Received______________;          Accepted_____________;

An observational study of the nova-like cataclysmic binary V347 Pup (LB1800) is presented. An analysis of optical and UV spectroscopy is made aiming to define the physical properties of the binary system and of the accretion disk. The study of the line profile behavior and the determination of the primary radial velocity are pursued using a variety of methods. We also present the detection of secondary spectral signatures which were used to derive its radial velocity curve. A tentative companion spectral classification and spectroscopic parallax are also given. A Doppler tomography study of Balmer and HeII lines lead us to an estimate of the average surface brightness distribution of these lines in the accretion disk. Exploratory modeling of the accretion disk in V347 Pup and comparison with UV observations is carried using the system parameters constrained by the radial velocity study.

Subject headings: cataclysmic variables, stars: accretion, stars: individual (V347 Puppis)

The determination of masses and orbital parameters of cataclysmic variables (CVs) is a field plagued with large observational uncertainties. Spectroscopic dynamic solutions for CVs are strongly affected by the interpretation of line profile velocities and require high signal-to-noise data (Warner 1995). Photometric mass determinations using eclipses on its turn are available for a small set of CVs and they also suffer from method and observational uncertainties. From the spectroscopic analysis perspective, the presence of intrinsic gas motions with velocities comparable to the primary orbital velocity and the uneven distribution of the absorption spectrum over the companion surface (e.g. Wade & Horne 1988) are only examples of the problems intrinsic to the task of finding reliable dynamical solutions from radial velocity curves. Although there are mass estimates for 78 out of 318 CVs with known orbital period, there are only 10 identified CVs that present both primary eclipses and two line systems (Ritter & Kolb, 1998). Closed solutions have been found for most of the members of this subset with a minimum number of assumptions and approximations. Well-established basic parameters allow a better constraining of accretion disk models and complement emission line analysis of individual systems. For example, the interpretation of Doppler Tomogram often involves the knowledge of the position of the stellar components and of the gas stream in velocity coordinates. Transforming those velocity maps in position coordinates also rely on the knowledge of the primary potential well. In a CV population context an unbiased and reliable parameter distribution over the known sample is needed to constrain evolutionary scenarios and for a better understanding the relations between the various CV subclasses.

The high galactic latitude (b = 27º) star LB1800 was identified as a bright nova-like cataclysmic variable (V347 Puppis) by Buckley et al. 1990 (BSRTC hereafter) on the basis of optical spectroscopic observations. Partial primary eclipses were observed defining an orbital period of 5.5 hours. In addition, these authors found the system to be located within the error box of the Uhuru X-ray source 4U0608-49, moreover visible in HEAO1 scanning experiment. These findings strongly suggested that V347 Pup is the optical counterpart of a transient X-ray source. Spectropolarimetric observations followed by this group ruled out the possibility of a highly magnetized white dwarf for V347 Pup. The serendipitous discovery of V347 Pup among Luyten & Anderson (1958) list of blue selected stars has been indeed important for characterizing the properties of CVs in general, since it is one of the brightest (m_V ~ 13.4) eclipsing nova-like systems known.

After its identification the system was observed in the UV and X-rays by the IUE, ROSAT (Mauche et al. 1994, Shlosman, Vitello, & Mauche 1996) and more recently by the HST. IUE, ROSAT and optical observations were globally analyzed by Mauche et al. (1994, M1994 hereafter) who modeled the accretion disk continuum using two characteristic temperature regimes to account for the concave UV spectral shape. V347 Pup is a faint, soft ROSAT source and was not detected by EUVE in any passband. These authors suggested the presence of an extended self-eclipsing disk with a "hidden" boundary layer. The behavior of UV lines during eclipse was successfully modeled (Shlosman, Vitello, & Mauche, 1996) by the emission from a rotating, line-driven wind. Their wind models may be calculated with reasonable mass-loss rates if the disk maximum temperature is ~50000 K. Despite the limited time coverage of spectroscopic observations, no low-states have been observed in this system while the emission line ratios and continuum flux seems remarkably stable (e.g. Mauche, Lee, & Kallman, 1997). This suggests that the accretion disk in this system is as a good test case for steady and high- disk models.

The next section detail the observational procedures used for obtaining and reducing the spectroscopic data. Section 3 describes the optical spectrum, a radial velocity study with resulting revised system parameters and the Doppler tomography analysis. The content of section 4 focus is on the accretion disk modeling. A discussion of the results and a comparison with previous works is given in section 5. Finally the conclusions are summarized in section 6.

The spectroscopic data described in this paper was obtained at the 1.6m B&C telescope of LNA at Itajubá, Brazil and at the 1m telescope of CTIO. Most of the data was taken at LNA with a Cassegrain spectrograph, employing a 1200 groves/mm grating in the blue and a 900 groves/mm grating in the red. The detectors were a front-illuminated GEC 1152´ 770 CCD and a back-illuminated SITe 1K´ 1K CCD. The instrumental spectral PSF was properly sampled by more than 2.0 pixels. Variable seeing was common during observations. To preserve resolution and wavelength stability the slit was opened to roughly match minimum seeing values. A typical FWHM resolution of about 2 Å was usually achieved (table 1). The CTIO observations were taken with the 2D-frutti detector at the Cassegrain spectrograph yielding similar blue coverage and resolution. Direct images in the R_KC filter of the target field were taken at LNA to evaluate the possibility of spectral contamination by a faint nearby star 9" southeast of V347 Pup. Its contribution was found to be negligible even during worst seeing conditions. Atmospheric dispersion and zero order slit loss corrections were made by opening the spectrograph slit during best atmospheric conditions in order to take a flux calibration exposure of the target. Standard stars from Hamuy’s et al. 1994 list were observed through wide slit widths for flux calibration. Bias, Flatfields and twilight slit-illumination frames were also taken. Dark current contribution was found to be negligible. During the time-series, target exposures were bracketed by He-Ar lamp calibrations and the corresponding wavelength solution was then interpolated for each target observation. Internal wavelength errors were typically smaller than 60 mÅ using low order functions. The basic data reduction followed standard procedures in IRAF including spectrum extraction by a minimum variance algorithm for gray-time nights. A total of 136 blue spectra were taken covering from 3950 Å to 5000 Å. After the observations in the blue were made, 98 red spectra ranging from 4900 Å to 6350 Å were taken in 1997 and 1998. Both blue and red spectra are well distributed along binary phase with no significant gaps. Exposure times were limited to 10 to 15 minutes (

