? 5 n

AJT66 + 2

( « )

Forr (hydrogen and helium) main-sequence stars n ~ 1, and for degenerate stars n ~ - 5 . Sincee for conservative mass transfer the orbital angular momentum is constant it follows fromm Eqs. (1.6) and (1.7) that for M2 > (§ + %) Mx the Roche lobe will shrink relative to the

secondary,, accelerating the mass transfer, and unstable mass transfer on a short time scale will occurr until the mass ratio has decreased sufficiently. On the other hand, if M2 < (§ + f ) Mu the

typicall case in LMXBs and CVs, the Roche lobe will actually expand relative to the secondary ass a result of mass transfer. In principle, mass will be transferred due to the slow evolutionary expansionn of the secondary on a nuclear time-scale. However, the secondary in this case will have aa low mass (less than ~1.4 MQ) and unless it has left the main sequence the evolutionary time

scalee is long, and the implied mass transfer rates are much less than the observed accretion rates. Therefore,, for these systems the only way of having sustained mass-transfer at the observed rates inn many of the LMXBs and most of the CVs is through the loss of orbital angular momentum. 1.1.33 A n g u l a r - m o m e n t u m losses

Thee question remains what drives the angular momentum losses from close binaries with periods off a few hours. The short periods of these binaries imply that gravitational quadrupole radiation (GR)) is significant (Kraft, Matthews and Greenstein 1962). Weak-field general relativity (e.g., Landauu and Lifschitz 1958) gives for the loss of angular momentum due to GR

JGRJGR _ 32 G3 MlM2{M1 + M2)

JGRJGR 5 c5 a5 " \ )

Usingg Eqs. (1.1) and (1.2) this implies for Roche-lobe filling main-sequence secondaries (R/RQ ~

M/M®) M/M®)

-Jtf2 l Cfl~10-1 0/(Jtfl ï g)^JJ MQyr~l , (1.10)

wheree / ( M i , q) is a slowly varying function of order unity. For systems below the period gap (^*orb~~ 2 hr) this gives mass transfer rates comparable to the observed rates. However, for systemss above the period gap {POT\,~ 3 hr) Eq. (1.10) gives values which are too low by one to two

orderss of magnitude (Patterson 1984). In particular they are inadequate to power short-period LMXBs,, which have Ix > 1036 ergs s"1. Therefore, a more efficient angular momentum loss

1.11.1 Close binaries containing a compact object 9 9 mechanismm is required above the gap. The exact nature of this mechanism and its dependence onn the mass of the secondary is currently a major source of uncertainty in the evolution of close binaries.. The most popular candidate is magnetic braking (Huang 1966, Mestel 1968, Eggleton 1976).. In this process the secondary loses angular momentum to the interstellar medium via aa weak (~ 10"1 4 M©/yr) stellar wind that follows the magnetic field lines anchored in the secondary,, and is at the same time kept in synchronous rotation (at the expense of orbital angularr momentum) by tidal interaction with the white dwarf. Thus if magnetic braking exerts aa spin-down torque on the secondary star this must extract angular momentum from the binary orbit.. The major problem is to give a quantative estimate of this torque. Prom the assumption thatt the observed correlation between age and equatorial velocity in G type stars (Skumanich 1972,, Smith 1972) is the result of magnetic braking Verbunt and Zwaan (1981) derived the followingg expression for the torque

JMBJMB = - 5 x l < T2 9r2*2M2f l4u ,3 . C1-11)

wheree k is the radius of gyration, u; the angular velocity of the star, and ƒ is a dimensionless parameterss of order unity. For main-sequence secondary it follows that

5/3 3

M2MBM2MB si 6 x 10

GeT»"-GeT»"-

11

(112)

Mestell and Spruit (1987) found a somewhat weaker dependence of -M on the period than given inn Eq. (1.12), which seems in better agreement with transfer rates derived from observations. However,, in view of the difficulties to derive accurate values of At for CVs the observational constraintss on the M - P relation are not very strong..

1.1.44 The period gap

Itt is very hard to think of an observational selection effect that might explain the lack of detection off systems with periods in the period gap. If systems evolve to shorter orbital periods it is clear thatt the mechanism that drives the mass transfer must become ineffective at a period of ~ 3 hr:: otherwise they would evolve into the gap and have mass transfer rates and luminosities comparablee with those systems on either side of the gap (note that the situation for LMXBs mightt be different; see Fig. 1.1 and the discussion below). Let us consider the thermal time scale off the secondary

TKHTKH =

whichh for main-sequence stars (L2/LQ ~ (M2/M0)3-5) becomes

(1.13) )

™ ** 1.6 X I * ( £ ) " % ( I - " )

Att Porb ~ 3 hr the mass transfer time-scale T^ = M2/M2 ^ rKH [here we assumed magnetic

braking,, but it holds for any driving mechanism giving the observed mass transfer rates]. Once T.yy ~ TKH the secondary star can no longer maintain its thermal equilibrium radius, and will be oversizedd for its mass (i.e., R2 - RL> RMS{M2)). If magnetic braking becomes significantly less

effectivee at Porb ^ 3 hr the mass transfer time-scale will become the one based on gravitational

radiation,, which is of order 4 109 _{yr at P}

o r b ~ 3 hr (see Eq. (1.10)), and thus substantially longer

thann TKH- The secondary will therefore shrink to its main-sequence radius, disconnect from the Rochee lobe, and mass transfer will cease. Only when the system's period, as a consequence of

10 0 _{11 Introduction and summary} thee continuing gravitational radiation, has decreased to about 2 hr does the secondary fill its Rochee lobe again, and mass transfer is resumed; the system has crossed the period gap without transferringg mass, and therefore without being detectable (Rappaport, Verbunt and Joss 1983, Spruitt and Ritter 1983).

Thee question remains what can cause the magnetic braking to switch off at Porb ~ 3 hr.

Robinsonn et al. (1981) noticed that at a period of P^ ~ 3 hr (M2 ~ 0.3 M0) the secondary

starss become fully convective. The idea arose that the vanishing of the radiative core either reducess the secondary's magnetic field by interfering with the dynamo process (assumed in somee models to require anchoring in the radiative core), or drastically reduces the secondary windd (again perhaps by interfering with the dynamo processes which are supposed to power the wind).. In this case one would expect also that isolated stars with M2 < 0.3 M0 (spectral later

thann M5) show a strong reduction in magnetic activity. However, observations do not provide compellingg evidence for such a reduction. Taam and Spruit (1989) proposed that the transition too fully convective secondaries does not switch-of the magnetic activity, but rather rearranges thee surface field, and they showed that such a rearrangement can reduce the effectiveness of the magneticc braking substantially.

