Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://jet.sao.ru/hq/lizm/conferences/pdf/2003/2003_p250.pdf Äàòà èçìåíåíèÿ: Wed Apr 21 16:48:55 2010 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 08:38:13 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: total solar eclipse

Magnetic stars, 2004, 250-258

Magnetic field connected fast line profile variability in spectra of bright O supergiants
Kholtygin A.1 , Brown J.2 , Fabrika S.3 and Surkov A.
1 2 3

3

Astronomical Institute of St.Petersburg University Glasgow University, UK Special Astrophysical Observatory of the Russian AS, Nizhnij Arkhyz 369167, Russia

Abstract. Results of study of fast line profile variability (lpv) in the sp ectra of selected bright O-stars are rep orted. A regular comp onent of lpv in the sp ectra of the star Ori A with estimated p erio d P 3 d have b een detected. We supp ose that the formation of long time-scale regular comp onents of lpv can b e explained in the framework of the magnetically confined wind-sho ck (MCWS) mo del of Bab el & Montmerle (1997a). In the context of testing the MCWS mo del the program of searching for weak magnetic fields in bright O and early B stars is outlined. The p ossibility of measuring weak longitudinal magnetic fields (B l 100 G) is demonstrated. Key words: stars: early typ e ­ stars: magnetic fields ­ line: profiles

1

Intro duction

Hot early-type stars have fast dense stellar winds driven by strong stellar radiation field in the lines and in the continuum. The winds of these stars are highly structured on both small (Eversberg et al. 1998) and large (Morel et al. 1998, Kaper et al. 1999, de Jong et al. 2001) scales. Several processes are attracted to explain the formation of these structures: wind instabilities, non-radial pulsations and co-rotation of largescale structures in the wind. The last can be explained with accepting a hypothesis that hot stars possess global dipolar magnetic fields (Babel & Montmerle 1997a,b). Nevertheless, in spite of the numerous attempts of detecting the magnetic fields of bright O and WR stars (de Jong et al. 2001, Donati et al. 2002, Chesneau & Moffat 2002, etc.), only for one star ( 1 Ori C) the measurements were successful (Donati et al. 2002). In this report we discuss the recent observations of hot stars in the light of magnetically confined windshock model of Babel & Montmerle (1997a). A review of attempts to reveal a magnetic field of O and early B stars is given. We also propose a program of searching for weak magnetic fields in such stars with using the large sequence of target stars with the Zeeman analyzer at the North Caucasus Special Astrophysical Observatory (SAO) 1 m telescope.

2
2.1

Observations and data reduction
Program of observations

Recently Kholtygin et al. (2003a) have proposed a program of spectral observations of bright O and early B stars for searching and analysing the fast line profile variations in their optical spectra with large signal-tonoise ratio and a high time resolution (5­30m). The list of program stars is given in Table 1.

c Sp ecial Astrophysical Observatory of the Russian AS, 2004


MAGNETIC FIELD CONNECTED FAST LINE PROFILE...

251

Table 1: Program stars list HD 3360 24212 24760 30614 36486 36861 37742 47839 91316 120315 156633 160762 163472 166182 180968 203064 205021 209975 210839 214680
1

Name Cas Per Per Cam Ori Ori A Ori A 15 Mon Leo Uma 68 Her 85 Her V2052 Oph 102 Her 2 Vul 68 Cyg Cep 19 Cep Cep 10 Lac

Sp.Type B2IV O7.5III B0.5IV O9.5I O9.5II O8III O8III O7Ve B1Iab B3V B1.5Vp+ B3IV B2IV-V B2IV B0.5IV O8e B2IIIevar O9.5I O6Iab O9V

V 3.67 4.04 2.90 4.29 2.23 3.66 1.79 4.66 3.84 1.85 4.80 3.79 5.83 4.35 5.47 5.04 3.22 5.11 5.09 4.87

V

1

V

rot

2330 1590 2060 2175 1860 2110 1110

-

2340 800 2010 2300 1140

sin i 17 213 134 129 144 74 124 67 75 108 127 105 127 84 84 115 27 95 219 35

2

2

from Kaper et al. (1997) data are taken from SIMBAD database (http://simbad.u-strasbg.fr/Simbad/) along with mean values from Abt et al. 2002 (marked with asterisks)