The disk modeling section made use of HST/Faint Object Spectrograph observations of V347 Pup. These data were obtained in November, 15, 1996 in rapid-readout mode, using grating G130H and a pair of square apertures. This configuration yielded a spectral resolution for a point source. Only the average continuum between 1150Å and 1600Å was used in the present analysis. A continuum signal-to-noise ratio above 30 is typical in the combined spectra.

3.1.  The average spectrum

The average out-of-eclipse spectrum of V347 Pup is shown in figure 1. It presents a high-excitation spectra with EW (HeII 4686Å) about 2/3 EW(Hb ) and a blue continuum. A tabulation of optical line fluxes and equivalent widths is given in BSRTC while the UV lines are measured by M1994. The spectrum shown represents a long-term average in the blue composed with later observations in the red (l  > 4950 Å). Average out-of-eclipse synthetic magnitude and color index: V = 13.6 and (B-V) = 0.15 match the broad band photometry by BSRTC quite well. The overall out-of-eclipse continuum shape over the 1000 Å blue coverage does not change significantly when compared to the full eclipse average. The HeII to Balmer flux ratio is also comparable among such averaged data. This may suggest a partial eclipse; in agreement with the photometric eclipse shape in which a flat bottom is not observed (BSRTC). These authors observed CaII H and K absorption and interpreted this effect as an absorption of the disk continuum by the stream. Many absorption features are seen in our data in the blue, including the G band. These features reach maximum strength relative to the continuum close to eclipse. After this analysis it was decided that a search for the companion spectra should be pursued between Hb and Ha as the secondary was assumed to be possibly too hot to show molecular bands.

 

Figure 1: Combined blue and red average spectra of V347 Pup from phase 0.1 to 0.9. The instrumental resolution is 2 Å (FWHM). Blue and red observations are not simultaneous (see text).

The eclipse average (-0.09 < phi < 0.09) spectrum (figure 2) shows more absorption features in the red like MgI, NaI D, and well defined CaI l l 6103,6122,6162 Å lines. These CaI multiplet 3 lines were found to be enhanced in the spectrum of the long-period dwarf nova DX And (Bruch et al. 1997). Unfortunately the Na D1 and D2 lines are broadened and not fully resolved in our data. They are also clearly affected by the red wing of HeI l 5876 Å emission. Its combined equivalent width is >1.4 Å. If previous reddening estimates (E(B-V) <=  0.05, M1994) and galactic latitude are taken into account, it is straightforward to conclude that the strength of these lines cannot be entirely explained by interstellar absorption. The eclipse spectrum was convolved with a Gaussian to match library spectra (Jacoby, Hunter, & Christian 1998) resolution (4.5Å FWHM) and the relative strength of the features was used to assign a spectral type between K0 V and K5 V to the absorption component. The possibility of a later subgiant spectral type cannot be completely excluded from the data at hand. The continuum depression in the red side of NaI D is due to the earth atmosphere (H₂O) as well as the molecular oxygen band at 6276 Å (Curcio, J. A., Drummeter, L. F., & Knestrick 1964).

 

Figure 2: Continuum-subtracted average spectrum of V347 Pup over the phase interval from -0.09 to 0.09. The presented resolution is 4.5 Å (FWHM). Each major tick mark separation corresponds to 10¹⁵ erg/s/cm²/Å. The observed spectrum of a K0 V star (HD23524) is shown scaled by 2.98´ 10^-3 to mimic the line strengths in V347 Pup. Its was also shifted down for comparison. The most prominent features found on both spectra were identified (Moore, Minnaert, & Houtgast 1966).

3.2.  Time Resolved Spectroscopy of Emission Lines

The binary phase in our study was computed using Baptista & Cieslinski 1991 ephemeris:  HJD_min = 2446836.96176(±25) + 0.231936060(±35) E . This ephemeris was tested by measuring the eclipse minimum position in continuum phase-folded light curves. It was found to be in good agreement with our data. To better understand the line profile and flux variations along the orbit phased grayscale images of Hb , Hg and HeIIl 4686 Å were computed. The Balmer and HeII lines are brighter relative to the continuum during phase interval from ~0.4 to ~0.8 suggesting that the continuum emission from the hot-spot is probably dominant (figure 3). The previous claim of the presence of absorption reversals in the Balmer lines (BSRTC) could not be confirmed while a double-peak appears in the HeII line during phase ~0.5. The Hb and HeII l 4686 trailed spectrograms also show that the emission lines are not totally eclipsed.

 

Figure 3: Hb trailed spectrogram. Continuum subtracted spectra are displayed with a phase resolution (FWHM) of 0.08 and a velocity resolution of 120 km/s.