1.1.55 T h e m i n i m u m p e r i o d

Theree is no reason why mass transfer should stop once a binary reaches a period of ~80 min. Inn fact gravitational radiation will operate more efficiently there than at longer periods (see Eqs.. (1.9) and (1.10)). The observed minimum period in CVs was independently explained by Paczynskii and Sienkiewicz (1981) and Rappaport, Joss and Webbink (1982). For P ^ t < 1.5 hr,, the time scale for mass transfer through gravitational radiation becomes smaller than the thermall time scale of the secondary, i.e. TGR = r^ < TKH- This means that the secondary

willl be oversized with respect to its mass, and implies a corresponding decrease in its central temperature.. The luminosity of the secondary will, therefore, decrease and the thermal time-scalee {TKH) will increase, driving the star still further out of thermal equilibrium. The secondary

evolvess to a degenerate state and the orbital period passes through a minimum period which dependss on the total mass of the system and the chemical composition (Sienkiewicz 1984). Since thee secondary continues to transfer mass the central temperature will drop further until nuclear burningg is extinguished and the secondary becomes degenerate. The secondary will then start too follow the mass-radius relation of the form R2 oc M2- 1 , and continuing mass transfer will

leadd to an increase of the orbital period.

1.22 Low-mass X-ray binaries

Persistentt LMXBs are typically 100-1000 times more luminous in X rays than in the optical. Thee optical continuum and line emission of luminous ( I x ~ 1036 ergs s_ 1) LMXBs is dominated byy reprocessing of X rays in matter surrounding the X-ray source. The phasing and amplitude of orbitall light-curves of LMXBs indicates that most of the optical continuum emission originates inn the accretion disk around the neutron star primary, with a significant contribution from the partt of the secondary not shielded from X rays by the disk (Van Paradijs 1983, Van Paradijs andd McClintock 1994b). The optical spectra of LMXBs consist of a blue continuum and a few ratherr weak high-excitation lines (in particular Hell 4686 A and the N m / C l l l 4630-4650 A Bowenn blend), and is unlike the spectrum of an ordinary star. The absolute visual brightness of LMXBss are in the range My ~ + 5 to -5 and show a clear trend with orbital period, consistent withh the idea that the optical brightness only depends on the X-ray luminosity and the size of thee accretion disk (Van Paradijs and McClintock 1994a).

1.21.2 Low-mass X-ray binaries 11 1 1.2.11 C l a s s i f i c a t i o n

1.2.1.11.2.1.1 Persistent sources

X-rayy studies of LMXBs have shown that these objects can be divided into distinct groups on the basiss of their X-ray timing and spectral characteristics. The different X-ray characteristics of the sourcess are probably related to the properties of the compact object: neutron star or black hole, strengthh of magnetic field and the mass accretion rate (see, e.g., Van der Klis 1994). Hasinger and Vann der Klis (1989) have shown that the persistent bright LMXBs can be divided into two groups, thee so-called "Z sources" and "atoll sources" (see Van der Klis 1991 for a recent review). The ZZ sources are so called because they describe a Z-shaped track in an X-ray "hardness-intensity" diagramm (comparable to a Hertzsprung-Russell diagram) and "colour-colour" diagram. Three branchess are distinguished in these diagrams, which are (for mostly historical reasons) called thee horizontal branch, the normal branch and the flaring branch. Optical and UV observations havee indicated that M increases from the horizontal branch, via the normal branch to the flaring branchh (Vrtilek et al. 1991). The noise properties and the presence of quasi-periodic oscillations (QPOs)) are correlated with the different branches. For two Z sources (Sco X - l and Cyg X-2) thee orbital period is known; both are relatively long (18.9 hr and 9.8 day, respectively). The atolll sources exhibit two branches in their X-ray colour-colour diagrams, called the "banana" (sometimess divided in the upper and lower banana) and the "island". Also in these sources the X-rayy noise properties are correlated with the different states. In atoll sources QPO's have not beenn observed. Optical evidence indicates that M increases from the island state, via the lower bananaa to the upper banana. The four atoll sources with know orbital period all have relatively shortt orbital periods (P<5 hr; the probable atoll source 2129+119 in the globular cluster M15 hass an orbital period of 17.1 hr).

1.2.1.21.2.1.2 Transient sources

AA separate group of LMXBs are the soft X-ray transients (SXTs, sometimes referred to as "X-rayy novae"). These sources are in a quiescent state for most of the time, then turn on and risee to maximum in typically a few days and afterwards decay more slowly on a time scale of usuallyy months. It is generally thought that the cause of the outbursts is the same as that whichh gives rise to dwarf nova outbursts in CVs (see below). At maximum SXTs strongly resemblee persistent LMXBs. In quiescence, the optical emission related to the X-ray source also disappears;; the secondary becomes visible and it has been possible in some systems to measure theirr radial-velocity curves. For four sources (0620-003, 2023+338,1124-684 and J0422+32) this providess evidence for the presence of a black-hole primary. Three of these sources have a 1-10 keVV X-ray spectrum which is "ultrasoft" (JbT ~ 1 keV). This confirms an earlier suspicion (White andd Marshall 1984) that an ultrasoft X-ray spectrum may be a good (but not perfect) way to selectt black holes in X-ray binaries. From four SXTs type I X-ray bursts (due to thermonuclear runawayss on the surface of a neutron star) have been observed which indicates that they contain neutronn star primaries. It is interesting to note that ~75% of the SXTs seem to contain black-holee primaries. The detected orbital periods in SXTs are all relatively long (P> 7 hr). Many SXTss which are thought to contain black-hole primaries show "glitches" in their X-ray and opticall outburst light curves that interrupt the exponential decay one to several months after thee main maximum. These glitches might be the result of enhanced mass loss from the secondary inn response to the outburst of X-ray illumination (Chen, Livio and Gehrels 1993; Augusteijn, Kuulkerss and Shaham 1993). Note however, that not all SXT X-ray light curves show the long approximatelyy exponential decay.

12 2 _{11 Introduction and summary} 1.2.22 F o r m a t i o n a n d e v o l u t i o n

1.2.2.11.2.2.1 Formatio n

Threee processes have been suggested for the formation of LMXBs in the galactic disk. Thee first model is accretion-induced collapse (AIC) of a white dwarf and was already sug-gestedd long ago (see, e.g., Whelan and Iben 1973, Van den Heuvel 1976). Since LMXBs have manyy characteristics in common with CVs, an evolutionary link between them by AIC may appearr quite plausible. However, so far there is no observational evidence to support this idea. AICC can only occur in CVs containing an O-Ne-MG (or possibly a CO) type white dwarf, which tightlyy constrains the allowed rate of mass transfer. For the formation of LMXBs with periods off a few hours (which do not obey these constraints) the white dwarf should have been formed withh a mass very close to the Chandrasekhar limit, because otherwise the accreted matter will havee been ejected in nova explosions, and the white dwarf will not cross this limit (see Van den Heuvell 1994 for a brief discussion of the different requirements for AIC to create a LMXB).