2.2

The data and their reduction

Spectra of program stars were obtained at the SAO 1.0 m telescope with CEGS spectrograph. The instrument configuration was described by Musaev (1996). Observations in 2001 were made with a 1242 â 1152 Wright Instruments CCD detector and those in 2003 -- with a 2048 â 2048 CCD detector. ° The grating was used in 49 ­ 122 orders to produce spectra covering 6000 A centered on H with a resolution of R = 45000 (0.08 ° A/pixel near H ). The ratio S/N from 200 to 400 per pixel in continuum in each spectrum (depending on the weather conditions and the instrumental configuration) was achieved for the target stars presented in Table 2. Reduction of data was made with standard MIDAS procedures. Flatfielding was done with using an image of the fast rotating star Leo (V sin i = 329 km/s) as a quasi-flat field. It appeared that the pixel inhomogeneity remaining after this procedure does not exceed 0.3% in continuum units (see Monin et al. 2002 for details). All spectra were normalized to the continuum level by means of an appropriate procedure.

3

Fast line profile variations

The profiles of most lines in the spectra of studied stars seem to be variable. This illustrates the set of the residual spectra (individual minus mean) of 19 Cep near H and CIII 4650 plotted in Fig. 1 a-d. It is easy to see the variable detail of the H line profile at the velocity 90 km/s from the line center. The similar details are revealed in the profiles of the lines H , and strong He lines in the spectra of the star (Kholtygin et al. 2003 a,b). On the other hand, the line CIII 4650 (Fig.1 c-d) in the spectra of the same star shows only the marginal line profile variations (lpv).


252

KHOLTYGIN, BROWN, FABRIKA, SURKOV

Figure 1: a: The residual spectra near H for 19 Cep (order 98). b: the same as in a, but al l spectra were smoothed with gauss filter (W = 0.2° c, d: -- the same as in a, b, but for order 102 (including CIII 4650 A), line). The time interval between the successive spectra is about 30 min. Time grows upward.


MAGNETIC FIELD CONNECTED FAST LINE PROFILE...

253

Table 2: List of stars observed in 2001 and 2003 September 4­7, 2001 texp . Number Total time (min.) exp. of observ. (h) Per 10 4 0.7 Cam 10 11 1.8 19 Cep 15 22 6.0 10 Lac 10/15 39 10.5 November 29 ­ December 4, 2001 Per 5 5 0.4 Cam 10 6 1.0 Ori A 10 75 12.5 Ori A 2 36 1.2 10 Lac 15 27 6.7 January 22­23, 2003 Cam 7/10 16 2.6 Leo 5 7 0.6 Ob ject

3.1
3.1.1

The cyclical comp onents of the line profile variability
Perio d Search

To investigate the time variations of the line profile, we make the modified CLEAN (see Roberts 1987 and Kholtygin & Shneiwais 2003) analysis of the residual line profile variations S () = I () - I mean for H, He and CIII lines in spectra of O8 star Ori A. For illustration we present in Fig. 2 a grey-scale plot of the Fourier power spectra for CIII 5696 lpv. The higher detected frequency 1 1.3d-1 gives the period of time variation P 18h. This period is intermediate between the typical one of non-radial pulsation (NRP) and the typical of O stars rotational periods. The nature of this Fourier component is unclear. It should be mention that the determined frequency is very close to the "dangerous" frequency 1.4d-1 , which can not be detected for a given set of moments of observations (see Kholtygin & Shneiwais 2003), so the reality of this component is not evident. The lower frequency 2 0.35d-1 (P 3d) which is very close to one half of the period 6.1­6.3 d, suspected by Kaper et al. (1997) from analysis of the UV spectra of this star. The close frequencies are found from our Fourier analysis of the line profile variations of H , HeI 4713 and HeII 4686.

3.2

Mo dels of cyclical comp onents of the line profile variability

The large time scale line profile variations are often explained via formation of the large-scale structures in the stellar wind. These structures are connected with corotating interaction regions (CIR, Cranmer & Owocki 1996) resulting from a localized "bright spot" on the stellar surface. These CIRs are thought to produce the cyclical modulation of the P-Cygni absorptions in optical and UV lines (e.g. Kaper et al. 1999, de Jong et al. 2001). Nevertheless, the formation of many variable lines occurs close to the stellar surface so that azimuthal extension of the line formation region due to bending of the CIRs structure must remain small. Moreover, the occultation effect for CIRs must be observed in the case of the multiple CIRs in the wind. The detected by us cyclical variability of the emission line CIII 5696 (see Fig. 2) hardly agrees with the CIR model. Alternative explanation of the cyclical line profile variations can be obtained in the framework of "confined corotating wind" model. In this model a star is an oblique magnetic rotator (see, for example, Fig. 15 in Rauw et al. 2001), the wind is confined in latitude and its symmetry axis is the tilted magnetic axis of the star. Such a model has been proposed to explain the lpv observed in the spectra of P up (Moffat & Michaud 1981) and 1 Ori C (Stahl et al. 1996).


