Astron. Astrophys. 323, 399-414 (1997)

1. Introduction

Cygnus X-2 (discovered by Bowyer et al. 1965) is a very well studied bright low-mass X-ray binary (LMXB). Hasinger & van der Klis (1989) classified the brightest LMXBs into two sub-classes, i.e. the "Z" sources and the "atoll" sources, on the basis of their correlated X-ray spectral and fast timing behaviour. Cygnus X-2 was classified as a Z source. In the X-ray colour-colour diagram (CD) Z sources trace out a Z shaped track and move smoothly, without jumps, through the branches. In all the six known Z sources, except GX 349+2, three branches have been identified, the horizontal branch (HB), the normal branch (NB) and the flaring branch (FB). The transition between the HB and the NB is called the hard vertex, and between the NB and the FB the soft vertex. The variation of one single parameter, i.e. the mass transfer rate to the compact object (), is thought to produce the tracks (e.g. Hasinger & van der Klis 1989, Lamb 1991).

Recent studies of EXOSAT data of the Z sources (see Kuulkers et al. 1994a, 1996a, b; Kuulkers & van der Klis 1996) indicate that the Z sources can be divided into two groups, the Cyg-like sources (Cyg X-2, GX 5-1 and GX 340+0) and the Sco-like sources (Sco X-1, GX 349+2 and GX 17+2). The Cyg-like sources show motion of the Z pattern in the colour-colour diagram and the hardness-intensity diagram (HID), their HBs are horizontal, and when the sources are on their short FBs the X-ray intensity generally decreases. The Sco-like sources, instead, do not display significant motion of their Z pattern in the CD, their HBs are almost vertical, and the X-ray intensity increases when they move onto their extended FBs. It was proposed (see Kuulkers & van der Klis 1995b, and references therein) that the Sco-like sources are viewed face-on (low orbital inclination) and the Cyg-like sources more edge-on (higher orbital inclination), and that the motion of the Cyg-like sources could be caused by accreting material, such as a precessing accretion disk getting into the line of sight.

Theoretical studies (Psaltis et al. 1995) indicate that a difference of magnetic field strength can explain some of the differences between the Cyg- and the Sco-like sources. In their model, the Cyg-like sources have a somewhat higher magnetic field strength (B Gauss) than the Sco-like sources (B Gauss). Their so-called unified model can explain the Z tracks in the CDs and HIDs and the difference in the Z shapes between the Cyg-like and Sco-like sources. However, no explanation is given for the fact that the Cyg-like sources display motion of the tracks, whereas the Sco-like sources do not.

Cygnus X-2 shows the most pronounced motion of the Z pattern in the CD and the HID of all Z sources. One of the first observations of Z pattern motion in the HID was found by Vrtilek et al. (1986) using Einstein MPC data. They observed a factor of two increase in intensity between different epochs but also an increase of the intensity on time scales less than a day. At that time the Z track behaviour of Cygnus X-2 was not yet known, so they interpreted the variations as due to orbital effects. Nowadays we know that most of the intensity variations of Cygnus X-2 are due to the non-periodic motion of the source through the Z track, and most likely associated with changes in . However, not all intensity variations that were found by Vrtilek et al. (1986) can be explained in this manner. In their Fig. 2, three branches are present which now can probably be identified as NBs. These branches are shifted with respect to each other, making this in retrospect the first indication that Cygnus X-2 displays motion of the Z-track in the HID.

Clear evidence for Z track motion in the CD and the HIDs was found when different EXOSAT observations of Cygnus X-2   were compared with each other (Hasinger et al. 1985b, Hasinger 1987, 1988). Hasinger (1987) discussed six EXOSAT observations of Cygnus X-2. He found that the Z tracks of different observations were displaced in the HIDs with respect to each other. Hasinger et al. (1985b) and Hasinger (1988) reported a state of Cygnus X-2 with very low count rates. The count rates, especially in the high energy bands, flared and the spectrum was hard. In the CD this observation looked like a large FB, and it is displaced to harder colours with respect to other Cygnus X-2 EXOSAT observations (see also Kuulkers et al. 1996a).

Vritlek et al. (1988) investigated the long-term temporal variability of Cygnus X-2 using six different instruments on three satellites (OSO 8, HEAO 1, and Einstein). Three count rate states between 2 and 10 keV were found. During the lowest state the spectrum was harder than during the other two states.

