SpringerLink
Forum Springer Astron. Astrophys.
Forum Whats New Search Orders


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

Previous Section Next Section Title Page Table of Contents

6. Discussion

We have found several characteristics in the behaviour of Cygnus X-2 which seem to depend on overall intensity level. In the following we discuss some of those characteristics in more detail and suggest possible explanations for the different overall intensity levels and associated phenomena.

6.1. The rapid X-ray variability during different overall intensity levels

We find conclusive evidence that the rapid X-ray variability at the same point on the NB differs between different observations. We find a different kind of high timing behaviour near the soft vertex when Cygnus X-2 is in the medium level and the high level (see Fig. 8). In the medium level weak, flat VLFN is detected and a pronounced NBO occurs. In the high level the VLFN is stronger and steeper, and the NBO is not detectable. In the intermediate intensity level, the VLFN is strong and steep in the FB, but weak and flat in the NB, and the NBO is about the same strength as during the medium level. Our results are the first unambigous detection of differences in the rapid X-ray variability at a specific position in the Z track (near the soft vertex) between different observations (see Dieters & van der Klis 1996 for a study of Sco X-1 with respect to this issue). So far, we find no evidence for any difference between the properties of the X-ray variability during different intensity levels on other parts of the Z track (such as the HB). However, our data do not allow a very detailed comparison of the timing behaviour on those parts of the Z track.

It is thought (van der Klis et al. 1985, 1987; Hasinger & van der Klis 1989; Lamb 1991) that variations in  [FORMULA]   are responsible for the changes in the rapid X-ray variability along the Z and the formation of the Z track, and that for this reason the rapid X-ray variability is closely related to the position of the source on the Z track. However, the shifting and shape changing of the Z track in Cygnus X-2 and other Z sources suggest that variations in  [FORMULA]   can not explain all aspects of the formation of the Z track. We, therefore, suggest that both the shape of the Z track and the rapid X-ray variability are not totally determined by the mass accretion rate onto the compact object, but also partly by a so far unknown process, which may also cause the long-term intensity variations (see Sect. 6.4).

Two other Z sources, GX 5-1 and GX 340+0, display motion of their Z track in the CD (Kuulkers et al. 1994a; Kuulkers & van der Klis 1996), although not as pronounced as Cygnus X-2 . Kuulkers et al. (1994a) and Kuulkers & van der Klis (1996) did not find any significant changes in the rapid X-ray varability in GX 5-1 and GX 340+0, respectively, when the Z track moved through the CD and HID. If the motion of the Z track in the CD and HID in Cygnus X-2, GX 5-1, GX 340+0 are caused by the same phenonemon, we then predict that the rapid X-ray variability in GX 5-1 and GX 340+0 changes when the Z tracks of those sources move through the diagram. The reason why so far no changes have been found is most likely due to the much lower amplitude of the motion of the Z through the CD and HID of GX 5-1 and GX 340+0, as compared to Cygnus X-2 , which suggest that the amplitude of the difference of the rapid X-ray variability would also be much smaller. Moreover, in Cygnus X-2 sofar only differences in the rapid X-ray variability were found near the soft vertex and not on e.g. the HB, while GX 5-1 and GX 340+0 were mainly observed in the HB and upper NB, and hardly near the soft vertex. Kuulkers et al. (1994a) and Kuulkers & van der Klis (1996) could not make a good comparison of the rapid X-ray variability near the soft vertex when the Z track of GX 5-1 and GX 340+0, respectively, moved in the CD and HID.

6.2. Velocity of motion along the Z track

Van der Klis (1991) suggested that the VLFN could be (partly) due to motion of the source along the Z track. In our analysis, we found that both the VLFN fractional amplitude and the velocity of motion along the Z track increase when the overall intensity level increases. However, the strongest VLFN was not observed during the most extreme high level, but during the less extreme, slightly lower one, whereas the highest velocity was indeed observed during the most extreme high level. Therefore, we conclude that the motion along the Z track is at most only partly responsible for the VLFN. As differences in NB slope in the HID can not explain this, part of the intensity variations causing the VLFN must take place perpendicular to the NB.

The increase in velocity along the Z track when the overall intensity increases indicates that the spectrum changes more rapidly during high level episodes than during medium level episodes. If the Z track is traced out by changes in the mass accretion rate onto the neutron star, then an increase in the velocity of motion along the Z track indicates that  [FORMULA]   changes more rapidly when the overall intensity level increases. However, it then seems unlikely that variations in  [FORMULA]   can explain the shifts and changes in shape of the Z track between intensity levels (see also Sect. 6.1). It is possible that the increase in velocity of motion along the Z track is due to the same phenomenon causing the Z tracks to shift and the shapes to change.

6.3. The width of the NB in the HID

The width of the NB in the HID increases when the overall intensity increases, while the width of the NB in the CD remained approximately constant. This indicates that the difference in the width of the NB in the HID is due to intensity variations and not due to spectral variations. However, at the same time the velocity of motion along the Z track increases, which indicates an increase in spectral variations. So, in the HID, when the source moves perpendicular to the Z track (change in intensity), it also moves along the Z track (change in spectrum). Perhaps colours are well correlated with [FORMULA]  , but the intensity varies also due to another (unspecified) process that is more prominent in the high overall intensity level.