254

KHOLTYGIN, BROWN, FABRIKA, SURKOV

nu,d 2

-1

Lambda Ori A, CIII5696

1.5

1

0.5

0 -300

-200

-100

0

100

200

300

V,km s-1

Figure 2: The CLEAN Fourier spectrum for CIII 5696 line profile variations in the spectra of Ori A. Only the points above the significance level = 0.999 for a hypothesis on a presence of the periodical components in the lpv are plotted.

4
4.1

Magnetic fields in O stars
Magnetically confined wind-sho ck mo del

The "confined corotating wind" model described in the previous section was developed by Babel & Montmerle (1997a) as a magnetically confined wind-shock (MCWS) model. In this model the wind streams from both magnetic hemispheres collide with each other and produce a strong shock, an extended X-ray-emitting postshock region and a thin dense cooling disc in the magnetic equatorial plane. This model can explain the rotational modulation of the X-ray luminosity due to a partial eclipse of the magnetosphere effect by the cooling disc. In support of this X-ray flux modulation we can mention a correlation between X-ray fluxes and optical line profile variability (see discussion in Kholtygin et al. 2003b). The crucial point for explanation of the line profile variations in the framework of "confined corotating wind" or MCWS model is the presence of a magnetic field at the surface of the star. The necessary values of the field are not very large. Babel & Montmerle (1997b) found that a magnetic field (B 270 - 370 G) can confine a significant fraction of the wind of 1 Ori C into a "circumstellar cooling disk" located in the plane of the magnetic equator. Waldron & Cassinelli (2000), based on their own X-ray observations of O9.7 star Ori, have estimated that at the value ne = 1012 cm-3 , typical of this star, and the temperature of the hot gas emitting in the X-ray lines in stellar wind, TX 5 â 107 K, obtained by them, the surface magnetic field strength can be estimated to be 180 G assuming a balance between gas and magnetic pressures. Parameters of this hot dense plasma revealed by Waldron & Cassinelli (2000) in the wind of Ori are comparable with those for solar flares. This confirms the reality of MCWS model for this star. Schulz et al. (2001) also suggested that the very high temperature of the hot gas in the wind of 1 Ori C (up to 6 â 107 K) obtained by them from the X-ray line intensity ratios of this star can be explained in MCWS model.

4.2

Searching for magnetic fields of O and early B stars

There exist some different techniques to determine the stellar magnetic field. One of the most popular methods is measurement of the line shift between the left and right circular components of spectral lines (detecting the first moment of Stokes V parameter) (e.g. Brown & Landstreet 1981, Monin 1999, Monin et al. 2002). This method is sensitive only to the line-of-sight component of the disc-integrated vector field. This technique obey to measure the magnetic fields from 102 G. The so-called Robinson technique (Robinson 1980) is based on measurements of the differential Zeeman broadening between magnetically sensitive and intensive spectral lines. Nevertheless, this technique has large systematic errors. A very effective methods of determination of


MAGNETIC FIELD CONNECTED FAST LINE PROFILE...

