Astron. Astrophys. 355, 607-616 (2000)

3. Discussion

3.1. WR 137

WR 137 = HD 192641 (WC7 +?) has been studied in the infrared (IR) and peaks in brightness were reported in 1984.5 and in 1997, probably caused by heated dust (Williams 1997). The dust emission has been directly IR-imaged at two epochs recently using the Hubble Space Telescope by Marchenko et al. (1999). The repetition of IR maxima occurs with a 13 yr period, suggesting a possible binary origin, as found for other WR periodic dust makers. WR 137 was discovered to be a spectroscopic binary by Annuk (1995). However, Underhill (1992) did not find any evidence for binary motion in her data. Therefore, the binary status of WR 137 remains uncertain.

Marchenko & Pikhun (1992) published a long-term photometric study for 1958 - 1989, but it is based on photographic plates and the accuracy is insufficient to reveal light variations below a few per cent. Our photometry is presented in Table 2 and the light curves are shown in Fig. 1. We searched for periodicities using the procedure of Lafler & Kinman (1965), in the period range from 1 d to 100 d, but no period could be found. The only photometric variations we can see in our data are random light variations with amplitudes of 0.02 mag (peak to peak) in V during each observing season and up to 0.03 mag (peak to peak) when we compare different years. (However, the peak to peak amplitude from all data is 0.05 mag in B, and 0.07 mag in U.) During 1991-1998 22 measurements of the check star HD 192987 were obtained. The mean values (N = 22) of the magnitude differences (HD 192538 minus HD 192987) and their standard deviations are mag and mag. The scatter in Fig. 1 is greater than the observational error ( in ) and, therefore, probably contains real erratic variations with small amplitudes.

Fig. 1. Light curves of WR 137 (data from Table 2)

Table 2. Differential photometry of WR 137 (= HD 192641) - in the sense comparison star HD 192538 minus WR 137

In 1997, when the last peak in the IR was observed (Williams 1997), no photometric effect can be seen, apart from small-amplitude random variations. Their origin should arise in the continuum, as the plots in Fig. 2 suggest: There are some correlations ( for B and V and for U and B) between the lightcurves in each of the three passbands, which would be difficult to explain by variability of emission lines. The origin of the small-amplitude random continuum variations of WR 137 is possibly related to dynamical wind instabilities, resulting in temperature effects at the "pseudo-photospheric" level.

Fig. 2. Random light variability correlations for WR 137 (data from Table 2)

3.2. WR 140

WR 140 = HD 193793 (WC7 + O4-5) is another periodic dust maker. Williams et al. (1978, 1987a, 1987b, 1990) and Williams (1997) reported variations in the IR, revealing brightenings in 1977, 1985, and in 1993, which they attributed to the building of dust grains in the WR 140 wind with a period of 7.94 yr. The re-occurrence of the heated dust has been interpreted as due to wind-wind interaction in a binary system. Earlier spectroscopic studies failed to reveal the binary motion. However, a re-analysis of earlier published radial velocities and using the period in the IR (7.94 yr) led to a successful determination of the orbit (Williams et al. 1987c). It was found that the grain formation coincides with the periastron passage (PP) in the system (actually occurring before PP). This discovery was later confirmed by Moffat et al. (1987) and now presents the basic model for studies of WR 140. Williams et al. (1990) and van der Hucht et al. (1991) reported on variability of WR 140 at X-ray, UV, IR and radio-wavelengths. Our photometry of WR 140 is presented in Table 3 and the light curves are shown in Fig. 3. From Fig. 3, there is clear evidence for a dip in the light in 1993, between orbital phases 0.9 and 1.1. The dip is seen in all passbands and should therefore be due to continuum light attenuation. The amplitude of the "eclipse" in the V passband is 0.03 mag.

Table 3. Differential photometry of WR 140 (= HD 193793) - in the sense comparison star HD 193888 minus WR 140. Orbital phases are calculated with  d and .

Fig. 3. Long-term light variations of WR 140 (data from Table 3)

Two remarkable features are to be mentioned. First, the very broad shape of the light minimum, assuming a smooth trend between yearly data. After 1993, the light gradually increased to reach the "pre-eclipse" level in 1997, or even 1998. Considering the "eclipse" to be caused by an obscuration of the star(s) by the wind, the light curves strongly suggest that a dust envelope was built up around the WR star by the wind-wind interaction at the PP, which was gradually dispersed in the following years. Possibly it is the same dust observed in the IR when still heated. Second, it is apparent (Fig. 3) that the amplitude of the eclipse increases towards shorter wavelengths. In the lower panel of Fig. 3, the variation of the colour is shown, which is in the sense: WR 140 colour gets redder when its light is attenuated. This conclusion is easily obtained when considering the magnitudes of the comparison star HD 193888, which are: , , , and . The amplitude of the colour variation in of WR 140 is 0.04 mag (again Fig. 3). As is well known, the main source of opacity, (non-relativistic) electron scattering, has no effect on colours. Thus, in this case electron scattering alone is not sufficient, and an additional opacity source should be introduced, possibly Rayleigh (or Mie) scattering by small carbon dust particles.