Thee second model involves three stars and was suggested by Eggleton and Verbunt (1986). Inn this model the original system consisted of a HMXB with a distant low-mass companion. Underr certain circumstances the massive secondary in a HMXB will engulf the neutron star primaryy which spirals inward to the core of the massive star and forms a Thome-Zytkov object. Thiss object has the radius of a red supergiant, and the low-mass companion will spiral down into itss envelope. If the mass loss rate from the envelope during spiral-in is sufficient a close binary remainss consisting of a neutron star (the core of the Thorne-Zytkov object) and the low-mass companion. .

Thee third model was suggested by Sutantyo (1975, 1992) and is similar to the model for thee formation of CVs (see below), but with a much more extreme mass ratio. In this model thee initial system consists of a massive primary star in a fairly wide orbit around a low-mass secondaryy star. The massive star will become a giant and engulf the secondary, and the system enterss into a common envelope phase in which the secondary spirals inward and the hydrogen richh envelope of the giant is ejected. After this phase the systems consist of a helium star, the coree of the primary, and the secondary in a ~0.5-1.0 day period. In this scenario the secondary mustt >1 MQ, because less massive secondaries cannot provide enough orbital energy to eject the commonn envelope without leaving a binary too small to accommodate the remaining evolution off the helium star. The helium star will ultimately explode as a supernova. In a symmetric explosionn the system receives a large runaway velocity due to the sudden mass ejection, and the orbitt becomes eccentric. Also, the neutron stars receive a kick velocity at birth. Due to tidal forcess and the loss of angular momentum the system will circularize and the secondary will fill itss Roche lobe and the system becomes an X-ray source.

LMXBss containing black-hole primaries cannot be formed by AIC. Also, accretion induced collapsee of a neutron star in an LMXB may be difficult since the secondaries in LMXBs may nott have enough mass to turn a neutron star into a black hole with a mass larger than 3Af0 (as

observedd in these systems). LMXBs containing a black hole might be formed by the last two modelss mentioned above if the original most massive star in the system was sufficiently massive (>4OM0;; Van den Heuvel and Habets 1984).

Inn globular clusters the dominating process for the formation of LMXBs is through close encounterss - either by tidal capture or by exchange collisions, e.g., binary-single star interactions whichh cause one binary component to be exchanged for a passing neutron star (see, e.g., Verbunt 1990,, Bhattacharya and Van den Heuvel 1991, Hut 1992). The reason why this process dominates iss the very high star density and low relative velocities in globular clusters.

LMXBss with periods below 80 min are thought to contain helium degenerate secondaries. Suchh systems are probably formed through a second common envelope phase in a similar way to thee AM CVn type CVs (see below). In this scenario the neutron star spirals into the envelope of

1.21.2 Low-mass X-ray binaries 13 3 thee secondary which has evolved into a giant. What remains is the degenerate core of the giant inn close orbit around the neutron star. Note however, that such short period LMXBs might also bee formed if the secondary started transferring mass (upon first contact with its Roche lobe) nearr the end of its main-sequence life (Iben and Tutukov 1984, Tutukov et al. 1985, Pylyser and Savonijee 1989). In this scenario the secondary has a degenerate helium core with a (relatively) hydrogenn rich envelope.

1.2.2.21.2.2.2 Evolution

Inn the case of an LMXB with a giant companion in a relatively long orbital period ( P > 1 day) thee mass transfer is driven entirely by the evolutionary expansion of the secondary which will ultimatelyy lose its envelope. Through the accretion of matter from the secondary the neutron starr will spin up, and after the exhaustion of the envelope we are left with a wide (orbital periods off days to years) detached system containing a white dwarf and a millisecond pulsar in a circular orbit.. In LMXBs with orbital periods in the range 10 hr < P < 1 day both magnetic braking andd evolutionary expansion of the secondary play a role in the mass transfer. In the case that thesee two are of comparable strength the end product of the evolution will be a radio pulsar inn circular orbit with a (very) low-mass companion and orbital period in the range 10 hr to 10 dayss (Pylyser 1988, Pylyser and Savonije 1988,1989).

Inn principle, one might expect that LMXBs with periods below 10 hr will, as result of the losss of angular momentum as described above, evolve down to periods of ~80 min. Ruderman, Shahamm and Tavani (1989), Ruderman et al. (1989) and Kluzniak et al. (1988), Kluzniak (1992)) pointed out that in LMXBs the high energy radiation produced by the interaction of the magneticc field of the neutron star with the accretion disk, may drive a wind from the companion byy heating of its outer layers. This evaporation might cause the accretion to stop. The spun-up neutronn star would then appear as a millisecond pulsar. The continuing gamma-ray flux from thee pulsar would eventually evaporate the companion completely, and leave a single millisecond pulsar.. This evaporation process might be relevant to the lack of observed LMXBs below the periodd gap. As shown in Fig. 1.1 there are, in contrast to CVs, virtually no LMXBs with periodss below 3 hr. If an LMXB enters the period gap the rapidly spinning neutron star will continuee to evaporate the secondary, and perhaps destroy it completely (Van den Heuvel and Vann Paradijs 1988, Ergma 1991, Fedorova and Ergma 1991). Recently it has also been suggested thatt the the radiation by the X-ray source may change the conditions in the outer envelope of thee secondary, change the mass transfer process, and effect the evolution (Podsiadlowski 1991, Harpazz and Rappaport 1991). By reducing the life time of LMXBs the above processes might be ablee to explain the apparent dearth of observed LMXBs, which suggest a birthrate insufficient too explain the number of their proposed descendants, i.e., the millisecond pulsars (Kulkarni and Narayann 1988).

1.2.33 S u m m a r y of r e s u l t s

Thee first part of this thesis is concerned with optical studies of LMXBs. In chapter 2 the identificationn is reported of the optical counterpart of the Z type source GX 349+2 (Sco X-2), whichh was suggested by Cooke and Ponman (1991) on the basis of an accurate radio position. GXX 349+2 is only the third Z source that has been optically identified, and the first such source inn 24 years. The orbital period of this source is probably in the range 1-20 days. The extinction towardss the source is A\ ~ 5 mag.

Inn chapter 3 we present the results of coordinated X-ray and optical observation of the Z typee source Sco X-l, which were part of a multiwavelength campaign. The main result was thee identification of the well known division at B ~ 12.8 between the optical bright and faint statess of Sco X-l, with the transition from the normal to the flaring branch in the X-ray

colour-14 4 _{11 Introduction and summary} colourr diagram. This confirms the earlier suggestion made by Priedhorsky et al. (1986). Prom archivall observations we find that in both optical states the orbital light curve is approximately sinusoidal,, with similar amplitudes.

Inn the fourth chapter we present a phase-resolved spectroscopic study of the optical counter-partss V801 Ara and V926 Sco of the atoll sources 1636-536 and 1735-444, respectively. We find thatt the radial-velocity variations in these sources are dominated by a component originating fromm the point where the mass stream from the donor star intersects the out disk. This compo-nentt seems to be relatively strong in LMXB with short (< 10 hr) orbital periods. The overall propertiess of the two sources are very similar.

