A&A 376, 1031-1038 (2001)
J. Greiner1 - G. Tovmassian2 - M. Orio3,4 - H. Lehmann5 - V. Chavushyan6 - A. Rau1 - R. Schwarz1 - R. Casalegno3,4 - R.-D. Scholz1
1 - Astrophysical Institute Potsdam, An der Sternwarte 16, 14482 Potsdam, Germany
2 - OAN, Instituto de Astronomía, UNAM, AP 877, 22860 Ensenada, México
3 - Osservatorio Astronomico di Torino, Strada Osservatorio 20, 10125 Pino Torinese (TO), Italy
4 - Dept. of Astronomy, Univ. Wisconsin, 1150 University Ave., Madison, WI 53706, USA
5 - Thüringer Landessternwarte, Sternwarte 5, 07778 Tautenburg, Germany
6 - Instituto Nacional de Astrofisica, Optica y Electronica (INAOE), Aptdo. Postal 51 y 216,
72000 Puebla, Pue., Mexico
Received 12 April 2001 / Accepted 16 July 2001
We report optical observations of the VY Scl star BZ Cam during its previous optical low state in 1999/2000. We find drastic variations in the line profiles. Narrow-band imaging observations show that its nebula extends farther than previously known and seems to be composed of two components. We determine the [O III] line intensity of BZ Cam's nebula to erg/cm2/s. We discover a proper motion of BZ Cam of mas/yr which together with the systemic radial velocity yields a space velocity of 125 kms-1. We re-interpret the nebula as being photo-ionized by hypothesized transient, luminous, supersoft X-ray emission during optical low states, and shaped by the transverse motion of BZ Cam.
Key words: X-ray: stars - binaries: close - reflection nebulae - stars: individual: BZ Cam - 0513 - CAL 83
BZ Cam is a binary system with a period of 221 min (Patterson et al. 1996). It is thought to contain an accreting white dwarf and a 0.3-0.4 main-sequence donor (Lu & Hutchings 1985). BZ Cam belongs to the group of variable stars called VY Scl stars, or anti-dwarf novae, due to its occasional drop in brightness. Most of the time it is observed at around V = 12.0-12.5 mag (0.2 mag), but during low states it is as low as V = 14.3 mag. Only one previous optical low state is known, which occurred in 1928 (Garnavich & Szkody 1988).
BZ Cam is surrounded by a faint emission nebula (Ellis et al. 1984) which has a bow-shock-like structure (Krautter et al. 1987; Hollis et al. 1992). This nebula is also detected at radio frequencies (Hollis et al. 1992), implying a 35 cm-3 density in the H II recombination region (assuming an electron temperature of 104 K). Based on the optical emission line ratios these authors argue that photoionization can not alone account for the excitation of the nebula, and that shock wave heating seems to contribute. Based on IUE data BZ Cam was found to exhibit a wind (Hollis et al. 1992), which was recently also detected in the optical as well as to display rapid variability (Ringwald & Naylor 1998), rare among canonical CVs.
Supersoft X-ray binaries (SSB) were established as new class of astronomical objects during the early 90ies (Greiner 2000). They contain a white dwarf (WD), accreting mass at rates high enough to allow quasi-steady nuclear surface burning (van den Heuvel et al. 1992). Their luminosities are of order of -1038 ergs-1, with typical temperatures of 30-50 eV. Two SSBs have particular properties which are worth mentioning in this context: (i) CAL 83: it is the only SSB (among a dozen SSBs searched) which is surrounded by a distinct nebula (Pakull & Motch 1989; Remillard et al. 1995) caused by ionization of the surrounding interstellar medium by the luminous X-ray radiation (Rappaport 1994). (ii) 0513: it shows quasi-periodic optical low states (Southwell et al. 1996) during which supersoft X-ray emission is "on'' (Schaeidt et al. 1993). Among the "classical'' CVs two systems have recently been shown to exhibit transient, supersoft X-ray emission, both during periods of optical low states: the VY Scl star V751 Cyg (Greiner et al. 1999) and V Sge (Greiner & Teeseling 1998).