Occasional "eclipses" have been observed in the carbon-rich late-type WC stars WR 103, WR 113, and WR 121 (for a history of "eclipses" see Veen et al. 1998). In these cases the "eclipses" were caused by occasional formation of dust in the line-of-sight. Although dust formation in the winds of late-type WC stars is now well established, the problem with grain condensation in the very hostile environments where the grains are believed to form remains unsolved. Clearly, a trigger is needed to start the grain formation. In the case of WR 140, this could be the shock compression in the colliding winds at PP. We assume that the fading of WR 140 shortly after PP is due to dust condensation in the wind of the WC star. After the condensation ceases the star brightens again because the dust is blown away and gradually dispersed. The "eclipses" studied by Veen et al. (1998) have typical amplitudes of several tenths of a magnitude and last from several days up to a month. In contrast, the amplitude of the light dip in WR 140 is much smaller and the recovery of brightness lasts several years. This implies continuing supply (expanding from the PP production + new?) of dust, even 2 - 3 years after PP. If there is new dust, this would be really surprising, since the trigger seems no longer to be effective. Following the procedure of Veen et al. (1998, using their Eqs. (5), (6), and (7)) and taking the terminal velocity from Eenens & Williams (1994), we obtain for the distance of the dust formation region from the WC star in WR 140: . This is only a rough estimate, but it is much larger than the respective distances for all "eclipses" studied by Veen and co-workers. It is also much larger than the radius of the shell of WR 140 obtained by Williams et al. (1987a) which is . Taking for the carbon particle density we get for the dust mass production rate (over unit area) the value . These results should be taken with caution because of the small amplitude of the "eclipse" in WR 140 and of possible deviations from the model used (e.g. continued supply of dust after PP).

WR 140 was observed photometrically during PP in 1977 by Fernie (1978) but no changes of brightness were found. This is likely due to his low precision data.

Like WR 137, the observations of WR 140 also show small-amplitude, day-to-day random light variations (amplitudes up to 0.02 mag), in addition to the eclipse variation. Fig. 4 shows the correlations of the random light variations in UBV, indicating that they are likely due to continuum rather than emission line variations (similar to WR 137, Fig. 2). Dynamical wind instabilities could be the origin, as in WR 137. Moffat & Shara (1986) suggested a 6.25 d period for the light variations they observed in WR 140, which, however, does not fit our data. Our observations during 1991 - 1998 cover 90% of the orbit. It remains to be seen whether the forthcoming PP in 2001 will repeat the light curve so far observed.

Fig. 4. WR 140. Random light variability correlations for 1994 (data from Table 3)

3.3. WR 148

WR 148 (= HD 197406, WN8 + c?) is a single-line spectroscopic binary, possibly hosting a compact companion. The star has been studied by Bracher (1979). She determined the orbital period as  d and also found light variations with the same period and an amplitude of 0.04 mag in V. Further spectroscopic studies by Moffat & Seggewiss (1979, 1980) revealed an unusually low mass function of the system, which was later confirmed by Drissen et al. (1986): f(m) = 0.28 . WR 148 has also an exceptionally large distance from the galactic plane, pc (Moffat & Isserstedt 1980; Dubner et al. 1990). Smith et al. (1996) found that WR 148 is a WN8 star. The low mass function and high z value led Moffat & Seggewiss (1980) to advance the idea that WR 148 harbours a compact companion as product of a supernovae explosion some 5 Myr ago. In their model, the companion is orbiting within the WR envelope. As the companion orbits around the WR star the projected envelope density varies. This is the origin of the light variations of WR 148, because electron scattering occurs in this envelope. Photometric studies by Antokhin (1984), Moffat & Shara (1986), and Marchenko et al. (1996) confirmed the light variations found by Bracher (1979) with an amplitude of 0.03 mag in V and also point to the very "noisy" appearence of the light curve. (With the ephemeris of Drissen et al. (1986), the light minimum occurs at phase zero with the WR star in front). Marchenko et al. (1996) noted the unusual wide-shaped light minimum, quite different from other known WR + O systems with atmospheric eclipses and a V-shaped light minimum (Lamontagne et al. 1996). For WR 148, Marchenko et al. suggested that the secondary light arises from an extended hot cavity in the WR envelope, near the companion, and which is ionized by X-rays. According to Marchenko et al., the rather weak X-ray source observed in WR 148 (Pollock et al. 1995) may be explained by the hot X-ray cavity being locally embedded in the WR envelope. Presently, the evolutionary status of WR 148 remains unclear and the companion could be either a B2-B4 III-V star or a relativistic object (as deduced from the mass function, Marchenko et al. 1996).

Our photometry is presented in Table 4 and the light curves are shown in Fig. 5, plotted with the ephemeris of Drissen et al. (1986). From Fig. 5 it is apparent that our light curves in 1993 are similar to the light curves published by Moffat & Shara (1986). The minimum occurs at phase zero. The 1994 light curves, however, show a remarkable change in their shape and mean light level. Random light variations, already noted in other works, could well contribute to the disturbance of the light curve shape, but it is unlikely that they would change the mean light. Furthermore, long-term changes in mean light appear to be correlated in U, B, and V (Fig. 5). Therefore, they too should be due to changes in continuum light.