1.33 Cataclysmic variables

Thee term "cataclysmic variables" is derived from the Greek word KaTa«.Auo-/«k, which means floodd or storm. This term has been chosen because all these stars are characterized by sudden increasess in brightness. The exception are nova-like stars which were added later (these sources doo not show outbursts, but their properties are basically the same as other CVs).

Thee optical continuum and line emission from CVs is generally dominated by the accretion diskk around the white dwarf primary. As the accreted material spirals inward through the disk itt releases its gravitational energy and heats the disk to temperatures of ~ 3 000 - 100 000 K. Thee bolometric luminosity of the disk is given by

11 GMWDM ^disk^disk

-ll it\VD

Inn most cases, this exceeds the visible luminosity of the two component stars, and hence the visiblee spectrum is dominated by the accretion disk. Only in systems with long orbital periods, andd hence larger and more luminous secondaries, or in systems with low accretion rates (e.g., dwarff novae in quiescence) can a contribution from either component star be detected. For recentt reviews on CVs the reader is referred to La Dous (1993) and Warner (1994).

1.3.11 C l a s s i f i c a t i o n

Theree is a large variety of types of CVs distinguished by their eruptive behavior (novae, dwarf novae),, the magnetic field strength of the white dwarf primary (AM and DQ Her stars), or their spectrall properties (nova-like variables). In many cases considerable overlap exists between the differentt types. Below I give a brief description of the different types (see, e.g., Warner 1994).

Classical novae have, by definition, only one observed eruption. Their amplitudes can rangee from 6 to 19 mag. The larger the amplitude of the eruption, the shorter its decay time.. The nova eruptions are caused by thermonuclear runaways of hydrogen-rich material accretedd on the surface of the white-dwarf primary.

Recurrent novae have more than one observed eruption. Their amplitudes range from 5 too 9 mag. Several of them have M giant secondaries.

Dwarf novae have outbursts in the range from 1 to 8 mag. The intervals between outbursts rangee from ~10 days to tens of years with a well defined time scale for each object. The outburstss last typical for 2-20 days. There are three distinct types of dwarf novae based onn the morphology of the outburst light curves.

ZZ Cam stars sometimes remain at an intermediate brightness level for periods of days to yearss ("standstill"). During these periods outburst do not occur.

1.31.3 Cataclysmic variables 15 5 SUSU UMa stars show normal outbursts at irregular intervals, and less often, but at more regularr intervals, superoutbursts which are somewhat brighter at maximum (~0.5-1.0 mag)) and last ~ 5 times longer. During superoutbursts SU UMa type dwarf nova show brightnesss variations ("superhumps") with a period which is slightly longer (~2-8%) than thee orbital period. All but one (TU Men) SU UMa type dwarf novae have orbital periods beloww the period gap.

UU Gem stars include all dwarf novae which have not been classified as a Z Cam or SU UMaa type dwarf nova. All definite U Gem and Z Cam stars have orbital period above the periodd gap.

Dwarff nova outburst are the result of a sudden increase in mass accretion onto the white dwarff either as a result of a mass transfer instability that arises in the secondary or an instabilityy in the accretion disk.

Nova-like Variables include all the CVs that do not show outbursts. This is a very hetero-geneouss group. It is thought to include pre-novae, post-novae and maybe Z Cam stars in standstilll which not yet have been observed to show a dwarf nova outburst. Nova-likes also includee VY Scl stars which show occasional reductions in brightness ("anti dwarf nova") fromm an approximately constant brightness level. Most nova-likes have emission line spec-tra,, but the subgroup of UX UMa stars have optical spectra that show, in addition, broad absorptionn lines.

Magnetic CVs are often included among the nova likes. In these systems the magnetic fieldd is sufficiently strong to channel the accretion stream onto the magnetic polar regions off the white dwarf. They fall into two groups:

Polars,Polars, or AM Her stars, show phase dependent optical polarization variations at the white dwarff spin period together with photometric, emission line and (usually) X-ray variations att the same period. In these systems the white dwarf spin period is equal to the orbital period,, which is thought to be due to magnetic interaction of the white dwarf with the secondary. .

IntermediateIntermediate polars, or DQ Her stars, show little or no optical polarization. The white dwarff spin period in these systems is (generally much) shorter than the orbital period.

Thee magnetic interaction is thought to be too weak to bring about synchronism, as in thee polars. Unlike polars these systems are thought to possess an accretion disk which is disruptedd near its inner edge by the magnetic field of the white dwarf.

AM CVn stars all have orbital periods below ~80 min and only show helium lines in their spectra.. The secondaries in these systems are thought to be helium degenerates.

1.3.22 Formation and e v o l u t i o n

Itt is believed that initially a system which is going to become a CV has a period of a few months too a few years, so that the two stars are widely separated. The component that is initially thee more massive of the two stars, the primary, proceeds faster with its nuclear evolution and evolvess all the way to a red giant or a red supergiant stage before overflowing its Roche lobe. Soo it may develop a massive degenerate core, or even a massive non degenerate core if the initiall mass of the primary was sufficiently large. As soon as the primary expands sufficiently too overflow its Roche lobe a phase of rapid mass transfer begins; since the primary is the more massivee component this mass transfer is not stable (see Eq. (1-6)) and leads to a build up of ann expanding envelope around the accreting secondary (a low-mass main-sequence star). The binaryy evolves via a contact configuration to one in which a common envelope engulfs two cores: a

16 6 _{11 Introduction and summary} degeneratee primary, and a main sequence secondary. The two cores transfer angular momentum too the common envelope and gradually spiral in, with ever decreasing period and separation. Finally,, either the two cores merge and we are left with a single star, or the envelope is lost andd a short-period binary is left. In the second case the primary will become a white dwarf, whilee the main sequence secondary is smaller than its Roche lobe and the system is detached (Eggletonn 1976, Paczynski 1976, Ritter 1976, Taam, Bodenheimer and Ostriker 1978, Meyer and Meyer-Hofmeisterr 1979, Livio, Saltzman and Shaviv 1979, Livio 1982, De Kool 1987).

Muchh later still, either as a result of angular momentum loss from the system, or as a resultt of nuclear evolution of the secondary, that star fills its Roche lobe, the system becomes semi-detached,, and a stage of mass transfer begins.

Itt is interesting to note that most CVs are expected to undergo nova explosions as a result off the continuing accretion of material onto the white dwarf. The only exceptions appear to bee the so-called "super soft sources" in which steady burning of the accreted material on the surfacee on the white dwarf is thought to occur. The presence of processed material in the ejecta off novae indicated that more mass is ejected during the outburst than the mass of hydrogen whichh is burned to helium to power the outburst. Thus the white dwarf is expected to decrease inn mass during the evolution of a CV.