The radial velocity measurements were made employing a variety of methods (table 2), some of them are not intended to give physically meaningful results but to clarify the phase-dependent behavior of the profiles. Hb , Hg and HeII l 4686 Å heliocentric velocities were measured using single Gaussian fits, flux weighted centroids, and line maximum. The line maximum was estimated by weighting the wavelength by a high power (8-12) of the profile flux. Strong rotational disturbance is clearly present in Hb centroids r.v. curves extending to +/- 200 km/s. This effect is also seen in Hg , while its is barely visible in HeII l 4686. The methods mentioned above tend to sample basically the line cores with increasing weight given to the intensity distribution. Sinusoidal fits to these curves have comparable semi-amplitudes of about 120 km/s. They agree with the Gaussian fit measurements presented for optical lines by M1994.

 

 

 

 

 

 

	TABLE 2 SINUSOIDAL FITS TO RADIAL VELOCITY CURVES
		Methoda		K (km/s)		gamma (km/s)	RMS (km/s)
		Hb   core & wings
		Centroid		124(±13)	0.04(±3)	2	29
		Single Gauss		109(±15)	0.05(±3)	-3	33
		Cross-corr		88(±12)	0.06(±3)	-	24
		S&Yc		207(±11)	0.02(±3)	16	23
		Doppler Mapd		196(±49)	0.08(±5)	-	-
		Hg   core & wings
		Centroid		130(±10)	0.03(±2)	43	20
		Single Gauss		122(±17)	0.01(±3)	3	33
		S&Yc		194(±21)	0.00(±3)	60:	41
		HeII l 4686 Å  core & wings
		Centroid		102(±13)	0.07(±3)	45	26
		Single Gauss		117(±14)	0.08(±3)	28	29
		S&Yc		180(±12)	0.04(±3)	29:	25
			absorption lines
		Cross-corre		205(±13)	0.48(±1)	-	64
		Xcorr binned		203(±15)	0.50(±2)	-	40
		CaI l 6162 Å		183(±22)	0.50(±2)	-	61
a Spectra in the phase interval [0.89,1.06] were excluded from fit. b Phases of (Vrad -gamma) zero crossing from positive to negative values. c Schneider & Young method for a = 1000 km/s (see text). d Obtained directly from Hb tomogram at ½ v½ = 660 km/s (see text).

Because the fast rotating gas in the innermost parts of a keplerian disk should be less susceptible to non-orbital motions its emission is believed to be kinetically coupled with the primary star. To accomplish the task of selecting the Doppler broadened emission, the classical convolution sampling method of Schneider & Young (1980) was used to measure the velocity of the line wings on 12 continuum subtracted, phase-binned spectra. All spectra close to eclipse (-0.11 < f  < 0.07) were removed from the sample to avoid the rotational distortions. Double-Gaussian filter functions with FWHM = 180 km/s were able to resolve the radial velocity variations along the profile from the line core to a = 1300 km/s in Hb and Hg , where a = Vsin(i) is the half-distance between the two convolving Gaussian centers. Following the common assumption of fast tidal orbital circularization in CVs these measurements were then fitted by a single sinusoid with free phase (f ₀), amplitude (K) and systemic velocity (g ). The resulting standard "diagnostic" diagram (Shafter, Szkody, & Thorstensen, 1986) for Balmer lines is shown in figure 4. Some discrepant values of the Hg systemic velocity for large velocities are produced by the rather uncertain continuum subtraction at the wings of this line. The HeII l 4686 Å profile could not be measure up to such high velocities because of the blend with the CIII/NIII complex. However, consistent (within 1s ) semi-amplitudes of about 180 km/s were found for HeII wings at 900 Km/s. The Hb velocity diagnostic was repeated for different binning and therefore with different S/N ratio, and also using subsets of the spectrum set aiming to check for inhomogeneous data and sampling problems. The outcome was self-consistent with the results previously described. Another test was made by repeating the line measurements using time-series of synthetic spectra with the same phase sampling, noise and emission line properties of the real data. It was concluded that the velocity error bars are not underestimated and that the small decreasing slope in the Balmer diagnostic diagram for |v| > 850 km/s can not be attributed to noise-induced systematic effects.

 

Figure 4: Wing velocity diagram for Hb (filled circles) and Hg (open squares). The abscissa is the half separation between the double Gaussian convolution filter. Each Gaussian has a FWHM = 180 km/s. From top to bottom the panels give the sinusoid fit parameters: the systemic velocity, phase of positive to negative crossing, RMS of fit, semi-amplitude percent error and semi-amplitude value.

Both Balmer and HeII lines showed strong velocity´ amplitude dependence displaying larger amplitudes as higher velocity emission is sensed (figure 3). Nonetheless, the three diagnostic diagrams show a plateau around K about  190 km/s. On the other hand, no phase shift between photometric and spectroscopic secondary conjunction phases was found. An independent method for measuring the primary radial velocity amplitude was pursued in this study. Doppler tomograms were proved to be useful and robust tools in the study of dynamics and emission properties of CVs (Marsh & Horne, 1988). Starting from the Hb Doppler maps of V347 Pup (see section 3.6) one may fit circular isophotes for increasingly higher velocity modulus in the tomogram. By sampling the high velocity emissivity in the map we are able to, at least in principle, select regions closer to the primary without including velocity projected components which are always present in the line wings. The error and orbital phase consistency may be then evaluated using the proposed tomographic diagnostic diagram (figure 5). Again, an increase in the semi-amplitude with profile velocity is seen leading to a maximum which is in agreement with the determinations derived from classical methods. Taking the error-weighted mean K values along the plateau for the four diagnostic diagrams a consistent set of semi-amplitudes are derived for Balmer and HeII emission. By averaging them one finally finds K₁ = 193± 16 km/s, where the error is the dispersion among various methods and lines. Previous cross-correlation measurements by BSRTC revealed smaller values (K₁ = 134± 9 km/s) which are difficult to conciliate with present determination within the quoted errors. Moreover, these authors found K to be almost independent on the profile position sampled or measuring method. However, their radial velocity curves were derived from only 1.3 orbital cycles continuous coverage and therefore their measurements are less robust to intrinsic spurious line variations. More uncertain radial velocity information is available from IUE UV lines (M1994) suggesting semi-amplitude values between 200 and 340 km/s. Due to the instrumental effects discussed by these authors extreme caution should be taken in the use of such velocities. These are flux weighted centroid velocities and the curves show large phase shifts with respect to the optical lines and eclipses. Finally, the recent work by Still, Buckley and Garlick (1998) shows V347 Pup displaying quite different line profiles while no outburst was recorded. Their Balmer line profiles are characterized by deep double-peaks during most of the orbit. Gaussian convolution velocities found by these authors indicate K₁ = 156± 10 km/s for a = 800 km/s.

