Peremennye Zvezdy (Variable Stars) 30, No. 4, 2010 Received 29 September; accepted 5 October.

Article in PDF

The progenitor and the remnant of the helium nova V445 Puppis

V. P. Goranskij¹, S. Yu. Shugarov^1,2, A. V. Zharova¹, P. Kroll³, and E. A. Barsukova⁴

1. Sternberg Astronomical Institute, University Ave., 13, Moscow, 199992 Russia

2. Astronomical Institute of the Slovak Academy of Sciences, Tatranska Lomnica, 05960 Slovakia

3. Sternwarte Sonneberg, Sternwartestrasse, 32, Sonneberg, D-96515 Germany

4. Special Astrophysical Observatory of the RAS, Nizhnij Arkhyz, Karachai-Cherkesia, 369167 Russia

1 INTRODUCTION

The outburst of V445 Pup was discovered on 30 December 2000 by Kanatsu (Kato & Kanatsu 2000). The earliest observation of V445 Pup in the outburst dated 19 November 2000 was found in ASAS archive. At that time, the brightness of the star was 8^m.8. The brightness maximum of 8^m.46 in the V band was reached on 29 November 2000. The first spectroscopic observations in the outburst by Wagner et al. (2001) showed that the Balmer emission and He I lines typical for classical novae were not present in the spectrum of V445 Pup. The spectra were dominated by emission lines arising from Fe II, Ca II, C II, Na I, O I. Line widths corresponded to an expansion velocity of about 1000 km s^-1. The ejecta produced during the outburst allow us to consider V445 Pup as a nova.

The nature of classical novae is known to be a thermonuclear explosion of hydrogen on the surface of a white dwarf in a semidetached binary system. Hydrogen accumulates on the surface of the white dwarf due to accretion from a donor, usually a red dwarf. As a result of hydrogen explosion, strong Balmer lines are observed in the spectra of classical novae. Ashok & Banerjee (2003) suggested that V445 Pup is the unique helium nova predicted theoretically by Kato et al (1989) and Iben & Tutukov (1994) who considered the case of a degenerate white dwarf accreting helium from a helium-reach donor. Note that a subclass of classical novae called helium-nitrogen (He/N) novae was introduced by Williams (1992). These novae have spectra with strong Balmer lines, and they also have He I and He II lines. CNO elements seen in the spectra were mixed by accumulated hydrogen envelope from the surface of the white dwarf through a dredge-up mechanism (see e.g. Glasner & Livne 2002). The case of helium nova suggests that the donor is a nucleus of an evolved star that previously lost its hydrogen envelope due to accretion.

In the outburst of V445 Pup, the decay of brightness by 1^m.8 continued gradually for 164 days and followed by a small rebrightening between 12 May and 21 June 2001. The last observation of V445 Pup in the outburst was registered on 11 July 2001 at visual magnitude 11.5. Then the star faded rapidly and was not seen in August 2001. On 4 October 2001, no object brighter than V = 20^m and I = 19^m.5 was found by Henden et al. (2001) at the position of V445 Pup. They remark: "The object is evidently shrouded in a thick and dense carbon dust shell, in view of the apparent over-abundance of carbon in ejecta previously observed in infrared and optical spectra". Lynch et al. (2001) detected the infrared radiation in the 3–14 µm range just 1 month after the object had been discovered. The spectrum revealed smooth and featureless continuum, which they treated as a thermal emission of dust with the temperatures ranged between 280 and 1300 K. They suggest that this dust was a product of previous outbursts, at least in part.

The detailed spectral evolution of V445 Pup in the outburst was studied by Iijima & Nakanishi (2008). They acquired both high- and medium-resolution spectra for the optical wavelengths 3900–7000 Å. They confirmed the absence of hydrogen lines and noted unusually strong emissions of carbon ions. Some metal lines had P Cyg-type profiles with absorption components blue-shifted roughly by –500 km s^-1; this velocity was assumed to be the outflow velocity. The cited authors measured large radial velocity of V445 Pup, +224±8 km s^-1, which suggested that the object belonged to the old disk population. The distance was estimated using the interstellar NaI D₁ and D₂ absorption lines to be 3.5 < d < 6.5 kpc; the reddening is E(B–V) = 0^m.51.