Afterr a CV has evolved past the minimum period the binary separation will start to increase, andd the mass transfer rate (as a result of gravitational radiation) will strongly decrease. The evolutionaryy time scale of such a system becomes longer than the age of the galaxy, and these systemss will still have orbital periods close to 80 min. The expected very low accretion rates ( ~ 1 0- 1 22 MQ/yr) make it very difficult to detect such a system.

CVss with periods below 80 min are thought to contain helium degenerate secondaries. Such aa system is thought to be formed through a second common envelope phase in which the white dwarff primary spirals into the envelope of the secondary which has evolved into a giant (see, e.g.,, Iben and Tutukov 1984). What remains is the degenerate core of the giant in close orbit aroundd the white dwarf.

1.3.33 S u m m a r y of r e s u l t s

Thee second part of this thesis is concerned with optical studies of CVs. In chapters 5 and 6 we presentt optical photometry of intermediate polars (IPs) type CVs. From the optical observation off BG CMi presented in chapter 5 we find that the white dwarf rotation period decrease on aa time scale of ~ 6105yr. Using independent estimates of the mass transfer rate and the magneticc field strength of the white dwarf, we compare the observed time scale of the change in thee rotation period with two different accretion models. We find that the rotation rate of the whitee dwarf deviates substantially from the equilibrium value corresponding to the current rate oss mass transfer. The time scale of the period change in BG CMi is comparable to what is found inn other IPs and suggests that one physical time scale drives these period changes. In chapter 66 we present extensive photometric observation of the IP TV Col. We find that the average brightnesss of the system varies by A S j ~0.4 mag over periods as short as two weeks. These variatiess are probably the result of variations in the mass accretion rate. During the periods of highestt mean brightness the source shows outbursts with an amplitude of 1-2 mag. The optical brightnesss of the source varies at three different periods: a 5.5 hr period which is identified with thee orbital period, a 5.2 hr period, and a 4 day period which is the beat period between the 5.22 hr period and the orbital period ( P ^ P ^ - P ^ i ) . The 4 day period is thought to be the precessionn period of the disk. We are unable to derive any constant-period ephemeris for the 5.22 hr period, and we suggest that this period is in fact not stable. Because the orbital period iss stable and given the beat relation above, this should also apply to the 4 day period. We investigatee the changes in the optical light curve as a function of the 4 day cycle.

References References 17 7 Thee remaining chapters are concerned with optical observations of dwarf nova type CVs. Inn chapter 7 I present time resolved spectroscopy of the SU UMa type dwarf nova VY Aqr in superoutburstt and in quiescence. Using various observational constraints I derive the inclination andd mass ratio of the system. VY Aqr has system parameters very similar to those of OY Car, whichh is also an SU UMa type dwarf nova. However, the amplitudes of the outbursts seen in VY Aqrr are ~ 3 mag larger than those of OY Car. I argue that this is due to a lower mass transfer ratee during quiescence in VY Aqr compared to OY Car.

Inn chapters 8 and 9 photometric and spectroscopic observations are presented of the dwarf novaa V485 Cen. We find that the orbital period of this source is 59 min. This is very surprising ass it is far below the minimum period of ~80 min observed in the orbital period distribution off CVs. The detection of Ha in the spectrum of the source excludes the possibility that it is ann AM CVn type CV, The most likely explanation is that the secondary has a low, but finite, hydrogenn content.

Thee final chapter of this thesis gives an outline of a project which I started in 1992. The goal off the project is to make a comparative study of disk and halo CVs. We have selected samples off dwarf novae which represent disk and halo populations. Our primary aim is to increase the numberr of Population II dwarf novae with known orbital periods. By comparing the orbital distributionn and space densities of our two samples we hope to gain a better understanding of thee formation and evolution of CVs.

References s

Abt,, H.A. 1983, ARA&A, 21, 343

Augusteijn,, T., Kuulkers, E., Shaham, J. 1993, A&A, 279, L13 Bhattacharya,, D., Van den Heuvel, E.P.J. 1991, Phys. Rep., 203, 1 Chenn W., Livio M., Gehrels N. 1993, ApJ 408, L5

Cooke,, B.A., Ponman, T.J. 1991, A&A, 244, 358 Dee Kool, M. 1987, Ph.D. Thesis, Univ. of Amsterdam

Eggleton,, P.P. 1976, in Structure and Evolution of Close Binary Systems, eds. P.P. Eggleton, S. Mittonn & J.A.J. Whelan, Reidel, Dordrecht, Holland, p. 209

Eggleton,, P.P., Verbunt, F. 1986, MNRAS, 220, 13p Ergma,, E.V. 1991, Comments Astrophys., 15, 239 Fedorova,, A.V., Ergma, E.V. 1991, Sov. Astron., 35, 640 Harpaz,, A., Rappaport, S.A. 1991, ApJ, 383, 739 Hasinger,, G., van der Klis, M. 1989, A&A, 225, 79 Huang,, S.S. 1966, Ann. d'Astr., 29, 331

Hut,, P. 1992, in X-ray Binaries and Recycled Pulsars, eds. E.P.J, van den Heuvel & S.A. Rap-paport,, Kluwer, Dordrecht, Holland, p. 317

Iben,, I, Tutukov, A.V. 1984, ApJ, 284, 719

Kluzniak,, W. 1992, in X-ray Binaries and Recycled Pulsars, eds. E.P.J, van den Heuvel & S.A. Rappaport,, Kluwer, Dordrecht, Holland, p. 413

Kluzniak,, W., Ruderman, M., Shaham, J., Tavani, M. 1988, Nature, 334, 225 Kraft,, R.P., Matthews, J, Greenstein, J.L. 1962, ApJ, 136, 312

Kulkarni,, S.R., Narayan, R. 1988, ApJ, 335, 755

Laa Dous, C. 1993, in Dwarf Novae and Nova-Like Variables, Cataclysmic Variables, NASA/CNRSS Monograph Series on Non-Thermal Phenomena in Stellar Atmospheres, eds. M.. Hack & C. La Dous

Landau,, L., Lifschitz, E. 1958, The classical theory of fields, Pergamon Press, Oxford Livio,, M. 1982, A&A, 105, 37

18 8 References References Lubow,, S.H., Shu, F.H. 1975, ApJ, 198, 383

Marinus,, M. 1994, private communication Mcstel,, L. 1968, MNRAS, 138, 359

Mestel,, L., Spruit, H.C. 1987, MNRAS, 226, 57 Meyer,, F., Meyer-Hofmeister, E. 1979, AfcA, 78, 167 Paczynski,, B. 1971, ARA&A, 9, 183

Paczynski,, B. 1976, in Structure and Evolution of Close Binary Systems, eds. P.P. Eggleton, S. Mittonn & J.A.J. Whelan, Reidel, Dordrecht, Holland, p. 75

