THE EXPANDING BIPOLAR SHELL OF THE HELIUM NOVA V445 PUPPIS

The long-term brightness evolution of V445 Puppis at optical and near-infrared wavelengths is shown in the lower panel of Figure 1. It shows V445 Puppis before outburst in the V band and the 2MASS J, H, and K_s single epoch (Skrutskie et al. 2006), during outburst in optical¹⁰ and near-infrared (Ashok & Banerjee 2003), and post-outburst where the initial H and K_s observation after decline is from Ashok & Banerjee (2003) and the remainder of the post-outburst light curve is derived from our IRSF/SIRIUS J, H, K_s, and Magellan/IMACS V-band observations. The exact date of outburst is uncertain, but is constrained by a lower detection limit (V < 12 mag) of V445 Puppis less than 2 months prior to its discovery in outburst, see the discussion in Ashok & Banerjee (2003). The outburst has occurred within the two vertical dashed lines in Figure 1. The four epochs of high-angular resolution imaging with NAOS/CONICA are indicated by the vertical dotted lines, the three epochs of IMACS observations are marked by the solid vertical bars.

More than 8 years after outburst, V445 Puppis is still highly reddened and ∼6 mag below its pre-outburst brightness as measured in the J band and 5.2 mag below its pre-outburst brightness in the V band; at this phase, optical light is dominated by line emission with a negligible continuum contribution. In the last 2–3 years, V445 Puppis has steadily declined in brightness in the H and K_s bands, but stayed constant in the J band. Where the J-band flux has almost no continuum component and is dominated by the 1.0830 μm He i recombination line emission, the H and K_s bands contain a thermal continuum component of T = 250 K (Lynch et al. 2004). The latter band also contains the 2.0581 μm He i recombination line which has an equivalent width of ∼−180 Å as derived from our SOFI spectrum. The upper panel of Figure 1 shows the (J − H)⁰ and (H − K_s)⁰ color evolution of V445 Puppis, where the photometry has been corrected for the Galactic foreground extinction of E(B − V) = 0.51 mag (Iijima & Nakanishi 2008). We refer to Section 4.1 for a more detailed discussion on the Galactic foreground extinction. The fading in the H and K_s band—at fairly constant (H − K_s)⁰ color—could indicate a cooling of the thermal continuum component over the last 2–3 years. Spitzer observations of V445 Puppis taken during this phase (PI: Banerjee) will be able to constrain the temperature of the cool thermal component.

Figure 2 shows the deconvolved near-infrared K_s (false color) images obtained at the four different epochs. The deconvolved field of view of the 2005 March and December images is 166 × 152, whereas the 2006 October and 2007 March observations are displayed on a wider scale of 348 × 348. The incredible detail seen on these small scales—the ejecta only span 2''–3'' on the sky—is possible thanks to the AO technology on large ground-based telescopes. This allows a clear morphological description and dynamical analysis of the evolving nova ejecta. In Figure 3, we show the shell in 2005 March (in contours) superimposed on the 2007 March observations to demonstrate the distinct expansion of the shell.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The images unambiguously show a bipolar shell—initially with a very narrow waist—with lobes on each side (northeast: NE, and southwest: SW) of a centrally obscured region covering the nova remnant. At all epochs, two knots are seen at either extreme end of the nova shell. This shell is unlike any previously observed classical nova shell, though strongly reminiscent of bipolar planetary nebulae (PNe), proto–planetary nebulae (pPNe; Balick & Frank 2002), and the symbiotic Mira nebula Hen 2-104 (Corradi et al. 2001). In the case of V445 Puppis, though, the large expansion velocity of the shell is clearly consistent with a nova explosion, rather than a PN outflow. The central obscuration appears to be confined to a plane nearly perpendicular to the two lobes of the bipolar shell. Such an obscuration has also been seen in various pPNe (e.g., in Hen 401; Sahai et al. 1999), where it is interpreted as a thick circumstellar (CS) dust disk. As the ejecta of V445 Puppis had a dominant carbon signature shortly following the outburst and as the nova remnant has now been obscured for the last 8 years following outburst, it is likely that this dust is largely concentrated in a plane perpendicular to the lobes where it continues to obscure the underlying binary. Given the large obscuration post-outburst (see the lower panel of Figure 1), this dust disk must be a direct consequence of the 2000 November outburst.

In Figure 4, we show the spatially resolved velocity profile of the He i 7065 Å line extracted from our IFU spectroscopy. These were observed close to the second epoch of the VLT NACO K_s-band imaging campaign. We averaged together several fibers across the minor axis of the nova shell to generate each plotted line profile, sampling the shell approximately along its major axis, from east to west on the sky. This sampling at 017 per row of fibers is seeing-limited with a FWHM of 05 during our observations. Two components are distinctly visible, associated with each side of the nova shell, as expected from a bipolar outflow (Solf & Ulrich 1985). Peak emission of the NE lobe is at −1270 km s⁻¹ and +1140 km s⁻¹, whereas peak emission of the SW lobe is at −760 km s⁻¹ and +1720 km s⁻¹; the typical measurement errors are ± 20 km s⁻¹. The offset is presumably due to a mild inclination to the line of sight, which results in a different mean radial velocity for the two lobes (−60 km s⁻¹ versus +480 km s⁻¹ for the NE lobe and the SW lobe, respectively). The mean of these velocity components agrees well with the systemic radial velocity of V445 Puppis of 224 ± 8 km s⁻¹, reported by Iijima & Nakanishi (2008). All the emission lines in the Magellan IFU spectra reveal the same velocity structure, see also the summed spectrum shown in Figure 5. The He i 7065 Å line was chosen since it is one of the strongest unblended lines and allows for a cleaner measurement compared to the strong but blended O features. Moreover, the overall structure of the emission lines (as deduced from a long-slit perspective) has not changed significantly over ∼2 years (see the right panel of Figure 4) which suggests that the bulk outflow pattern has not evolved substantially.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Motivated by the observed kinematics, we model the velocity profile with a simple bipolar velocity field (non-variable in time) following the description in Solf & Ulrich (1985), namely,