255

the field structure is the Zeeman­Doppler imaging (e.g. Donati et al. 1997). The brightest O star Puppis (O4I(n)f ) is the most attractive target in searching for the magnetic field. Barker (1981) tried to measure the longitudinal Zeeman effect in the wings of the H line with a photoelectric Pockel cell polarimeter. The obtained mean value of Bl = -44 ± 105 G means that the field is too small to be detected in this experiment. Recently Chessneau & Moffat (2002) have reported a new attempt to detect the magnetic field in Puppis. The disc-averaged value of the longitudinal field component was estimated using line profile integration of continuum normalized Stokes V parameter for four lines HeII and NIV. During four observational nights the following values were obtained: B l = -220 ± 225, 88 ± 359, -236 ± 201 and 143 ± 215 G, respectively. It means no detection of the magnetic field. Moreover, no short-time scale variability of the Stokes V parameter in the investigated line profile were detected within a 0.1% level of noise per resolution element. De Jong et al. (2001) have reported an attempt to measure the longitudinal component of the field for the bright O7.5 star Persei. The average value of all measurements is 27 ± 70G. The error bars of these measurements appeared to be too large to detect a field. Recently Donati et al. (2002) have informed on the longitudinal field (B l ) estimations for O6 star 1 Ori C obtained by measuring the first moment of the Stokes V profile, normalized by the line equivalent width, with using the least-squares deconvolution technique (Donati et al. 1997). The derived B l values are equal to 357 ± 46 G, 37 ± 35 G, 257 ± 49 G, 191 ± 63 G and 257 ± 73 G for rotational phases 0.033, 0.621, 0.789, 0.920 and 0.177, respectively. The phase dependence of B l can be cosine fitted with the mean value B0 = 172 ± 30 G. The modeling of the Stokes V parameter variations indicate that the obtained B l values are consistent with the assumption that the magnetic field can be presented as a tilted dipole with the polar field strength Bp = 1.1 ± 1.1 kG inclined at 42 ± 6 with respect to the rotation axis, which, in turn, is assumed to be inclined at 45 to the line of sight. The attempts to determine the magnetic fields for the selected early B stars had no results (Monin et al. 2002). Recently Donati et al. (2001) have detected a moderate variable magnetic field of |B l | < 100 G and with a polar strength Bp 360 ± 30 G in the bright B1 star Cephei. A weak varying longitudinal magnetic field with a strength between 10 G and -46 G, corresponding to Bp 335 G, was detected in the B2IV star Cas (Neiner at al. 2003). MacGregor & Cassinelly (2003) have discussed the possibility of generation of magnetic fields in hot stars. They found that fields in hot main-sequence stars generated by dynamo mechanism at the interface between the radiative core and convective envelope of the star could rise to the surface and reach the values necessary for the wind confining.

4.3

Program of detection of magnetic fields

It is clear from the previous section that the moderate values of B l 100 - 200 G and even smaller for O and early B stars are expected. This means a necessity for the procedure of measuring the fields of this value for O and B stars. For this purpose, we propose a program of searching for the magnetic fields of bright O and early B stars, presented in Table 1. As the main method of determination of the magnetic field, we plan to use the simple "photographic technique" (see subsection 4.2) as a procedure most suitable for numerical targets. Using this procedure, the mean longitudinal magnetic field B l of the star is determined via the wavelength shift between the right and left circular polarization components of the line (e.g. Monin 1999): = R - L = 2 k0 g B l 2 , 0 (1)

where R and L are the wavelengths of the centre of gravity of the right and left circularly polarized components of the line, respectively, g is the effective Land´ factor of the line, 0 is the rest wavelengths of e -1 the line, and constant k0 = 4.67 â 10-13 ° G-1 . A The value of 1 R = r d , (2) W where W is the line equivalent width, r is the residual intensity of the line at wavelength . A similar expression can be written for the value of L .


256

KHOLTYGIN, BROWN, FABRIKA, SURKOV

Figure 3: Left panel: Mean longitudinal magnetic field strength Bl as a function of rotational phase for me values Bp = 1100 G and i = 45 . Right panel: Phase averaged longitudinal magnetic field strength B l an versus the angle between rotational and magnetic axes. The required mean longitudinal magnetic fields B l is the line intensity weighted mean over the visible stellar disk (Eversberg 1997): 1 Bl = W
2 /2

d
0 0

Bl cos() sin()d â

r (, )d .

(3)

Here Bl = Bl (, ) is the line of sight component of the magnetic field, and are the coordinates of the point on the stellar disk and r (, ) is the residual intensity of the line in this point. For a tilted dipole magnetic field and the linear law of limb darkening r (, ) = 1 - u + u cos() (u is the parameter of limb darkening for the wavelength considered) the variation of B l with rotational phase can be obtained by integration Eq.3. Finally (e.g., see Preston 1967 for details): Bl = B
p

15 + u [cos cos i + sin sin i cos 2 ( - 0 )] , 20(3 - u)