The motion of the Z trough the CD and HID was again reported in early Ginga observations of Cygnus X-2 reported by Hasinger et al. (1990). They found that the Z track in the HID during the October 1988 observation was shifted to higher intensities (by a factor two) with respect to the July 1988 observation. In the CD the Z track of the October 1988 observation was shifted downward parallel to the NB with respect to the July 1988 observation. The shape of the Z track in both the CD and the HID was different between these two observations. During the July 1988 observation the Z track performed a loop in the HID at the NB-FB transition, such that along the FB the source first rose in intensity and later decreased. During the October 1988 observation no such loop was found. Instead, along the FB the source immediately started to decrease in intensity. In the CD of the July 1988 observations the HB is diagonal and the FB is large. In the October 1988 CD the HB is horizontal and the FB is very small.

Kuulkers et al. (1996a) reanalysed all EXOSAT observations of Cygnus X-2. They also reported motion of the Z pattern and different shapes of the Z track. They compared their results with all previous reports on Cygnus X-2, and described the phenomenolgy in terms of three different intensity levels (possibly part of a continuous sequence). In the rare low level Cygnus X-2 displays a hard spectrum at the lowest intensities. These episodes seemed to occur at binary phases 0.8-0.2 (with phase 0.0 defined as the X-ray source superior conjunction), however not in each orbital cycle. Vritlek et al. (1988) and Kuulkers et al. (1996a) discussed various possible explanations, e.g. that at these times the secondary hides its heated face and/or part of our view of the inner disk region, so that most of the observed radiation then comes from a scattering hot corona surrounding the inner accretion disk, or that a tilted disk or a hot spot at the outer disk hide much of the inner disk region from our view. The other two intensity levels were called the medium and the high level. A periodicity in the occurence of these intensity levels could not be found. The shape of the Z track in the HIDs was dependent on these intensity levels. As the source moves onto the FB during high level episodes, the intensity immediately starts to decrease and the FB in the HID is almost horizontal. They called this a 'colour-independent' dip. When the source moves up the FB during medium level episodes the intensity first increases and then decreases. The FB in the HID is also pointed upwards and therefore referred to as a 'colour-dependent' dip. No Z track was observed during the low level.

The fast timing behaviour of Z sources is closely related to the position of the sources in the CD and HID (Hasinger & van der Klis 1989). On all branches a very-low frequency noise (VLFN) and a high frequency noise (HFN) component in the power spectra can be identified. On the horizontal and normal branches also low frequency noise (LFN) can be identified which decreases in amplitude from the HB to the upper normal branch and disappears on the lower normal branch. Quasi-periodic oscillations (QPOs) are found on the horizontal branch (called HBOs) with a frequency from 12 Hz at the left end of the HB up to 55 Hz near the hard vertex, and then remains constant upto halfway the normal branch. However, Wijnands et al. (1996a) found a decrease of the frequency of what is probably the HBO down the NB in GX 17+2. On the NB QPOs (NBOs) are observed with a frequency between 5-7 Hz. Sometimes the HBO is seen simultaneously with the NBO, indicating different mechanisms are responsible for these QPOs. On the flaring branch sometimes a QPO (FBO) is seen with a frequency near 7 Hz at the beginning of the FB, but rapidly increasing in frequency up to Hz further up the FB. The 7-20 Hz flaring branch QPO has not been found on Cygnus X-2 . However, Kuulkers & van der Klis (1995a) detected a QPO on the FB of Cygnus X-2 with a frequency of 26 Hz. This QPO was found in an intensity dip on the FB and is believed to be a different phenomenon from the 7-20 Hz FBO.

Two Z sources are known to show bursts. GX 17+2 (Tawara et al. 1984; Kahn & Grindlay 1984; Sztajno et al. 1986; Kuulkers et al. 1994b, 1996b) shows bona fide type I burst (thermonuclear flashes on the neutron star surface, Hoffman et al. 1978). In Cygnus X-2 burst-like events have been reported (Kahn & Grindlay 1984; Kuulkers et al. 1995). Kuulkers et al. (1995) detected 9 burst-like events in the archival EXOSAT data of Cygnus X-2. These burst-like events had a short duration ( 3 s), did not show evidence for cooling, and did not occur in specific regions of the Z track. To date it is unclear wether these burst-like events are genuine type I bursts.

In this paper we present an uniform analysis of all available data obtained by the Large Area Counter (LAC) instrument on board the Ginga satellite.

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de