BZ Cam recently entered a very rare optical low state.
Figure 1 shows the optical light curve of BZ Cam over the last
2 years covering the recent optical low state.
|Figure 1: Optical light curve of BZ Cam with most of the data taken from VSNET. The times of optical spectroscopy and narrow-filter imaging are marked by vertical dotted lines.|
|Open with DEXTER|
We report here on our observations triggered by this rare optical low state.
We acquired photometric and spectroscopic observations (see Table 1
for a log) in the optical low state,
|INAOE 2.1 m||Jan. 9-12||4000-7500||1.0(3)||1500|
|AIP 0.7 m||Jan. 21||white||1.7||10|
|AIP 0.7 m||Feb. 02||V||0.5||30|
|Tbg 2.0 m||Feb. 27||3500-9500||1.9||1800|
|INAOE 2.1 m||Feb. 26-28||4000-7500||1.0(3)||1800|
|OAN 1.0 m||Mar. 04||V/R||0.5||60|
|INAOE 2.1 m||May 05||4000-7500||4.0||1500|
|Tbg 2.0 m||Jun. 7/8||4000-8500||1.9||1800|
|WIYN 3.5 m||Sep. 28||[OIII], H||-||600|
Low resolution spectra of BZCam were obtained at the 2.1 m telescope at Cananea, México in winter of 2000. The LFOSC spectrograph (Zickgraf et al. 1997) was deployed in mid-January (8-11) observations to cover the 4000-7500 Å wavelength range with 13 Å FWHM resolution. A 3 arcsec wide long slit, as projected on the sky, was used at the entrance of this multi-object spectrograph. Later in the month (27-29 January) the object was again observed using the same telescope, but with the B&Ch spectrograph instead of the LFOSC. The resolution was slightly better than 10 Å, and the coverage of wavelengths was about the same. Again, the B&Ch spectrograph with the same telescope and same settings was used in May, when BZ Cam was back to its high state, to acquire a few spectra. Spectrophotometric standard stars were observed at each night in order to provide flux calibration. A He-Ar arc lamp was utilized in order to calibrate the spectra for wavelengths. IRAF standard procedures of long slit spectroscopy were used for data processing.
The photometric measurements of the object in the V and R bands were done at the 1.0 m telescope of OAN in Tonantzintla, México on March 4.
Observations with the 3.5 m WIYN telescope at Kitt Peak (USA), operated by the University of Wisconsin, Indiana University, Yale University, and the National Optical Astronomy Observatories, were done using a mosaic of four 2K2K CCDs.
Spectroscopic observations at Tautenburg were done using the newly developed spectrograph for the Nasmyth-focus of the 2.0 m Schmidt telescope. Grisms with 200 Å/mm and 100 Å/mm were used, respectively. Reduction of the spectra was performed using standard MIDAS routines from the long-slit package. Wavelength calibration was done using night sky lines.
Photometric observations at the 0.7 m telescope at AIP employed a 1K1K TEK CCD camera. In the Cassegrain focus the 24 m pixel size corresponds to a plate scale of 0 5/pixel, completely sufficient for the typically bad seeing (2-4 ). Initial basic reduction was done using standard MIDAS programmes, while the photometry was done with DOPHOT (Mateo & Schechter 1989).
Sample spectra are shown in Fig. 2,
|Figure 2: Top: BZ Cam spectra in 2000: low state (Jan. 10+11), beginning of the rise out of the low state (Jan. 26) and high-state (May 5). The flux is given in units of 10-14 erg/cm2/s/Å. Note the different slopes and emission line strengths. Bottom: blow-up of rectified spectra of the high, intermediate and low state (from top to bottom). P Cygni profiles are present only during the high-state, while in the intermediate state the emission lines appear on top of broad absorption lines.|
|Open with DEXTER|
|Spectral Line||Relative Flux|
see e.g. Ringwald & Naylor 1998, so we do not present emission line ratios). Several spectral differences are immediately recognized from the examination of the spectra in different brightness states:
The non-canonical intensity ratios of the Balmer emission lines during the optical low state (Table 2) is most probably due to a Balmer absorption system which is obvious during the intermediate state, but possibly has not completely vanished during the other states. We mention here that it is a general feature in short-period SSBs that e.g. H and H are clearly seen in absorption while H and H are progressively filled. Certainly, this effect is much more drastic in SSBs, but it could potentially serve as an explanation of the emission line ratios in BZ Cam as well.