Lynch et al. (2004) reported that in January, 2004 the object had faded to fainter than J = 18^m, so that they could not take its spectrum in the visible range. In the infrared, they detected only two He I lines at 1.0830 and 2.0581 µm, both showing doubled profiles due to bipolar outflow. The very red continuum was detected only at λ ≥ 1.5 µm. It was produced by emission of hot dust.

Woudt et al. (2009) published the results of post-outburst JHK photometry, adaptive optics imaging in the K band, and optical-range spectroscopy of V445 Pup. They discovered an expanding and narrowly confined bipolar shell, the outflow characterized by large velocity of 6720±650 km s^-1. Some knots were moving with larger velocities, 8450±570 km s^-1. They derived an expansion parallax distance of 8.2±0.5 kpc. They noted that the expansion velocity measured by Iijima & Nakanishi (2008) with the high resolution spectra in outburst was only 500 km s^-1. Such a big difference may be due to strong collimation of bipolar ejection located just in the plane of the sky and inclined to this plane only by 5°.8–3°.7 (Woudt et al. 2009). The authors assume that the small inclination angle may confirm the presence of an orthogonal dust structure closely aligned to the line of sight and causing the strong extinction observed after the outburst. In their spatially resolved optical spectrum obtained with VLT in 2006 January in the 4465–7634 Å range, only the emission lines of [O I], [O II], [O III] and He I were seen, but not the continuum.

The presence of a bright progenitor of V445 Pup having a visual magnitude of 13.1 was first noted by Platais et al. (2001). Its absolute proper motion was small, µ = 0".008±0".004. With the distance derived by Woudt et al. (2009), the luminosity of the progenitor proved to be very large, log L/L_Sun = 4.34±0.36 which is consistent with the absolute bolometric magnitude value M_bol = –6.1±0.9. Woudt et al. (2009) note that the derived luminosity suggests that V445 Pup probably contains a massive white dwarf accreting at high rate from a helium star companion. But they did not exclude that the companion was also a bright star. Liller (2001) reminded of three hydrogen-deficient cataclysmic variables, CR Boo, CP Eri, and V803 Cen, all of them being hot blue objects showing no hydrogen, but revealing He I emission lines. The absolute magnitude of –6.1 is unprecedented high for a cataclysmic variable, making us to think about the nature of the progenitor.

The observations of the light curve in the outburst and light curve modeling by Kato et al (2008) reveal that the CO white dwarf in V445 Pup is very massive and close to the Chandrasekhar mass limit (M_wd ≥ 1.35 M_Sun); a half of the accreted matter remained on the white dwarf after the outburst. Therefore, V445 Pup was considered a strong candidate for a type Ia supernova progenitor. P. Woudt and D. Steeghs called V445 Pup a "ticking stellar time bomb" in the ESO Science Press Release 0943. Taking into account the observations with adaptive optics by Woudt et al. (2009) which show only spatially resolved products of eruption but no stellar component, it is hard to maintain the concept that the mass of the system increases. The question is what mass of the components was left after the explosion.

The scenario for V445 Pup may be quite different. Recent dynamic 3D simulations by Guillochon et al. (2010) discovered a new mechanism for the detonation of a core of a sub-Chandrasekhar CO white dwarf (with a mass lower than 1.4 M_Sun) in the system with a pure He white dwarf or a He/CO hybrid secondary. Fink et al. (2010) found that secondary core detonations were triggered for all of the simulated models ranging in core mass from 0.810 up to 1.385 M_Sun with corresponding helium shell masses from 0.126 down to 0.0035 M_Sun. In that paper, the double detonation scenario remains a potential explanation for type Ia supernovae. But the destruction of the CO white dwarf means that V445 Pup, after its outburst in 2000, will not be a type Ia supernova progenitor. It is of great interest if the narrowly confined bipolar cones observed by Woudt et al. (2009) are debris of the detonated white dwarf. The overabundant carbon in the outburst will also be an evidence for CO white dwarf detonation.