6.4. The overall intensity variations

Several models have been proposed (Priedhorsky & Holt 1987, and references therein) in order to explain long-term intensity variations in X-ray binaries, e.g. long-term variations in [FORMULA]   and precessing accretion disks. Precessing neutron stars have also been proposed to explain the variations (see Priedhorsky & Holt 1987; Schwarzenberg-Czerny 1992, and references therein).

6.4.1. Variations in the mass accretion rate

Long-term changes in the mass accretion rate have been proposed (see Priedhorsky & Holt 1987) to explain the long-term intensity variations in low-mass X-ray binaries other than Cygnus X-2 (e.g 4U 1820-30, the Rapid Burster, Aql X-1, excluding Her X-1). However, Kuulkers et al. (1996a) and Wijnands et al. (1996b) argued that the long-term intensity variations in Cygnus X-2 can not be due to variations in the mass accretion rate. The main argument is that variations in the mass accretion rate are thought to produce motion of the source along the Z track (see e.g. Hasinger & van der Klis 1989), while the Z track is observed during several different intensity levels.

6.4.2. A precessing accretion disk

In order to explain long term intensity variations, not related to orbital variations, in high-mass X-ray binaries and Her X-1, precessing accretion disks have been proposed (see Priedhorsky & Holt 1987 and references therein). The recent detection (Smale et al. 1996, Wijnands et al. 1996b) of a 78 day period in the RXTE, Vela 5B (see also Smale & Lochner 1992) and Ariel V all sky monitor data of Cygnus X-2 favours a precessing accretion disk in Cygnus X-2.

However, explaining the long-term X-ray variations of Cygnus X-2 with a precessing accretion disk is not without serious contradictions. If we assume that, during the medium overall intensity level, the emission region is blocked by (part of) the accretion disk and much radiation is absorbed, scattered and/or reflected, the power spectrum should be blurred and quasi-periodic oscillations (both the NBO and the HBO) should be harder to detect than when the emission region is not blocked by the accretion disk (the high level). However, we see exactly the opposite: no NBO is detected during high overall intensity levels. The upper limits are significantly lower than the actually detected values during the medium levels. If we assume that not during the medium level but during the high level the emission region is blocked by the accretion disk, then the difference between the power spectra are not totally unexpected. However, it is difficult to explain the increase of the count rate when the accretion disk is blocking our view of the emission region.

Another possibility is that the accretion disk blocks, during the medium overall intensity levels, a localised emission area (e.g. the neutron star surface), which is not blocked during the high levels. In this region no NBO occurs and therefore the fractional amplitude of the NBO during the high level is diluted by the additional flux. However, not only the strength of the NBO should be affected, but also the strength of the HBO, except when the HBO would originate from this localised area, in which case we would expect an inverse effect. This, however, is unlikely, because during all intensity levels HBOs are observed at approximately the same fractional amplitude.

Therefore, it is unlikely that the increase in count rates from the medium to the high level is caused by an localised emission region, which is sometimes hidden from our view by a precessing accretion disk, or by a precessing accretion disk, which sometimes blocks part of the total X-ray radiation.

Kuulkers et al.(1996a) proposed that the difference in overall count rates in the Cyg-like sources is caused by anisotropic emission, the radiation being scattered preferentially into the equatorial plane by a puffed-up inner disk structure. In this model, with increased scattering in the high level, the decrease in NBO amplitude follows naturally due to light travel time effects in the scattering process. Our observations therefore support their model.

6.4.3. A precessing neutron star

Another option is that the differences between the levels are due to changes in the properties of the emission region. A precessing neutron star could produce changes in the inner disk region. Due to the precession of the neutron star the orientation of the magnetic field with respect to the accretion disk changes. According to numerical calculations (Psaltis et al. 1995) the strength of the magnetic field is enough for the field to have a profound effect on the spectrum of the source and possibly also on the count rate. A change in the effective magnetic field, caused by a different orientation of the field, could maybe explain the different Z tracks during the different levels. Also, the rapid X-ray variability would be affected, although it is not clear if this could explain the differences that we found. The behaviour of the HBO should differ between the different intensity levels due to the different effective magnetic field, but no difference is found. It is also not clear why the NBO should differ between intensity levels.

6.5. The decrease of the HBO frequency down the NB

The decrease of the HBO frequency from the hard vertex down the NB in the June 1987 and October 1989 PC data allows us to apply the method described by Wijnands et al. (1996a), in order to derive an upper limit for the equatorial magnetic field strength. Assuming that the radius of the neutron star is [FORMULA]  cm we derive an upper limit of [FORMULA] G on the star's magnetic field at the magnetic equator. This value is similar to the value found for GX 17+2 (Wijnands et al. 1996a). This upper limit is consistent with numerical computations on the X-ray spectrum of Cygnus X-2 (Psaltis et al. 1995). As noted by Wijnands et al. (1996a), a further decrease of the frequency down the NB will reduce this upper limit. The fact that we see the HBO frequency decrease on the NB in Cygnus X-2 supports the interpretation that the QPO on the NB in GX 17+2, which also decreases in frequency (Wijnands et al. 1996a), is the HBO.


[TABLE]

Table 4. Results of power spectral fits of the MPC3 data [FORMULA]

[TABLE]

Table 5. Results of power spectral fits of the PC data [FORMULA]



Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de