Another noteworthy property is the rather small intensity of the He II emission. This is in contrast to the fact that in SSB the He II emission line is usually the strongest line, or at least stronger than H. However, this can be understood in terms of different temperatures of the white dwarf and correspondingly different ionizing flux for He II. The mass of the white dwarf in BZ Cam is certainly smaller than 1 , and more probably in the range of 0.4-0.7 (Lu & Hutchings 1985). In contrast, white dwarf masses in SSB are thought to be 1 , and consequently the effective temperatures are higher for SSB white dwarfs as compared to BZ Cam. While this difference has little effect on the number of ionizing photons for hydrogen (<912 Å), it has a drastic effect on the number of ionizing photons for He II (<228 Å).
In order to estimate the proper motion of BZ Cam, we looked for
Digitized Sky Survey (DSS) data using the plate finder service at the
Space Telescope Science Institute. On the 8 Palomar Schmidt plates
with a time baseline of more than 40 years (see Table 3)
found in the DSS we measured the position of BZ Cam using the
plate constants provided with the FITS images and the ESO Skycat tool.
Searching the VizieR database in Strasbourg for more independent measurements of the position of BZ Cam, we were lucky to find an early epoch from the Astrographic Catalogue AC2000 (Urban et al. 1997; Urban et al. 1998). In addition, there are two other positions from the HST archive and from the FONAC catalogue (Kislyuk et al. 1999), both with similar epochs as with the POSS2 data (see Table 3).
Combining the POSS measurements with the AC2000 position the resulting proper motion becomes more accurate in both directions, with reaching Hipparcos-like accuracy. The error in remains twice as large, mainly due to the very different positions measured on the POSS1 plates. When the additional positions from the HST and FONAC are included, the errors of remain at the same level whereas those of increase, especially after including the FONAC position. As can be seen from Table 3, the value from FONAC is a clear outlayer compared to all other values. This is also the reason, why the proper motion of BZ Cam given in the FONAC catalogue (as determined from only two positions - from the Astrographic Catalogue and from one second epoch plate measurement of the Kiev wide-angle astrograph) is distinctly different in (see also Table 4). The errors of the FONAC proper motion correspond to the catalogue precision given in Kislyuk et al. (1999).
For our further analysis we adopted the proper motion solution obtained from 8 POSS plates and the AC2000: mas/yr and mas/yr. This solution is in good agreement with other solutions (except the FONAC ) and shows the smallest total proper motion error.
|Figure 3: The proper motion of BZ Cam obtained from measurements over a 100 years time baseline. The AC2000 position from 1897 and 8 positions as measured on POSS plates between 1953 and 1997 are shown by dots. The lines represent the proper motion fit using these data. The data not used in this finally adopted proper motion solution are shown by other symbols: FONAC () and HST (+) (see also Tables 3 and 4).|
|Open with DEXTER|
At a distance of pc (Naylor & Ringwald 1998) this corresponds to a transverse velocity of km (solely towards the South), where most of the uncertainty stems from the distance error. Adding a systemic radial velocity component of about -75 km s-1 (Lu & Hutchings 1985; Patterson et al. 1996) we derive a space velocity of BZ Cam of 125 km s-1.