Fig. 5. Light curves of WR 148 (data from Table 4

Table 4. Differential photometry of WR 148 (= HD 197406) - in the sense comparison star HD 197619 minus WR 148. Orbital phases are calculated with  d and JD 2432434.4 (Drissen et al. 1986).

There is a strong evidence for a long-term variation of the mean light. Although the time-base is too short, there are some indications that the long-term variation is periodic, possibly with a cycle of about 4 years. Marchenko et al. (1998b) point to a possible "overall brightening" of WR 148 in 1994 and 1995. As shown in Fig. 5, it is obvious that in 1993 the mean light was even some 0.05 mag higher, as in 1994. This long-term variation completely masks the short-term binary variations if the whole data set is depicted in one plot. Therefore we plotted the data separately for each year in Fig. 5.

Taking into account the model of Marchenko et al.(1996), the long-term light variations in WR 148 could be due to variations of the size of the hot X-ray cavity. Further conclusions at that time seem premature. A comment should be given on the observation at JD 2449585.4, phase = 0.57 (the companion in front), which strongly deviates from the regular light curve of 1994. As we can exclude observational errors as a reason for this measurement, it has to have some astrophysical origin. For instance, an event of accretion onto a compact companion could be invoked to explain this flare-like burst.

Flickering and flaring of WR 148 on different time-scales have been reported by Antokhin & Cherepashchuck (1989), Zhilyaev et al. (1995) and Khalack & Zhilyaev (1995). Matthews et al. (1992) looked for flares in the WR star EZ Canis Majoris (WR 6 = HD 50896, WN5) and reported one flare event. Flare-type activity of EZ CMa was also observed by Duijsens et al. (1996). This star is in many respects similar to WR 148, e.g. showing light variations with a 3.77 d period, long-term changes in the light curve, and a possible WR + c binary status (Firmani et al. 1980; Balona et al. 1989; Duijsens et al. 1996).

3.4. WR 153

WR 153 (= HD 211853 = GP Cep) is a quadruple system (Massey 1981) with orbital periods 6.6884 d (pair A, WR + O) and 3.4696 d (pair B, WR + O or O + O). Earlier spectroscopic studies by Hiltner (1945) and Bracher (1968) revealed radial velocity variations due to binary motion with a period of 6.68 d. Panov & Seggewiss (1990) reanalysed Hiltner`s velocity data and found evidence for two WR stars, one in each pair. WR 153 has been observed photometrically by Hjellming & Hiltner (1963), Stepien (1970), Moffat & Shara (1986), Panov & Seggewiss (1990), and Annuk (1994), all detecting eclipses with both periods, 6.6884 d and 3.47 d. Finally, Annuk (1994) refined the second period to 3.4696 d, in agreement with the velocity data of Massey (1981). However, in the recent analysis of WR star light curves by Lamontagne et al. (1996) the 3.47 d variation of pair B could not unambiguously be extracted from their data.

Our photometry of WR 153 is presented in Table 5 and the light curves are shown in Fig. 6a and b, with the 6.6884 d and 3.4696 d periods, respectively. From Fig. 6, our data are consistent with the ephemeris of Massey (1981) and Annuk (1994), respectively. Since the true shape of both light curves is unknown, no allowance is made for the 3.47 d period in Fig. 6a, where it is superimposed on the 6.69 d light variations. In Fig. 6b, the data points around the 6.69 d period minimum (at phases from 0.96 to 0.13 in Fig. 6a) have been removed.

Fig. 6. Light curves of WR 153 with a  d and b  d (data from Table 5)

Table 5. Photometry of WR 153 (= HD 211853). The comparison star is HD 211430 with , , and . Orbital phases "1" are calculated with  d and JD 2443690.32 (Massey 1981), orbital phases "2" with  d and JD 2443689.16 (Annuk 1994)

The light curve with the 6.69 d period (pair A) is probably due to an atmospheric eclipse (only one, V-shaped light minimum!). In pair B, two light minima are seen due to a core eclipse in that pair. Independent evidence can be obtained from the HIPPARCOS photometry data. We made a period search, using 122 data-points from HIPPARCOS. The analysis was performed with the PERIBM procedure, developed at the Astronomical Institute of the University of Vienna (latest version from: ftp://dsn.astro.univie.ac.at./pub/PERIOD98/current/ ). Our analysis clearly shows that there are peaks at d^-1, corresponding to a 1.735 d period, and at d^-1, corresponding to the 6.69 d period. The 1.735 d period is exactly 1/2 of the 3.47 d period and the reason that it shows up in the amplitude spectrum is because of the double-wave light curve (two eclipses in pair B), consistent with our ground-based photometry. Moffat & Shara (1986) also deduced that pair A had a single minimum at phase 0.00 ( d) and pair B had a double minimum at phases 0.00 and 0.50 ( d).

© European Southern Observatory (ESO) 2000

Online publication: March 9, 2000
helpdesk@link.springer.de