(4)

where Bp is the polar magnetic field strength, is the angle between magnetic and rotational axes, i is the rotational axis inclination angle and 0 is the phase of the maximal longitudinal field. The mean longitudinal magnetic field strength Bl versus the rotation phase is plotted in Fig. 3 (left panel). We accept the values Bp = 1100 G for the polar field strength and i = 45 for the inclination angle which have been obtained by Donati et al. (2002) for 1 Ori C. From this figure we see that the maximal Bl value never exceed 1/3 of the Bp value. mean are plotted in Fig. 3 (right panel). For almost all values of the angle The phase average values of B l mean and rotational axis inclinations i the phase averaged field B l does not exceed 100 G. By this means, for detecting the magnetic fields in O stars we have to measure very small lambda differences between the right and left circularly polarized components of the lines. For example, for the strong HeI 7065 line (effective Gaunt factor g = 2) the value of 0.01 ° With the parameter of A. the instrument intended to be used in searching for the magnetic field (see subsection 2.2) we are unable to determine the value of with the necessary accuracy from a single line. The accuracy of determination can be largely improved when a large number of lines are used. In this case we have to use for each line i its weight i (rL + rR )/2, where rL and rR are residual intensities of left and right circularly polarized components of line at their center of gravity (Monin 1999). Then the effective number of lines used for determinations can be estimated as Nline = n â
n i=1

w

i



,

o

where o is the weight for some reference line (e.g. HeI 7065).


MAGNETIC FIELD CONNECTED FAST LINE PROFILE...

257

Figure 4: Left panel: Mean effective magnetic field Beff as a function of input mean longitudinal field Bl for signal-to-noise ratio S/N=300 and Neff =200. Right panel: Dependence of mean effective magnetic field Beff versus the effective number of observations Nef f for S/N=500. The input value of B l = 100 G. The error bars (at 3 level) and mean values of Beff are plotted.

However, the number of the suitable lines in the spectra of O stars is not very large (about 30­50) and to increase the gain in the line displacement measurements, we can use a number of successive spectra N sp , obtained during the observational run. Looking at Fig. 3 we can see that it is possible to effectively use not more then 1/4 of the total rotational period, when changes of the value of B l are not too large. From the above reasoning, it is clear that the stars with the maximal rotational periods are the most suitable targets. In our list of program stars (Table 1) only two stars (19 Cep and Ori A) have rather large rotational periods ( 6 d). For these stars we can use spectra of 2 observational nights. This gives a total of Nsp 2 â 30 60 spectra. Thus, the effective number of lines which can be used for measurement is Neff = Nline â Nsp . Substituting in this expression the typical values Nline 50 and Nsp 60 we have a total of Neff 3000. To estimate the possibility of measuring weak fields using the above described procedure, we have made a numerical experiment to reproduce the parameters of real observational runs (see subsection 2.2). For a better determination of the line center of gravity we have used the Gauss fit of the line profile. We also suppose that all instrumental correction to the values are made. Supposing that for a given line i and for an observational run j the value of = ij is measured, we can estimate the mean longitudinal magnetic field via the following relation: B
(ij ) (ij ) l

= 0.5 â (k0 g2 ) 0

-1

.

(5)

Averaging the obtained B l values over all possible values of i and j , we can determine the effective longitudinal magnetic field Beff and its standard deviation by usual way. The resulting dependence of Beff on the input value of B l is presented in Fig. 4 (left panel). We can see that for a not very large number of Neff values the scattering of Beff is too large, and for the low values of B l the mean effective magnetic field appeared to be overestimated. A similar effect also exists for weak line intensity determination with A/N < 6, where A is the maximal line intensity (see Rola & Pelat 1994 for details). In this work it is shown that in this case the measured line fluxes have not normal, but log.­normal distribution with displacement to the larger fluxes. The nature of this effect is very simple: if the observed line flux (real flux plus noise) appears to be smaller than the noise level, we do not detect it and do not include this measurement in the total set of line profile measurements. In our case we measure the line shifts rather than fluxes, but the nature of measurements seems to be very close. Fig. 4 (right panel) demonstrates the possibility of improving the accuracy of determination of the B eff value with increasing the effective number of observations Nef f . For the large values of Nef f the weak input field B l = 100 G can be detected at 3 level.