Fortunately, there are many photographic images of V445 Pup in the world astronomical plate collections suitable for resolving the puzzle of the progenitor. Woudt et al. (2009) verified the plate archives at the Harvard-Smithsonian Center for Astrophysics (USA) and found no prior outbursts in 1897–1955. The progenitor of V445 Pup was identified on many plates at approximately constant brightness (from visual comparison with surrounding stars). We found many plates in archives of the Sternberg Astronomical Institute (SAI) of the Moscow University (Russia) and in archives of the Sonneberg Observatory (Germany). Two of us (V.P.G. and S.Yu.Sh.) performed eye estimates of V445 Pup independently, V.P.G. for Moscow plates, and S.Yu.Sh. for Sonneberg plates. The two sets showed similar behavior and marginal variability. But unexpectedly, the preliminary frequency analysis revealed the same periodicity in both sets with the period of 0^d.650653, coinciding to the 6th significant digit. Both light curves were of low quality. Therefore we decided to digitize the images of V445 Pup and to perform digital processing.

2 OBSERVATIONS AND DATA PROCESSING

In the Moscow SAI plate collection, we found 51 plates with images of V445 Pup taken with the SAI Crimean Station 40-cm f/4 astrograph and dated between 15 November 1969 and 4 November 1989. AGFA ASTRO and ORWO ZU-2 photographic plates, produced in the former GDR and having high sensitivity in blue light, were used, the exposure times were 45 minutes. The geographic position of the SAI Crimean Station is 2^h16^m08^s +44°43´42". The declination of V445 Pup is about –26°. This means that the highest altitude of the star above the horizon is 19°. Observations were limited to a 3-hour visibility time around this point. Photographic plates were mostly centered at τ CMa, they cover an area of 10°×10°. The region of about 20´×20´ centered at V445 Pup was digitized for each plate using the SAI CREO EverSmart Supreme scanner. CREO scanner frames are in the TIFF format.

We found 56 measurable images of V445 Pup on the plates of the Sonneberg Observatory collection dated between 19 March 1984 and 17 January 1991. These plates were taken with the 40-cm f/4 astrograph having optics basically similar to that of the SAI Crimean Station astrograph. Also, plates of basically the same type produced in the former GDR were used, so all our photographic material is very uniform. The Sonneberg Observatory has the geographic position 0^h44^m46^s +50°22´39", it is located about 5° to the north by latitude compared to the SAI Crimean Station. Thus, the star rises only to 14° for this geographic point, and its visibility time is less than that for the Crimean Station. Sky images are evidently affected with variable atmospheric extinction across the plate field. In these plates, the star is located near the center. The typical exposure time for these plates is 20 minutes. Images of V445 Pup were digitized using the Fuji FinePix F10 CCD camera and an ordinary biconvex lens. Frames made with this camera are in the JPEG format. After several experiments, we found that this method of digitizing gave the quality of measurements near the field center as good as that of a scanner. To increase the S/N ratio, we co-added several subsequent frames in night series. For this purpose, frames were put together by matching two stars, with the needed field rotation. The combined exposures of co-added frames were between 40 and 60 minutes. The total number measurements for the star, including co-added ones, is 31.

Additionally, we measured all the Internet-accessible Digital Sky Survey images of V445 Pup in B,R, and I bands and used 2MASS JHK magnitudes to study the spectral energy distribution of the progenitor.

All the frames were processed in the Windows BITMAP format. Extraction of images was made using the aperture method with star-profile correction; the WinPG software developed by V.P.G. was utilized. Special software was written by V.P.G. to approximate the characteristic curves with an nth-degree polynomial, with graphical output. Practically, the approximations with n = 1 or 2 were optimal. The total number of comparison stars used to build a characteristic curve was 23; a few stars with the largest deviations were eliminated from calculations, and the characteristic curves were re-calculated in such cases. The r.m.s. deviation of comparison stars from the polynomial fit was formally taken for the uncertainty of V445 Pup measurement.

Fig. 1. The finding chart of V445 Pup and comparison stars. This is a copy of a digitized image obtained with the UKST Schmidt telescope on 4 April 1980 on IIIaJ emulsion with a GG 395 filter. The progenitor is indicated as "var". V magnitudes, colour indices of marked stars, and corresponding uncertainties (in units of thousandths of a magnitude) are given in Table 1.