Paczynski,, B., Sienkiewicz, R. 1981, ApJ, 248, L27 Patterson,, J. 1984, ApJS, 54, 443

Podsiadlowski,, P. 1991, Nature, 350, 136 Popper,, D. 1980, ARA&A, 18, 115

Priedhorsky,, W., Hasinger, G., Lewin, W.H.G. et al. 1986, ApJ, 306, L91 Pylyser,, E. 1988, Ph.D. Thesis, Univ. of Amsterdam

Pylyser,, E., Savonije, G.J. 1988, A&A, 191, 57 Pylyser,, E., Savonije, G.J. 1989, A&A, 208, 52

Rappaport,, S.A., Joss, P.C., Webbink, R.F. 1982, ApJ, 254, 616 Rappaport,, S., Verbunt, F., Joss, P.C. 1983, ApJ, 275, 713 Ritter,, H. 1976, MNRAS, 175, 279

Ritter,, H., Kolb, U. 1994, in X-ray Binaries, eds. W.H.G.Lewin, J. van Paradijs & E.P.J, van denn Heuvel (Cambridge Univ. Press), in press

Robinson,, E.L., Barker, E.S., Cochran, A.L., Cochran, W.D., Nather, R.E. 1981, ApJ, 251, 611 Rudermann M., Shaham J., Tavani M. 1989, ApJ, 336, 507

Rudermann M., Shaham J., Tavani M., Eichler D. 1989, ApJ 343, 292 Sienkiewicz,, R. 1984, Acta Aston., 34, 325

Skumanich,, A. 1972, ApJ, 171, 565 Smith,, M.A. 1979, PASP, 91, 737

Spruit,, H . C , Ritter, H. 1983, A&A, 124, 267 Sutantyo,, W. 1975, A&A, 41, 47

Sutantyo,, W. 1992, in X-ray Binaries and Recycled Pulsars, eds. E.P.J, van den Heuvel & S.A. Rappaport,, Kluwer, Dordrecht, Holland, p. 293

Taam,, R.E., Bodenheimer, P., Ostriker, J.P. 1978, ApJ, 222, 269 Taam,, R.E., Spruit, H.C. 1989, ApJ, 345, 972

Tutukov,, A.V., Fëdorova, A.V., Ergma, E.V., Yungel'son, L.R. 1985, Pis'ma Astron. Zh., 11, 123 3

Vann der Klis, M. 1991, in Neutron Stars: Theory and observations, eds. J. Ventura & D. Pines, NATAA ASI C344, Kluwer, Dordrecht, Holland, p. 27

Vann der Klis, M. 1994, in Lives of the Neutron Stars, eds. A. Alpar, U. Kiziloglu & J. van Paradijs,, NATO ASI Series C (in press)

Vann den Heuvel, E.P.J. 1976, in Structure and Evolution of Close Binary Systems, eds. P.P. Eggleton,, S. Mitton & J.A.J. Whelan, Reidel, Dordrecht, Holland, p. 35

Vann den Heuvel, E.P.J. 1994, in Interacting Binaries, eds. H. Nussbaumer fe A. Orr, Springer, p.. 263

Vann den Heuvel, E.P.J, Habets, G.M.H.J. 1984, Nature, 309, 698 Vann den Heuvel, E.P.J., Van Paradijs, J. 1988, Nature, 334, 227

Vann Paradijs, J. 1983, in Accretion Driven Stellar X-ray sources, eds. W.H.G. Lewin & E.P.J. vann den Heuvel, Cambridge University Press, p. 189

Vann Paradijs, J. 1994, in X-ray Binaries, eds. W.H.G. Lewin, J. van Paradijs, & E.P.J, van den Heuvell (Cambridge University Press) (in press)

References References 19 9 Vann Paradijs, J., McClintock, J.E. 1994, A&A, 290, 133

Vann Paradijs, J., McClintock, J. E. 1994, in X-ray Binaries, eds. W. H. G. Lewin, J. van Paradijs, && E. P. J. van den Heuvel (Cambridge University Press) (in press)

Verbunt,, F . 1990, in Neutron Stars and Their Birth Events, ed. W. Kundt, Kluwer, Dordrecht, Holland,, p. 179

Verbunt,, F., Zwaan, C. 1981, A&A, 100, L7 Vrtilek,, S.D., et al. 1991, ApJ, 376, 278

Warner,, B. 1994, Cataclysmic Variable Stars, in preparation Whelan,, J., Iben, I. 1973, ApJ, 186, 1007

Partt I

Low-masss X-ray binaries

2 2

Thee optical counterpart of the Z source GX 349+2

W.. Penninx, T. Augusteijn

AstronomyAstronomy & Astrophysics 246, L81 (1991)

Abstract t

Wee have identified the optical counterpart of the Z source GX 349+2 with an 18thh magnitude star, whose spectrum shows strong Ha-emission. If this emission originatess from rotating material in an accretion disk around the neutron star we derivee a lower limit to the orbital period of 1.0 X sin3i day. If the companion is a giantt we derive for spectral type G5 and M2 upper limits to the orbital period of 11.22 and 19.5 days, respectively. The redenning towards the source is A y ~ 5.

2.11 Introduction

Thee persistently bright low-mass X-ray binaries can be divided into two groups, called Z sources andd atoll sources, on the basis of their X-ray spectral and X-ray timing behaviour (Hasinger and vann der Klis 1989; hereafter HK). The Z sources, which are the more luminous ones (~ 1038 _erg

s_ 1)) show a fairly uniform behaviour in the X-ray, radio and optical/UV bands (HK; Penninx 1989).. They show three 'spectral branches' in an 'X-ray colour-colour diagram' with correlated timingg characteristics. Of the six known Z sources, five have been detected as weak and strongly variablee radio sources (see also Cooke and Ponman 1991; hereafter CP). Only two Z sources havee been identified optically so far, Cyg X-2 (Giaconni et al. 1967) and Sco X-l (Giaconni ett al. 1962). Of the Z sources that have not been optically identified three have very large interstellarr extinction, as derived from X-ray observations (Schulz, Hasinger and Trümper 1989). Thee remaining source GX 349+2 (Sco X-2) has a relative low value of interstellar absorption (Schulzz et al. 1989) and might therefore be relatively easily detectable at optical wavelengths. Thee X-ray characteristics of GX 349+2 strongly resemble those of Sco X-l (see e.g. HK). One mightt therefore expect that the intrinsic properties of the optical counterpart of GX 349+2 wouldd also resemble those of Sco X-l.

Wee obtained two spectra of the optical candidate of GX 349+2, which was suggested by CP onn the basis of an accurate radio position, and looked for spectral characteristics that could supportt an optical identification of this Z source.