 

Figure 5: Tomographic diagnostic diagram. Here the radial velocity parameters are plotted as a function of the velocity modulus in the Hb Doppler image (see text).

3.3.  The Secondary Radial Velocity Curve

The source of absorption lines in the V347 Pup was investigated by taking time-resolved spectroscopy of the apparently featureless continuum between Hb and Ha . Initial attempts to use the G band and other features in the blue for obtaining secondary radial velocity curves failed for two reasons: the continuum S/N ratio was not sufficient to measure weak lines and some of the absorption features identified were affected by nearby emission lines. By examining the radial velocity curves of the G band for instance it is difficult to rule out the possibility of a significant contribution from the stream and/or the cool outer parts of the accretion disk. However, the red spectrum of V347 Pup allowed us to obtain radial velocity curves of other features like the CaI l l 6103,6122,6162 Å lines. The measured equivalent widths of these features are respectively 190, 320 and 980 mÅ. The radial velocity curve derived from CaI l 6162 Å has a 0.50± 0.03 phase shift with respect to the emission lines indicating that the companion spectrum has been detected. Besides measuring individual line velocities, the features redward of Na D were cross-correlated with a template prepared from the average eclipse spectrum (figure 6). This alternative method yields similar results. Again, exploratory binning variations were applied to the data confirming the cross-correlation semi-amplitude K₂ = 205± 13 km/s.

 

Figure 6: Emission line wings and absorption velocities measurements. Top panel shows the unbinned absorption cross-correlation values (dots) and binned spectra Hb wing velocities at a = 980 km/s (pluses). Middle panel presents the binned absorption cross-correlation values (dots) and binned spectra Hg wing velocities at same "a" (pluses). Individual measurements of CaI l 6162 Å centroids (dots) and HeII l 4686 Å wings at 940 km/s (pluses) are in the bottom panel.

3.4.  The K₂ Correction

Before using the derived semi-amplitude for describing the motion of companion center it is important to be aware that the Roche lobe filling secondary in such system may have a dimension comparable to the orbital separation. Secondary illumination, gravity darkening and the presence of spots may have considerable consequences on the line intensities which, on its turn, may cause large displacements in the effective center of measured velocities. In particular, the Ca I line strength should depend on the secondary atmosphere irradiation as its ionization potential is low. To investigate this scenario we plotted the CaI l 6162 Å flux as a function of the orbital phase. A strong dependence is seen in the sense that the line flux has a minimum at phase 0.4 - 0.6 and reaches a maximum during the inferior conjunction of the secondary. The observed line flux ratio between phase 0.0 and 0.5 is ~4. This indicates that the line is mostly produced in the back side of the secondary as seen from the disk. A first order approximation to evaluate the effect on the observed K₂ is made by considering that a spherical cap on the secondary surface has no contribution to the line while the line production over the rest of the surface is uniform. In our simple model we also assumed a limb darkening law in the Eddington approximation and neglected gravity darkening and anisotropic irradiation (Davey & Smith 1992). The intensity integrals were performed adjusting the cap size to account for the observed flux ratio. By summing the line over our bright/opaque model one finds an effective absorption center located at 1.11 x2; where "x2" is the distance between the secondary center and the binary center of mass. This procedure certainly introduces uncertainties associated with the simplified physical treatment of the problem. However, the evaluation of such errors is itself uncertain and only the measuring uncertainties will be retained in the final value of K₂ (eff) = 187± 12 km/s.

3.5.  System Parameters

It is nowadays clear in literature that, in general, the major uncertainty involved in measuring the white dwarf velocity amplitude in CV's does not come from the radial velocity curve scatter but instead, have their origin in the interpretation of the phase-dependent line wing modulations as a consequence of the primary orbital motion. From a statistical point of view, there is indication that spectroscopic masses may be biased towards large values of q º  M₂/M₁ (Bailey 1990), which may be explained if the semi-amplitudes found from emission lines are systematically higher than the true K₁ values. As mentioned in section 3.2 the measurements from Still, Buckley and Garlick (1998) for V347 Pup uses the same technique and sampled about the same velocity in the line wings obtaining K_EM = 156± 10 km/s. Instead of adopting one or other value with their small uncertainties or averaging measurements that are possibly intrinsically different, we decided to explore the whole range between both determinations of K₁. Then we used the K₂ value from previous section for constraining the basic system parameters. From the semi-amplitudes we directly find a range in the mass ratio 0.81 <= q <= 1.05. The orbital inclination may be derived using the above mass ratio and the eclipse width (D j  = 0.104± 0.006 from BSRTC and assingning a more conservative uncertainty) by a Roche lobe filling secondary assuming a geometrically thin disk (Horne 1985). This simple constraint gives "i" in the range from 77° to 84° . All the uncertainties involving the mass determination are represented in the resulting mass diagram (figure 7). Kepler’s third law may be applied to find the orbital separation while the primary and secondary Roche-lobe radii follow from standard geometry (e.g. Eggleton 1983). From the secondary Roche lobe volume radius we have 0.58 <=  R_L2  <=  0.68 R¤ while the mass-radius relation of Lacy (1977) yields 0.54 <=  R₂<=  0.75 R¤ . The former implies that the mass of the secondary given by dynamical solutions is apparently in good agreement with a normal main-sequence star contained by the Roche surface except for the high M₂ edge in figure 7. Nonetheless, there is significant intrinsic scatter (certainly > 0.05 R¤ ) in this empirical mass-radius calibration. Alternatively, the observed mass-radius relation given by Hoxie (1973) may be adopted yielding similar results (0.61 <=  R₂ < 0.72 R¤ ). A good match for the available secondary Roche lobe radius can be found among theoretical main sequence models depending on the chemical composition and mixing length chosen (e.g. Copeland, Jensen, & Jørgensen 1970).