Carrying out oure photographic measurements, we used the CCD BVR_CI_C standard sequence in the vicinity of V445 Pup published by A. Henden for VSNET. We present the finding chart of the progenitor in Fig. 1. The standard stars chosen by us are also marked in this Figure. We give Henden's magnitudes and their uncertainties for the chosen comparison stars in Table 1 because they are no longer accessible at the VSNET address.

The Moscow archive observations of the V445 Pup progenitor are presented in Table 2; the Sonneberg ones, in Table 3; and those from digitized sky surveys are collected in Table 4.

We performed our observations of the V445 Pup remnant between 31 March 2003 and 20 October 2009. These observations were acquired in the Special Astrophysical Observatory (SAO), with the 1-m Zeiss reflector and CCD UBVR_CI_C photometer equipped with an EEV 42-40 CCD chip. The geographic position of SAO is 2^h45^m46^s +43°39´12". The highest altitude of the star over the horizon is 20°. This object is difficult for observations and needs good sky transparency and seeing. Some constructions of the 6-m telescope dome located to the south of the 1-m reflector humper observations of objects with such a southern declination. Additionally, a part of our observations were obtained with the SAI Crimean Station's 60-cm reflector and UBVR_JI_J photometer with the Princeton Instruments VersArray CCD. Both devices are cooled with liquid nitrogen to a temperature stabilized at –130°C, allowing to record signals from very faint astrophysical objects. The frames were reduced in the FITS format. The extraction of images was made using the same aperture method with star-profile correction, the WinFITS software by V.P.G. was utilized. Our CCD observations are presented in Table 5.

3 ANALYSIS

The light curve of V445 Pup in the B band excluding the outburst is shown in Fig. 2. The variability of the progenitor is evident, and its full amplitude exceeds the mean uncertainty of the observations more than thrice.

Fig. 2. The pre- and post-outburst light curve of V445 Pup in the B band. The start and the end of the 2000–2001 outburst are indicated with a double vertical line.

The frequency analysis of the progenitor observations was performed using two independent methods: (1) the discrete Fourier transform for arbitrarily distributed time series (Deeming 1975), and (2) the phase dispersion minimization (PDM) method (Lafler & Kinman 1965). Implementations of these methods are provided by the EFFECT code developed by V.P.G. We analysed the combined time series including Moscow and Sonneberg photographic observations. The periodograms are shown in Fig. 3 a-c. The panels (a) and (b) of this figure present the amplitude spectrum and the spectral window for this series. We estimate significance levels for peaks of the amplitude spectrum using the empirical method suggested by Terebizh (1992). This method is based on a statistical analysis of simulated chaotic series generated by mixing the original series. In the chaotic series, each Julian date gets an accidental magnitude chosen from the same original series and, as a result, the chaotic series includes the same magnitudes and times. When we compute the amplitude spectrum of the chaotic series, we make more than a million of accidental light-curve implementations with arbitrary periods and estimate their amplitudes. The software provides the analysis of the cumulative probability distribution function for amplitudes in the spectrum of the chaotic series. The amplitude levels corresponding to cumulative probabilities of 90, 99, 99.9 and 99.99 percent for the chaotic series are plotted in Fig. 3a as straight lines.

Fig. 3. Periodograms of the V445 Pup progenitor. (a): The Deeming amplitude spectrum in the (10–0.3)-day period range. The parameter is the half-amplitude. (b): The spectral window of the same time series in the (1000–0.3)-day period range. The parameter is the half-amplitude. (c): The Lafler–Kinman periodogram. The parameter is θ^-1, θ being the normalized sum of squared magnitude differences between each two subsequent points of the phased light curve calculated with a trial period.

The presence of strong peaks in the amplitude spectrum of the original series exceeding 99.99 percent amplitude level of the chaotic series means that the probability of casual appearanse of these peaks is less than 0.01 percent. The progenitor of V445 Pup was evidently a periodic variable star. The multiplicity of peaks means that we have multiple solution for periodicity with the Moscow and Sonneberg series.