We also obtained narrow-band filter images in H and the [O III] 4959/5007 Å emission lines, showing the large nebula around BZ Cam (Krautter et al. 1987; Hollis et al. 1992). Based on our higher sensitivity we find that the nebula seems to be composed of two components: (i) a large, smooth component with bow-shock like shape towards the South, and a (ii) smaller, filamentary component which also has a bow-shock like shape towards the south, but with a smaller curvature radius, and in addition with cometary-tail like extensions towards the North-East and North-West (Fig. 4). A nebula spectrum around the [O III] line is displayed in Fig. 5. In certain regions of the nebula the [O III] line is very strong relative to H, whereas in other regions it is not. Figure 6 shows these relative intensity variations for a few line ratios across the nebula. One prominent example is the spectral difference of the two arcs marked as "A'' and "B'' in the right panel of Fig. 4. While both arcs are similarly bright in H, only "A'' is bright in [O III]. This argues against shock-excitation, but is consistent with photo-ionization from BZ Cam, since "B'' is at a much larger distance from BZ Cam.
We have estimated the total nebular flux in [O III] to be erg/cm2/s (after a 16% extinction correction when using E(B-V)=0.05 from Verbunt 1987). At a distance of 830 pc (Ringwald & Naylor 1998) this corresponds to erg s-1.
|Figure 4: The BZ Cam nebula in [O III] 5007 (left) and H 6563 (right), obtained with the WIYN telescope at 0 8 seeing. The orientation of the slit during the Tautenburg spectrographic observations, from which the line ratios have been derived (see Fig. 6) is marked in the left frame. North is up and East to the left.|
|Open with DEXTER|
We have earlier speculated that based on the behaviour of the SSB 0513 and the VY Scl star 7 possibly also other VY Scl stars could be emitters of supersoft X-ray emission during the optical low-state (Greiner et al. 1999). In the case of BZ Cam, the similarity in the wind properties with VSge adds even more support to this conjecture. We will argue in the following that by assuming supersoft X-ray emission during the optical low-states we can explain the hitherto puzzling properties of BZ Cam's nebula.
We first note that
all measured nebular emission line ratios ([OIII], [NII], [SII] vs.
of the nebulae surrounding BZCam (our own, Table 5,
|Figure 5: Spectrum of the BZ Cam nebula in the region of the [O III] line, plotted for three different regions in the nebula: Region "A'' (solid line), a region at the same position angle as "A'' but 4 times further away from BZ Cam (dashed line), and region "B'' (dotted line).|
|Open with DEXTER|
To make this expectation somewhat more quantitative, we used the XSTAR 2.0 code (Kallman 2000) as distributed in the HEAsoft package, and computed the emission line luminosities over a grid of the following input parameters: (i) hydrogen density n = 35 and 100 cm-3; (ii) temperature of the central, ionizing source T = 15 and 25 eV. The following parameters were adopted and not varied: luminosity of the central source erg s-1; outer radius (in [O III]) of the nebula pc. We find that the luminosity of the [O III] 5007 line is in the range of 0.1-2% of the X-ray luminosity. Thus, based on our measured [O III] flux, the mean (over the last 10000 yrs) ionizing (X-ray) luminosity would be - erg/s.
|Figure 6: Relative line ratios R across the BZ Cam nebula along the direction shown in Fig. 4. The central part has been omitted (BZ Cam itself). Negative values in the abscissa correspond to the North-western part of the slit in Fig. 4, positive values to the South-eastern part. The letters denote the following ratios: a) R = 0.5*[OIII]/[NII]6583 Å, b) R = [OIII]5007 Å/H, c) R = [NII]6583 Å/H, d) R = [OIII]4959 Å/H.|
|Open with DEXTER|
Only 2 optical low-states of BZ Cam are known over the last 110 years, separated by 71 yrs. As mentioned in the introduction, the duration of the 1999/200 optical low state is difficult to determine, but is shorter than 280 days. The duration of the 1928 low-state was less than about 100 days. Assuming 180 days as a mean low-state duration, results in a duty cycle of about 1/150. Thus, if supersoft X-ray emission occurs only during the optical low-states (as e.g. in 0513), then we can deduce an ionizing luminosity (for [O III]) during the optical low-state of - erg/s. Such an X-ray luminosity is very similar to that observed from the VY Scl star V751 Cyg (Greiner et al. 1999) or V Sge (Greiner & Teeseling 1998) during their optical low-states. Note that [O III] is only ionized if the effective temperature of the white dwarf is hot enough, say 10 eV, which is expected to primarily happen during optical low-states when X-ray emission is on. In contrast, hydrogen requires a lower ionizing potential, and therefore can be expected to be ionized also during the optical high states.