 

Figure 7: Mass diagram for V347 Pup. The vertical dotted line represent a lower limit to the primary mass taken from the maximum red wing velocity of HeII l 4686 Å and assuming a keplerian disk. Dotted vertical line at high M₁ edge is the Chandrasekhar limit for non-rotating white dwarfs. The curved quasi-horizontal dotted line shows the ZAMS mass-radius calibration for a Roche lobe filling secondary (Patterson 1984). Curve labeled "a" shows the secondary mass-function computed from the emission line radial velocities found in this work and i = 77° . The "b" curve represents the same mass relation when the K_EM value from Still, Buckley and Garlick (1998) and i = 84° are used. The thick almost vertical curve indicate the best primary mass-function while parallel curves correspond to the maximum uncertainty produced by "conspiring" errors in K and i. Finally, the long-short dashed straight lines "c" and "d" show the constant mass ratio for i = 84° and primary eclipse widths  = 0.110 and  = 0.098, respectively. The line labeled "e" gives the mass ratio implied by i = 78° and mid-range  = 0.104.

The accretion disk in V347 Pup may be sized by a simple geometric relation (Warner 1995). Such equation is written in terms of a value for (R_s/a), the orbital inclination "i" and the average phase of first and last contact ( = 0.105± 0.005) which was extracted from the eclipse profiles by BSRTC. The radius of the optically thick disk found is ~50% of R_L1 (the distance between the primary and the inner Lagrangian point). This value will be compared with the optically thin disk in the next section. The Hamada & Salpeter (1968) mass-radius relation for cool carbon white dwarfs applied to the dynamical white dwarf mass (0.55£  M₁ £  0.71 M¤ ) completes the set of parameters presented in table 3.

 

 

 

 

 

 

	TABLE 3 RANGES ON SYSTEM PARAMETERS
		Sec. Spectrum	K0V - K5V
		KEM	156 - 193	km/s
		K2	205±13	km/s
		K2 (eff)	187±12	km/s
		q º	0.81 - 1.05	-
		i	77 - 84	degrees
		M1	0.55 - 0.71	M¤
		RL1	0.79 - 0.92	R¤
		R1  (Mass-radius)	0.011 - 0.013	R¤
		M2	0.47 - 0.68	M¤
		RL2  (Roche lobe)	0.58 - 0.68	R¤
		R2  (Mass-radius)	0.54 - 0.75	R¤
		a	1.60 - 1.77	R¤
		d	510±160	pc
		MV  (Out-of-eclipse)		-

 

 

 

 

Its interesting to note that the dynamic mass of the companion (0.47<=  M₂ <=  0.68 M¤ ) found is in agreement with the expected mass for a normal late-K main-sequence star. According to the Schmidt-Kaler (1982) calibration the K5V spectral type corresponds to 0.67 M¤ . This fact supports the hypothesis that the observed absorption spectrum is mostly produced by the companion. Although there is a possibility of spectral contamination by the outer disk and knowing the intrinsic uncertainties in the spectral type matching process, we proceed assuming that the secondary exhibit a "true" K0-K5 V spectrum. In this case, the task of finding a spectroscopic distance by scaling a suitable template is straightforward. Special care was taken in the needed equalization of resolution made to achieve the same line widths before comparing their depths. A distance of 510± 160 pc is found for A_V = 0.15 where the 1s errors were estimated from the quadratic sum of the estimated scaling error with the spectral type error. These flux errors are favorably propagated into the distance 1s uncertainty quoted above. The implied absolute magnitude out of eclipse is not surprising for a nova-like CV (Warner 1987). A small but significant fraction (about 14%) of the total flux in this passband may be due to a secondary with, for instance, M_V = 7.0 with a spectral type K4 V (Schmidt-Kaler 1982).

3.6.  Doppler Tomography

Doppler tomograms from the Hb and HeII l 4686Å profiles were calculated using the filtered backprojection method (Rosenfeld & Kak 1982). The coordinate system definition follows the usual form; the X-axis points from the primary to the secondary while the Y-axis points in the direction of motion of the secondary.

The Doppler mapping of Hb line in V347 Pup was calculated from 110 out-of-eclipse line profiles. Typical continuum signal-to-noise ratio of individual spectra used in the reconstruction is ~15. An approximately symmetric emission can be seen in a 230 km/s (FWHM) resolution map (figure 8). Additional lower resolution maps were computed to explore the high velocity domain, however, no other significant structures could be found. Since the spectrum was taken at different epochs, the following results must be regarded as the average behavior of the system among the 3 observational time samples. There is no photometric evidence for eruptive or VY Scl-like behavior between the runs. However, the derived average Doppler map should be interpreted as the average of three independent epochs that may correspond to peculiar states of the disk.

 

Figure 8: Average Doppler tomogram of the Hb emission. The central "+" marks the binary center of mass. The map rotation is defined by the photometric conjunction of the primary. The estimated resolution of the reconstruction is 300 km/s (FWHM).