2.22 Observations and Analysis

Wee observed the optical candidate of GX349+2 (star 6 in the finding chart published by Jernigan ett al. 1979), with EFOSC on the ESO 3.6m telescope on August 1, 1990. We made two spectra

24 4 _{22 The optical counterpart of the Z source GX 349+2} X X 3 3

4000 0

50000 6000

W a v e l e n g t hh (A)

7000 0

F i g u r ee 2 . 1 . The 90 min spectrum of the proposed optical counterpart of GX 349+2. The foUowingg features have been indicated: the Ha emission and Na(D) absorption line; broad featuress due to the fringes in the CCD image (B) and two night-sky lines (NS). A comparison withh a heavily absorbed flat spectrum is also given. The two Unes correspond to deredenned fluxesfluxes of 1.5 and 40 mJy and A\ of 3 and 7, respectively

w i t hh t h e B300 grism (dispersion 230 A m m -1; 3.2 A p i x e l "1) a n d a high resolution R C A C C D c a m e r aa covering t h e r a n g e ~ 3 6 0 0 - 7 0 0 0 A . B o t h spectra were taken t h r o u g h a 1.5" slit giving a r e s o l u t i o nn of 13 A ( F W H M of t h e Helium-Argon lines of t h e comparison s p e c t r u m ) . T h e first s p e c t r u mm was t a k e n at U T Aug 1 1990 1:20 (45 m i n integration t i m e ) , a n d the second was taken a tt U T Aug 1 1990 2:24 (90 m i n ) .

D u r i n gg t h e 45 m i n exposure, p a r t of t h e s p e c t r u m fell on some b a d columns of t h e C C D , whichh caused p r o b l e m s in t h e night-sky s u b t r a c t i o n of t h e blue p a r t of t h e spectrum. T h e sky wass b r i g h t a n d strong night-sky lines are visible on the images. Fringes could not be fully taken o u t ,, a n d resulted in b r o a d artificial features in t h e s p e c t r u m . T h e 90 m i n s p e c t r u m is shown inn F i g . 2.1. T h e s p e c t r u m was wavelength calibrated using a Helium-Argon spectrum, and flux c a l i b r a t e dd using a n observation of Wolf 4 8 5 (Oke 1974). T h e s u b t r a c t i o n of the background s p e c t r a ,, in which fringes are somewhat shifted in wavelength with respect t o the star s p e c t r u m , r e s u l t e dd in t h e b r o a d features in Fig. 2.1. The background s u b t r a c t i o n was not perfect, a n d r e s u l t e dd in two features of t h e night-sky lines. We detect H Q emission at 6556.2 0.6 A, a n d N a ( D )) interstellar a b s o r p t i o n at 5890.6 0.6 A. T h e H a emission is also detected in t h e 45 m i nn s p e c t r u m . T h e equivalent w i d t h s of b o t h H a lines are 7.0 0.3 A; t h e full w i d t h at half m a x i m u mm is ~ 13.2 A. T h e central wavelength of t h e H a line in t h e 45 m i n spectrum is shifted w i t hh respect t o t h e s a m e line in t h e 90 m i n s p e c t u m by 4 A. T h e equivalent w i d t h of t h ee N a ( D ) line is 3.7 0.4 A. T h e quoted errors are l<r-errors.

O t h e rr stars t h a t were visible on the C C D , show no H a in emission; some show H a in a b s o r p t i o n . .

Additionallyy we m a d e B a n d V band images of t h e optical counterpart of GX 3 4 9 + 2 . We u s e dd s t a r E7-u as a calibration s t a r ( G r a h a m 1982). T h e m a g n i t u d e s derived for t h e optical c o u n t e r p a r tt of GX 3 4 9 + 2 , are B = l a n d V = . C P derived B = 20.2 a n d V = 18.7.. T h e y used s t a r s A a n d B (Penston et al. 1975) as comparison s t a r s . We derive for star B i nn o u r B b a n d i m a g e a brightness of 15.28, consistent with t h a t given by P e n s t o n ( mB = 15.35).

2 . 33 D i s c u s s i o n

I nn view of t h e k n o w n properties of the optical c o u n t e r p a r t s of low-mass X-ray binaries, t h e p r e s e n c ee of H a emission in t h e spectrum of star 6 is evidence t h a t this object is t h e optical

2.32.3 Discussion 25 5 counterpartt of the X-ray source. The radial velocity, as derived from the central wavelength of thee Ha-line 0 km/s) would not be expected from a single star being a member of the diskk population, and supports this identification.

Thee strength of the Ha line (EW ~ 7.0 A) is on the high side when compared to the Z sources Cygg X-2 (EW ~ 3.3-6.9 A, van Paradijs et al. 1990) and Sco X-l (Willis et al. 1980). If other spectrall features as found in Sco X-l and Cyg X-2 (He n lines, H/3, H7, H5, A4640) were present inn GX 349+2 with similar strengths, we would not have detected them. In contrast to the Z sources,, the atoll sources show no such Ha emission lines (Canizares, McClintock and Grindlay 1979).. Ha absorption Unes are possibly observed in the atoll sources 1636-53 and 1735-44 by Canizaress et al. (1979). It is not yet possible to decide whether the difference in presence of Ha emissionn lines between the atoll sources and Z sources imply fundamental differences in their structuree or merely differences of degree in one or another fundamental parameter of the system (e.g.. size of the disk).

Thee intrinsic optical spectra of disks are in general fairly flat (see e.g. van Paradijs 1983; Neugebauerr 1969 for Sco X-l). We have added in Fig. 2.1 flat spectra that are strongly absorbed. Thee derived spectrum is consistent with being a heavily absorbed flat spectrum (given the limitedd quality of the spectrum), in which case the spectrum is absorbed by Ay ~ 4-6, and a dereddenedd flux is ~ 5-20 mjy. This is similar to what was found by CP, who derived Av~ 5,

andd a dereddened mv=13.7 (13 mJy).

Thee wavelength difference 6 A) between the observed Na(D) line and rest wavelength (assumingg both Na(D) lines have equal strenghts) gives an average radial velocity of -120+30 km/ss for the absorbing medium. This velocity is probably the result of galactic rotation of the absorbingg medium. The strength of the Na(D) line (equivalenth width of 3.7 A) is consistent withh the derived interstellar absorption (^4v~ 5, CP, see also Fig. 2.1).

Thee FWHM of the Ha line (~ 13 A) is dominated by instrumental broadening. If we assume thatt this emission originates from an accretion disk this gives an upper limit for the velocity of rotatingg material of 300 km/s. Observed velocities of the rotating material are up to ~ 1000 (possiblyy 10 000 km/s) in other low-mass X-ray binaries (Canizares et al. 1979). The upper limitt to the rotating velocity is determined by the Kepler's third law, and gives a lower limit to thee distance to the neutron star of the region from which the Ha emission region originates of ~ 2.11 106X sin2i km (assuming a neutron star mass of 1.4 M0; i is the inclination). If we take this

distancee as lower limit to the radius of the Roche-lobe of the neutron star, and assume a lower masss limit of 0.08 M0 for the companion star, we derive a lower limit to the orbital period of

-vv 1.0x sin3_{i day. However, if the Ha emission originates from a (X-ray heated) region on the}

companionn star, or another fixed region in the binary frame (like the hot spot on the outside of thee disk), than the derived lower limits are not valid.