258

KHOLTYGIN, BROWN, FABRIKA, SURKOV

5

Conclusion

We report the results of a study of the fast lpv in spectra of selected bright O stars. The regular long time-scale components of lpv in the spectra of the star Ori A with the estimated period P 3 d have been detected. The formation of such components of lpv can be explained in the framework of the MCWS model of Babel & Montmerle (1997a). A program of detection of the weak magnetic fields in bright O and early B stars with the Zeeman analyzer at the North Caucasus Special Astrophysical Observatory (SAO) 1 m telescope is proposed. We demonstrate a principal possibility of measuring weak magnetic fields (B l = 100 - 200 G) for the program stars.
Acknowledgements. The work was supported by grants of NATO CLG 6978036 and RFBR 01-02-16858.

References
Abt H.A., Levato H., Gross M., 2002, Astrophys. J., 573, 359 Brown D.N., Landstreet J.D., 1981, Astrophys. J., 246, 899 Brown J.C., 1994, In: Moffat A.F.J., St-Lois N. (eds.) Quebec Workshop in Instability and Variability in Hot Star Winds. Ap&SS, 221, 357 Babel J., Montmerle T., 1997a, Astron. Astrophys., 323, 121 Babel J., Montmerle T., 1997b, Astrophys. J., 485, L29 Chessneau O., Moffat A.F.J., 2002, Publ. Astr. Soc. Pacific, 114, 112 Cranmer S.R., Owocki S.P., 1996, Astrophys. J., 462, 469 Donati J.-F., Semel M., Carter B.D., Rees D.E., Cameron A.V., 1997, Mon. Not. R. Astron. Soc., 291, 658 Donati J.-F., Wade G.A., Babel. J., Henrichs H.F., de Jong J.A., Harries T.J., 2001, Mon. Not. R. Astron. Soc., 326, 1265 Donati J.-F., Babel J., Harries T.J., Howarth I.D., Petit P., Semel M., 2002, Mon. Not. R. Astron. Soc., 333, 55 Eversberg T., 1997, Universit´ de Montr´ Thesis e eal, de Jong J.A., Henrichs H.F., Schrijvers S., Gies D.R., Telting J.H., Kaper L. and Zwarthoed G.A.A., 1999, Astron. Astrophys., 345, 172 de Jong J.A., Henrichs H.F., Kaper L., Nichols J.S., Bjorkman K. et al., 2001, Astron. Astrophys., 368, 601 Kaper L., Henrichs H.F., Fullerton A.W., Ando H. et al., 1997, Astron. Astrophys., 327, 281 Kaper L., Henrichs H.F., Nichols J.S., Telting J.H. et al., 1999, Astron. Astrophys., 344, 231 Kholtygin A.F., Monin D.N., Surkov A.E., Fabrika S.N., 2003b, Astron. Lett., 29, 177 Kholtygin A.F, Brown J.C. , Cassinelli J.P., Fabrika S.N., Monin D.N., Surkov A.E., 2003a, Astron.& Astrophys. Transact., 22, 499 Kholtygin A.F., Shneiwais A.B., 2003, Astrophysics, in press Eversberg T., L´ epine S., Moffat A.F.J., 1998, Astron. Astrophys., 494, 799 MacGregor K.B., Cassinelly J.P., 2003, Astrophys. J., 586, 480 Mathys G., 1991, Astron. Astrophys. Suppl. Ser., 89, 121 Morel et al., 1998, Astrophys. J., 498, 413 Monin D.N., 1999, Bull. Spec. Astrophys. Obs., 48, 121 Monin D.N., Fabrika S.N., Valyavin G.G., 2002, Astron. Astrophys., 396, 131 Moffat A.F.J., Michaud G., 1981, Astrophys. J., 251, 133 Neiner C., Geers V.C., Henrichs H. F., Floquet M., Fremat Y., Hubert A.-M., Preuss O., Wiersema K., 2003, Astron. Astrophys., 406, 1019 Preston G.W., 1967, Astrophys. J., 150, 547 Rauw G., Morrison D.M., Vreux E.G., Gosset E., Mullis C.L., 2001, Astron. Astrophys., 366, 585 Rola C., Pelat D., 1994, Astron. Astrophys., 287, 677 Roberts D.H., Lehar J., Dreher J.W., 1987, Astron. J, 93, 968 Robinson R.D., 1980, Astrophys. J., 239, 961 Schulz N.S., Canizares C.P., Huenemoerder D., Kastner J.H., Taylor S.C., Bergstrom E.J., 2001, Astrophys. J., 549, 441 Stahl O., Kaufer A., Rivinius T. et al., 2000, Astron. Astrophys., 312, 539 Waldron W.L., Cassinelli J.P., 2000, Astrophys. J., 548, L45