Using the derived system parameters it is possible to invert Doppler tomograms into position coordinates assuming a velocity prescription for the disk. A simple Keplerian disk around a 0.8 M¤ white dwarf was assumed to transform our tomogram in an emissivity distribution within the primary Roche lobe. Although the observational noise is amplified in this process (because the brightest central part of the disk is affected by the largest profile measuring uncertainties) it is possible to obtain the median emissivity as a function of the radius with acceptable significance. This method provides us with a radial emissivity profile that is robust to the hot-spot emission and eventually, to other localized anisotropies. The radial profiles for Hb and HeII l 4686 are compared in figure 9. The radius of the optically thin disk is larger than the value obtained for the optically thick continuum found in the previous section. In principle, almost all the available space in the primary Roche lobe may be in use by the optically thin disk. On the other hand, it is evident that the Hb local emissivity is always larger than HeII across the disk. Moreover, the HeII line production increases faster towards the disk center. This behavior has yet to be understood in terms of the phenomena responsible for the disk temperature inversion and emission line formation. A comparison of irradiated disk models (Williams 1995, Ko et al. 1996) or dynamo heating scenarios (Horne & Saar, 1991) with the observed emissivity profiles for various species may yield some clues to the emission line production mechanism in CVs. The relative slope of Balmer and HeII emissivities is qualitatively consistent with the hypothesis of line formation by irradiation from the boundary layer. Nonetheless, the disk shearing and viscous dissipation should also increase towards the center.

 

Figure 9: Radial disk emissivity profiles for Hb and HeII l 4686 Å. Derived from Doppler map inversions for M₁ = 0.8 M¤ (see text). Flux units are intrinsic line emissivities per cm² of disk assuming d = 500 pc, i = 80° , and E(B-V) = 0.03.

4. DISK MODEL

4.1.  Disk Structure and Spectrum Synthesis

Previous modeling of the UV (IUE) spectra of V347 Pup was made by M1994 by composing stellar atmospheres with the standard disk model temperature profile and a recipe for the local effective gravity. In this section, the updated V347 Pup parameters given above were used as a base for an exploratory model analysis that include the disk vertical structure and spectral synthesis of its atmosphere. The accretion disk model used here is described in detail by Hubeny, 1989, 1990a, 1990b. Their behavior over a wide range of parameters is detailed in Wade and Hubeny 1998. Some sample applications of these model disk atmospheres may be found in Knigge et al. 1997 and Diaz, Wade, & Hubeny 1996.

The disk is assumed in steady state, geometrically thin and in Keplerian rotation. A self-consistent solution for the structure is computed for a set of axially symmetric concentric rings. Each ring is approximated by a plane-parallel 1D atmosphere. Such atmosphere is in hydrostatic equilibrium with the disk depth-dependent gravity while the mass surface density S is given by the standard disk model (Shakura & Sunyaev 1973). An important ingredient of the modeling is the depth-dependent viscosity which is allowed to vary roughly as a power law of the column density m(z) above a certain physical depth z:

. (1)

were is the depth-averaged kinematic viscosity, parametrized in terms of the Reynolds number of the flow Re (Lynden-Bell & Pringle 1974). The energy balance between radiation and mechanical dissipation produced by viscous shearing is attained at disk surface. The atmosphere solution in local thermodynamic equilibrium (LTE) is found by the complete linearization method using the program TLUSDISK (Kriz & Hubeny 1986) for a simple H plus He chemical composition.

Once the temperature, density and opacity are described for each ring, the radiative transfer equation is solved for the entire wavelength range by the general spectral synthesis program SYNSPEC (Hubeny, Lanz, & Jeffery 1994). All lines were considered in LTE and the line list of Kurucz (1990) was used to compute the specific intensities In (m ,R) for the expected range of orbital inclinations. The integrated disk spectrum is obtained by co-adding the specific intensities in the observed wavelength scale. Finally the resolution of the synthetic spectrum is matched to the observed spectrum.

4.2.  Model Parameters and Results

An extensive grid of disk models was computed aiming to match the UV spectrum of V347 Pup. In this exploratory analysis we tried to fit only the continuum as defined by recent high S/N ratio HST/FOS spectra between 1150Å and 1600Å. The continuum shape measured between the strong wind lines is rather flat, suggesting a low temperature disk. A wide range of parameters was searched for a non-degenerate solution: 0.55 <=  M₁ <=  0.90 M¤ ; 10^-10 <=  <=  10^-8 M¤  /yr., 200 <=  d <=  700 pc; and 77°  <=  i <=  84° . Three sets of fittings over this model grid were made for E(B-V) = 0.01, 0.05 and 0.12. Such extinction upper limit is indeed well above the maximum extinction assumed from reddening in the galaxy at this particular line of sight (E(B-V) = 0.06) as discussed by M1994. Disk structure models were computed with z  = 2/3 and a Reynolds number Re = 5000. Metals in solar abundance were included during the spectral synthesis, thought the differences between the line blanketed and the "true" continuum are usually less than 15%.