Orbitall velocity measurements of optical emission lines of Z sources have led to semi-amplitudess K ~ 60 km/s for Sco X-l (P = 18.9h; Cowley and Crampton 1975), and K ~ 2000 km/s (H/3) for Cyg X-2 (P = 9.8d; Cowley, Crampton and Hutchings 1979). For the atoll sourcess 4U1636-53, 4U1735-44 and GX 9+9 (all of which have orbital periods near 4 hr) semi-amplitudess K ~ 200 km/s have been found (Cowley, Hutchings and Crampton 1988). Assuming thatt GX 349+2 has a velocity amplitude of 200 km/s, one would expect that the binary sys-tematicc velocity is between -100 - -500 km/s. Sco X-l and Cyg X-2 do not rotate with galactic rotation,, and the indicated systematic velocity is no surprise.

Thee observed differences between the atoll and Z sources have been interpreted in terms off a difference in neutron-star magnetic field strength, ~ 1010 Gauss for Z sources, and <108-5 Gausss for atoll sources (HK). HK have suggested that these differences may have an evolutionary connection,, as the Z and atoll sources also seem to differ with respect to the stellar type of the masss donor star (both known companions of Z sources are [sub-jgiants, whereas 4 (out of 10)

26 6 _{References References}

knownn companions of atoll sources have [main-sequence or degenerate] dwarfs as mass donors). Thee suggestion of HK can be tested by determining the character of the mass donor of GXX 349+2. The best way to check this would be to determine the orbital period using velocity measurementt of the Ha-line.

Iff the companion were a giant (as in the case of Cyg X-2), one might be able to detect spectrall absorption features of the giant companion. We did not find any absorption features in ourr spectra. However, this non-detection is not very stringent; absorption features as observed inn the spectrum of Cyg X-2 (see e.g. van Paradijs et al. 1990, e.g. K(3 with equivalent width off ~ 2.5-7.5 A), would not be detectable in the present spectrum of star 6.

Iff the companion were a large giant, we would also expect that it would be a major contributor off the IR-light. The dereddened H-band flux of CP mH = 14.2 (possible completely due to

reprocessedd X rays) can be used as an upper limit to the IR light from a mass donor. We willl use an upper limit of mj>14.2 (J is more commonly used than H as a reference band) to derivee an upper limit for the luminosity and orbital period from a possible giant companion. Usingg mv= mj + (V-J), with V-J as given as a function of spectral type by Johnson (1966),

wee obtain mv£ l 5 . 7 2 and mv£ l 7 . 2 8 for assumed spectral types of the companion star of G5

andd M2 respectively. Using a distance of 9.2 kpc (Penninx 1989), this gives absolute visual magnitudess Mv>0.90 (G5) and Mv >2.46 (M2). Using the relation between absolute magnitude,

stellarr radius and (stellar-type dependent) surface bightness, given by Popper (1980), we find correspondingg upper limits to the companion star of GX 349+2 of 9.7 RQ and 13.7 RQ, for

assumedd spectral types of G5 and M2, respectively. Finally, using the relation between orbital periodd and average density of the companion star (and assuming q=Mo p t/Mn s<0.8, see Paczynski

1971),, we find corresponding upper limits to the orbital period of GX 349+2 of 11.2 and 19.5 days,, respectively.

Vrtilekk et al. (1990,1991) showed that for the Z sources Cyg X-2 and Sco X-l the intensity of thee reprocessed X rays (optical/UV radiation) varies by a factor of three between flaring, normal andd horizontal branch. Since GX 349+2 has never been observed in the horizontal branch, we expectt that a brightness variation of the optical counterpart (which is correlated with the X-ray variability)) is less than in the case of Cyg X-2 and Sco X-l, probably ~ 50 - 1 0 0 %; this assumes thatt a possible mass donor contibutes insignificantly.

AA project to find colour changes as a result of changing ratios of the brightnesses of a blue diskk and a possible red giant, and a study to derive an orbital velocity curve are under way.

2.44 Conclusion

Ourr observations of star 6 in the finding chart of Jernigan et al. (1979) support the proposal off CP that this star is the optical counterpart of GX 349+2. The source shows strong Ho in emission,, typical for Z sources.

Acknowledgements Acknowledgements

Wee thank B. Cooke and T. Ponman for providing us with their results before publication. We alsoo would like to thank Prof. Jan van Paradijs for carefully reading the manuscript.

References s

Canizares,, C.R., McClintock, J.E., Grindlay, J.E., 1979, ApJ 234, 556 Cooke,, B.A., Ponman, T.J., 1991, A&A, in press (CP)

Cowley,, A.P., Crampton, D., 1975, ApJ 201, L65

References s 27 7 Cowley,, A.P., Hutchings, J.B., Crampton, D., 1988, ApJ 333, 906

Giaconni,, R., Gorenstein, P., Gursky, H., Usher, P.D., Waters,J.R., Sandage, A., Osmer, P., Peach,, J.V., 1967, ApJL 148, L129

Giaconni,, R., Gorenstein, P., Paolini, F., Rossi, B., 1962, Phys. Rev. Letters 9, 439 Graham,, J.A., PASP 94, 244

Hasinger,, G., van der Klis, M., 1989, A&A 225, 79 (HK)

Jernigan,, J.G., Apparao, K.M.V., Bradt, H.V., Doxsey, R.E., Dower, R.G., McClintock, J.E., 1979,, Nature 272, 701

Johnson,, H.L., 1966, ARA&A 4, 193 Neugebauerr et al., 1969, ApJ 155, 1 Paczynski,, B., 1971, ARA&A 9, 183 Oke,, J.B., 1974, APJS 27, 21

Penninx,, W., 1989, in 'Proceedings of the 23r d ESLAB Symposium', Bologna, Italy, Sept. 1989, ESAA Publications, ESA SP-296, p. 185

Penston,, M.V., Penston, M.J., Murdin, P., Martin, W.L., 1975, MNRAS 172, 313 Popper,, D.M., 1980, ARA&A 18, 193

Schulz,, N., Hasinger, G., and Trümper, J., 1989, AfeA 18, 115

vann Paradijs, J., 1983, in 'Accretion Driven Stellar X-ray sources', eds. W.H.G. Lewin and E.P.J. vann den Heuvel, Cambridge University Press

vann Paradijs, J. et al., 1990, A&A 235, 156 Vrtilek,, S.D., et al., 1990, A&A 235, 162 Vrtilek,, S.D., et al., 1991, ApJ, in press Willis,, A. et al., 1980, ApJ 237, 596