The spectral window (Fig. 3b) demonstrates the periodicity in time discontinuities in our series amounting to the sidereal day (P_sd = 0^d.997262) and of P_sd/n, where n = 2, 3, 4.... The amplitudes of these peaks decrease when the period decreases because of increasing phase window. The phase window for P_sd is 0.2. Thus, for each peak in window spectrum, we have a pair of symmetrically located alias peaks in the amplitude spectrum. The light curves corresponding to this pair of peaks have reverse phase count, so they look as mirror-reflection ones. The list of periods and frequencies of aliases is given in Table 6. One can see that the dominating peaks are sidereal-day-related. The formula of corresponding interdependence is given for each peak in the last column, `Remark', of Table 6. For f₀, we chose the lowest-frequency wave with the highest amplitude. Peaks of a lunar month (29.5 days) and of about a year (363 days) in the spectrum of the window are also present, which are responsible for combs of small peaks located around sidereal-day-related peaks in the star spectrum.

Fig. 4. The light curves plotted for the periods determined with Deeming (1975) method and presented in Table 6. The elements used to calculate phases are given below each curve.

The light curves corresponding to all alias periods are given in Fig. 4. These are single-wave curves. The light curves are approximately sinusoidal, the full amplitude of the sine wave is about 0^m.4. Formally, the Deeming method reveals the highest-amplitude light curves for two periods, 1^d.871862±0^d.00009 and 2^d.134469±0^d.00011 with equal half amplitudes of 0^m.22. The scatter of all the light curves reveals essential intrinsic variability. A few points do evidently contradict the sinusoidal solution. We verified these points and confirmed their Julian dates and magnitudes. These light curves may represent the case of reflection effect on the surface of a secondary star due to heating of a part of its surface by the X-ray or short-wavelength radiation coming from the primary star. However, no X-ray source was associated with V445 Pup before its outburst. In principle, such light curves may arise due to a large hot spot on the surface of a star. FK Com-type stars may be examples of a rotating star having a hot spot on the surface. These stars are considered to be close binary systems with a common envelope.

Fig. 5. The light curves plotted for double periods determined with the Deeming (1975) method and given in Table 6. The elements used to calculate phases are given below each curve.

Double-wave light curves for periods found by the Deeming method presented in Table 6 were also calculated and are shown in Fig. 5. Double-wave light curves are exhibited by W UMa-type binaries, these are also stars with common envelopes. In most observational cases of photometry without additional spectroscopic information, like radial velocity curves or double lines in the spectra, we can distinguish between single-wave and double-wave orbital periods only by differing alternate minima. The difference of minima depths appears due to difference of surface brightness of the components. Our observations do not show alternate minima of different depth. Unfortunately, the accuracy of photographic observations is insufficient to make a reliable choice between single- and double-wave curves. However, taking into account that this system contains a massive accreting CO white dwarf (Kato et al. 2008), we think that a W UMa light curve is not an acceptable solution because components of such a system should have different brightness.

Fig. 6. The light curve for the best single-wave period found with Lafler & Kinman (1965) method.

The Lafler–Kinman (L–K) method reveals periods not exceeding one day as preferable: 0^d.650654±0^d.000011, 0^d.679686±0^d.000012, and their double-wave aliases 1^d.301269±0.000044 and 1^d.359423±0.000048. The 0.650654-day period determined with the L–K method differs essentially from that determined by the Deeming method because the light curve plotted with the PDM solution shows a local detail at the phases between 0.9 and 0.1 that looks like a shallow eclipse (Fig. 6). However, additional photographic material is needed to verify if this detail is real. Certainly, a detail like a shallow eclipse cannot be revealed in a light curve with the Deeming method. The double-wave light curves found by the L–K method seem more irregular and asymmetric. Additionally, the light curve with the period of 0^d.650654±0^d.000011 is the most symmetric one and has the lowest dispersion, so we choose it as the best solution for the present time.

Our investigation shows that one can find a single final solution for the orbital period of V445 Pup only using observations taken at different geographic longitudes, thus increasing the observational phase window of the sidereal