Slices showing c ² values from a 4D parameter space were analyzed by comparing thousands of limb-darkened model spectra with the observed continuum (see example in figure 10). The synthetic spectra were plotted against data for various positions along the unique c ² valley seen in the ( i, M₁, ) space. The immediate result from this analysis is that the model continuum flux distributions are always too blue, when the overall flux level is matched or too faint, when the right slope is achieved. At the wavelength range covered by FOS spectra the continuum flux points from IUE observations (M1994) are 30-35% above the HST/FOS measured values. Therefore, more severe discrepancies were found when such IUE data were fitted to our models. A similar color-flux behavior was found by Wade (1988, 1984) when comparing composite stellar atmosphere models to the spectrum of other nova-like systems. In fact, the majority of viscous dissipation in our disk models occurs deep in the atmosphere while it presents stellar-like radiative transfer at upper layers. Much lower values of z were also tried yielding similar results. This may indicate that the viscosity prescription in the form given by equation 1 is not appropriate for describing the disk structure in V347 Pup. To illustrate the flux-color discrepancy, figure 11 shows the behavior of two low c ² sample models, presented with different distance scaling to match the continuum at 1481 Å. It is clear that the observations present a flux deficit at shorter wavelengths or an excess at larger wavelengths when compared to the simulations. The lower temperature model gives a better agreement for an extremely low distance value (204 pc). Another candidate explanation for the UV spectral mismatch may be raised in terms of the radial temperature profile of the disk. If the central parts are significantly cooler than the standard model prediction or alternatively, its continuum emissivity is suppressed (e.g. by magnetic field disruption or wind "evaporation" of the inner disk) then one may possibly cope with the observed flat energy distribution.

 

Figure 11: Sample c ² contour plot for disk spectrum fitting. Levels are labeled by the log of the estimated reduced-c ² of the fit. The distance in this plot is scaled between 200 pc and 700 pc. A constant E(B-V) = 0.05 is assumed. Note the steep increase in c ² for luminous disk models.

 

Figure 11: Model disk spectra and UV continuum measurements of V347 Pup. The observations (triangles) are high S/N continuum average points taken from HST/FOS observations. They are derredened by E(B-V) = 0.05. The doted line represents a model spectrum for a disk with = 6.3´

5. DISCUSSION

The presented scenario for V347 Pup indicates that the system may have a mass ratio close to 1 or even slightly larger. The calculations of de Kool (1992) suggest that a dynamically unstable mass transfer is implied by the mass configurations found in the lower left corner of the most probable polygon in the mass diagram. For the highest secondary mass, however, dynamically and thermally stable regimes may be sustained for mass ratios up to ~1.2. Therefore, we are possibly dealing with a young CV where the secondary will lose most of its mass through magnetic braking along its long-term evolution. Unitary or "inverted" mass ratios have been encountered in other double-lined eclipsing systems (c.f. AC Cnc Schlegel, Kaitchuck & Honeycutt 1984 and EM Cyg Stover, Robinson, & Nather 1981). Nevertheless, in none of them the hypothesis of mass transfer on a dynamical time-scale was firmly claimed. The solution found in this work indicates a less massive white dwarf when compared to BSRTC value (1.2 M¤ ). This finding has some consequences on the previous wind and disk modeling of the system. In particular, the disk temperature would be slightly lower at the same mass transfer rate. Such an effect may help to attain better wind models avoiding the ad hoc hypothesis of inner disk disruption needed to keep a reasonable mass loss rate while explaining the observed wind ionization (Shlosman, Vitello, & Mauche 1996). UV continuum shape close to Lyman-a is bluer in models with higher white dwarf masses. Therefore, they are more difficult to reconcile with observations using the disk model assumptions discussed in section 4. These optically thick models give some suggestion that the steady disk has an extended low temperature radial profile. In fact, some observational studies have also revealed that real nova-like disks show large departures from the classical T_eff µ  r^-¾ relation (e.g. Long et al. 1994 and references therein). The flux deficit may be also tentatively explained if the free-free and radiative recombination continuum emission from the optically thin gas is taken into account (see discussion in Knigge et al. 1998). Another phenomenon potentially capable of changing the present model flux distribution is the irradiation of the outer disk (Wade 1988). This effect may modify the local continuum contribution in the UV making the integrated spectrum redder. The Hb emissivity profile has a slope comparable with SS Cyg (Smak 1981) and Z Cha (Marsh, Horne, & Shipman 1987). Considering that in a keplerian disk the rotation velocity W scales as R^-1.5 these radial distributions are more centrally concentrated than the empirical relation I µ  W between the surface brightness of CaII lines in chromospherically active stars and their spin velocity (Horne & Saar 1991). This discrepancy is even larger for HeII 4686. If the dynamo process is invoked to explain the energy release at disk upper layers then its is interesting to note that such mechanism in V347 Pup shows no saturation up to local rotation frequencies larger than 10^-3 Hz. Up to our knowledge, there is no disk chromosphere models available that directly predicting radial emissivities for optical recombination lines.

 

6. CONCLUSIONS

Extensive emission-line profile measurements were performed in the present work leading to new radial velocity curves for the primary component in V347 Pup. These observations indicate that K_EM = 193± 16 km/s. A search for the secondary absorption spectrum resulting in a well defined K₂ = 205± 13 km/s was followed to complement the primary radial velocity study in this eclipsing binary. A simple model for the secondary absorption line distribution was computed to correct the observed semi-amplitude to K₂(eff) = 187± 12 km/s. The secondary spectrum also yield some interesting byproducts like its spectral type (K0V-K5V) and spectroscopic parallax (d = 510± 160 pc). The dynamical solution for this CV indicates a mass ratio around 0.9 with a primary mass in the range from 0.55 to 0.71 M¤ (see section 3.5). In this configuration, the binary inclination ranges from 77° to 84° , implying partial eclipses of the accretion disk. Doppler tomography of Hb and HeII4686 confirmed our classically estimated value for K₁ and allowed the measurement of the disk radial emissivity profiles of these lines. Evidence of a central enhanced HeII emission was found when comparing its distribution across the disk with the Hb surface brightness curve. This study is complemented by the application of disk models and synthetic spectra to the observed UV continuum, making use of the homogeneous set of system parameters previously defined. We conclude that the model disk structure tend to produce alternatively faint or blue UV continuum distributions when compared to the data. These models suggest that additional sources of UV continuum other than the disk photosphere may be present in the system. If the distance estimate is forced beyond the formal errors to about 200 pc then one may find consistent results for very low temperature disk models.