Since the size of the ionization zone is proportional to the ionizing flux (Rappaport et al. 1994), a rough estimate of the expected size of BZ Cam's nebula can be made by comparison to the canonical SSB CAL 83 and its nebula (Remillard et al. 1995). With our X-ray flux, the expected size of the ionization nebula (in H) of BZ Cam should be a factor 50 smaller than that of CAL 83, i.e. 0.4 pc. Indeed, the size of the BZ Cam nebula in H is 2 , corresponding to pc at 830 pc distance. Thus, the East-West extension of the nebula (perpendicular to the proper motion) is in perfect agreement with the prediction for an ionization nebula.
With the knowledge of the proper motion of BZ Cam, the bow-shock-like shape of the nebula as well as the two-component appearance can be explained based on a moving ionizing source, very similar to the theoretical considerations of Chiang & Rappaport (1996). Note that the time scale for illumination/photo-ionization is generally smaller than the time scale for recombination. In addition, the ionization time scale is fairly insensitive to the source luminosity (during the on-state) so that we can use the analogy to the more luminous, canonical 1037-38 erg/s SSBs (Chiang & Rappaport 1996). As a source emits ionizing photons, the ionization front propagates radially outward and decelerates as the ionizing flux attenuates due to both geometric (1/r2) dilution and photoelectric absorption. If a source moves substantially before the ionization front nears its equilibrium radius, the ionization nebula will become elongated, i.e. while ahead of the motion fresh gas is continually ionized, a slowly fading wake of recombining ions is left behind. Using formula (7) of Chiang & Rappaport (1996), and using cm-3 and , this will happen for BZ Cam at any velocity larger than 50 kms-1. With our above derived space velocity of 125 kms-1 (assuming 830 pc distance) and a homogeneous surrounding interstellar medium one theoretically would expect an axis ratio of about 1.4 (see Fig. 8 in Chiang & Rappaport 1996). Again, this is in surprising agreement with the value for BZ Cam of 1.3 as measured from the H image (right panel of Fig. 4).
As shown earlier (Fig. 4), BZ Cam's nebula seems to consist of two components. We believe that the above description, i.e. ionization by a moving source, applies to both components. However, the smaller components with the filaments (which are very bright in the line emission) may represent regions of higher density which could be shaped by both the ram pressure of the bow-shock (ahead of BZ Cam's motion) as well as episodes of transient major ejection events (in the far tail of BZ Cam's nebula). Both the larger size of low-excitation emission as compared to e.g. [O III] as well as the distance dependence of the emissivity in [O III] argue in favor of photo-ionization also of the small, filamentary component which is just shaped, but not excited by the hydrodynamic shock formed by BZ Cam's motion.
Our conclusions can be summarized as follows:
This naturally explains the nebular emission line ratios, the shape, size and flux of BZ Cam's nebula, and avoids the many complications related to a shock-excitation interpretation as proposed by Krautter et al. (1987), and elaborated by Hollis et al. (1992), which among others include (1) the radio emission is thermal, (2) the radio nebula anticorrelates with [OIII], (3) the large [OIII]/H ratio, (4) and the necessity to truncate the recombination zone by e.g. assuming an environment with larger rather than lower density.
G. T. acknowledges J. R. Valdès for doing part of the observations at Cananea Observatory. Much of the optical data presented in Fig. 1 were taken from the VSNET which we kindly acknowledge. This research has made use of the SIMBAD database and the VizieR Catalogue Service, Strasbourg, of the Digitized Sky Survey data produced at the Space Telescope Science Institute, Baltimore and of the ESO Skycat Tool, version 2.1.1. RDS gratefully acknowledges financial support from the Deutsches Zentrum für Luft- und Raumfahrt (DLR) (Förderkennzeichen 50 OI 0001). We thank the referee, K. Mukai, for constructive comments.