We would like to thank Richard Wade for helpful discussions on the accretion disk modeling. This work is partially based on optical data obtained at LNA/CNPq and CTIO/NOAO. The UV spectral information was taken from HST public data archive and IUE archive. This study made use of bibliographical information provided by the SIMBAD database. Generic satellite data search was performed using HEASARC/GSFC databases. Research partially supported by CNPq grant No. 301029 and FAPEMIG grant No. 183696.

 

REFERENCES

Bailey, J. 1990, MNRAS, 243, 57

Baptista, R., & Cieslinski, D. 1991, IAUC, 5407

Bruch, A., Vrielmann, S., Hessmann, F. V., Kochsiek, A., & Schimpke, T. 1997, A&A, 327, 1107

Buckley, D., Sullivan, D. J., Remillard, R. A., Tuohy, I. R., & Clark, M. 1990, ApJ, 355, 617 (BSRTC)

Copeland, H., Jensen, J. O., & Jørgensen, H. E. 1970, A&A, 5, 12

Curcio, J. A., Drummeter, L. F., & Knestrick 1964, Applied Optics, 12, 1401

Davey, S., & Smith, R. C. 1992, MNRAS, 257, 476

de Kool, M. 1992, A&A, 188, 202

Diaz, M., Wade, R. A., & Hubeny, I. 1996, ApJ, 459, 236

Hamuy, M., Suntzeff, N. B., Heathcote, S. R., Walker, A. R., Gigoux, P., & Phillips, M. M. 1994, PASP, 106, 566

Horne, K. 1985, MNRAS, 213, 129

Horne, K., & Saar, S. H. 1991, ApJ, 374, L55

Hoxie, D. T. 1973, A&A, 26, 437

Hubeny, I. 1989, in Theory of Accretion Disks, ed. F. Meyer, W. Duschl, J. Frank, & E. Meyer-Hofmeister (Dordrecht:Kluwer), p. 455

Hubeny, I. 1990a, ApJ, 351, 632

Hubeny, I. 1990b, in IAU Colloquium 129, Structure and Emission Properties of Accretion Disks, p. 227

Hubeny, I., Lanz, T., & Jefferys, C. S. 1994, CCP7 Newsletter

Jacoby, G. H., Hunter, D. A., & Christian, C. A. 1984, ApJS, 56, 257

Knigge, C., Long, K. S., Blair, W. P., & Wade, R. A. 1997, ApJ, 476, 291

Knigge, C., Long, K. S., Wade, R., Baptista, R., Horne, K., Hubeny, I. 1998, ApJ, 499, 414

Ko, Y., Lee, Y. P., Schlegel, E. M., & Kallman, T. R. 1996, ApJ, 457, 363

Kriz, S., & Hubeny, I. 1986, Bull. Astr. Inst. Czechoslovakia, 37, 129

Kurucz, R. L. 1990, Trans IAU 20B, 1968

Lacy, C. H. 1977, ApJS, 34, 479

Long, K. S., Wade, R. A., Blair, W. P., Davidsen, A. F., & Hubeny, I. 1994, ApJ, 426, 704

Lynden-Bell, D., & Pringle, J. E. 1974, MNRAS, 168, 603

Luyten, W. J., & Anderson, J. H. 1958, A Search for Faint Blue Stars XII, (Minneapolis, University of Minnesota)

Marsh, T. R., & Horne, K. 1988, MNRAS, 235, 269

Marsh, T. R., Horne, K., & Shipman 1987, MNRAS, 225, 551

Mauche, C. W., Lee, Y. P., & Kallman, T. R. 1997, ApJ, 477, 832

Mauche, C., Raymond, J. C., Buckley, D. A. H., Mouchet, M., Bonnell, J., Sullivan, D. J., Bonnet-Bidaud, J., & Bunk, W. H. 1994, Ap J, 424, 347. (M1994)

Moore, C. E., Minnaert, M. G. J., & Houtgast, J. 1966, The Solar Spectrum 2935Å to 8770Å, U.S. Government Printing Office, Washington

Ritter, H., & Kolb, U. 1998, A&AS, 129, 83

Rosenfeld, A., & Kak, A. C. 1982, Digital Picture Processing, Academic Press, New York

Schlegel, E. M., Kaitchuck, R. H., & Honeycutt, R. K. 1984, Apj, 280, 235

Schlosmann, I., Vitello, P., & Mauche, C. W. 1996, ApJ, 461, 377

Schmidt-Kaler, Th. 1982, in: Landolt-Börnstein, Numerical Data and Functional Relationships in Science and Technology, Vol. 2b, K. Schaifers, H. H. Voigt (eds.), Springer-Verlag, Heidelberg

Schneider, D. P., & Young, P. 1980, ApJ, 238, 946

Shafter, A. W., Szkody, P., & Thorstensen, J. R. 1986, ApJ, 308, 765

Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337

Smak, J. 1981, Acta Astr., 31, 396

Still, M. D., Buckley, D. A. H., & Garlick, M. A. 1998, MNRAS, 299, 545

Stover, R. J., Robinson, E. L., & Nather, R. E. 1981, ApJ, 248, 696

Wade, R. A. 1984, MNRAS, 208, 381

Wade, R. A. 1988, ApJ, 335, 394

Wade, R. A., & Horne, K. 1988, Ap J, 324, 411

Wade, R. A., and Hubeny, I. 1998, ApJ, in press

Warner, B. 1987, MNRAS, 227, 23

Warner, B. 1995, Cataclysmic Variable Stars, Cambridge University Press, Cambridge

Williams, G. 1995, AJ, 109, 319