$$\begin{equation} v_{ex} (\phi) = v_e + (v_p - v_e) \sin ^{\alpha } (|\phi |), \end{equation} \tag{ 1 }$$

where ϕ is the latitude angle (the poles are at |ϕ| = 90°, and the equator at ϕ = 0°), v_e is the equatorial velocity, and v_p is the polar velocity. The exponent α controls the degree of bipolarity: large α implies strong bipolarity. In our models we kept v_e constant at 500 km s⁻¹ based on the (equatorial) expansion velocity deduced from the P Cygni profile during outburst (Iijima & Nakanishi 2008). From the velocity profile alone (Figure 4), no unique solution to the bipolar velocity field (read: α) exists, although each value of α gives a best-fit value for the inclination i, v_e, and v_p. A wide range of models (α = 6–14) all indicate a bipolar shell which is closely aligned with the plane of the sky (i_shell = 58–37, respectively), giving further support to the presence of an orthogonal dust structure closely aligned along the line of sight causing the observed extinction.

Measurements of the angular separation of peak emission perpendicular to the polar axis were obtained at Δx = ±03, ±06, and ±08 (the latter only for the last two epochs) from the center, where x is the distance along the polar (major) axis (i.e., ϕ = ±90°); they are listed in Table 3 and shown in the left panel of Figure 6. Typical errors on the positional measurements are ± 001 for the first two epochs and ± 002 for the latter two epochs. Models with α < 10 are excluded from the high-resolution imaging (too round). Similarly, models with α>14 are excluded as they would have resulted in such high elongation of the shell that it should have been visible at Δx = ±08 at the first epoch. This was not observed.

Zoom In Zoom Out Reset image size

Table 3. Dimensions of the Nova Shell of V445 Puppis

Epoch	Δy at |x| = 03		Δy at |x| = 06		Δy at |x| = 08		Δx Knots		Errors
	('')	('')	('')	('')	('')	('')	('')	('')	('')
	NE	SW	NE	SW	NE	SW	NE	SW
1	0.19	0.19	0.25	0.24	⋅⋅⋅	⋅⋅⋅	0.76	0.75	0.01
2	0.25	0.27	0.31	0.31	⋅⋅⋅	⋅⋅⋅	0.88	0.89	0.01
3	0.29	0.29	0.34	0.33	0.32	0.34	1.08a	1.10a	0.02
4	0.32	0.34	0.36	0.36	0.36	0.36	1.15	1.18	0.02

Note.^a The uncertainties in the positions of the knots at this epoch are ± 005.

Download table as:  ASCIITypeset image

From the combined spatio-kinematic observations and models, a distance to the nova shell can be obtained; x and y values from the models have been derived from Equations (3) and (4) from Solf & Ulrich (1985) and differences with the observed values minimized by varying the distance to the nova. Even though we are comparing the velocity field of optical emission lines with the near-infrared K_s-band images of V445 Puppis, it should be noted that the velocity structure of the optical and near-infrared recombination lines of He i are identical (see the right panel of Figure 4); we are therefore probing the same outflow component at optical and near-infrared wavelengths. The distance thus derived is relatively insensitive to the choice of α; acceptable model fits (α = 10–14) lead to distances in the range of 7.8–8.4 kpc. These values are consistent with the lower limit of 6 kpc deduced from the presence of interstellar Na D absorption lines at v_lsr = +73.5 km s⁻¹ (see, Brand & Blitz 1993) in the high-resolution spectrum of V445 Puppis during outburst (Iijima & Nakanishi 2008).

We superimpose the best-fit model (α = 12) on the measurements in the left panel of Figure 6. The parameters of the α = 12 model are given in Table 4, where the formal measurement error is given as well as the systematic error for different allowed values of α. The expected radial velocity profile from the α = 12 model—along the major axis of the shell—is shown in the lower right panel of Figure 6 and can be compared with the observed profile shown in Figure 4. The bulk velocities (v_p = 6720 ± 650 km s⁻¹) are on the high end compared to hydrogen-rich novae, though not unreasonable, see, e.g., RS Oph with v_p = 5600 ± 1100 km s⁻¹ (Bode et al. 2007). We thus derive a distance to V445 Puppis of 8.2 ± 0.5 kpc (including systematic errors) by combining our dynamical study of the emission lines from the shell with on-sky expansion rates measured in AO images.

It is of interest to note that the knots behave independently from the shell. Their apparent linear expansion on the sky (0217 ± 0010 yr⁻¹), as shown in the upper right panel of Figure 6, results in a deprojected polar velocity at 8.2 kpc of ∼8450 km s⁻¹, larger than the 6720 km s⁻¹ derived from the bipolar velocity field of the shell. Moreover, a simple extrapolation of the expanding knots back to the center of the nova remnant coincides closely in time with a strong radio flare observed at ∼ HJD 245 2185 (Rupen et al. 2001). This suggests that the knots originate from a ∼345-day delayed post-outburst outflow rather than the nuclear burning event itself. The bipolar shell, on the other hand, is consistent with uniform expansion since the start of the nova outburst.

Detailed one-dimensional hydrodynamical models following the evolution of PNe (Schönberner et al. 2005; Mellema 2004) indicate that in the interstellar wind model of PNe the measured edges of the PN shells are due to either ionization or shock fronts, which move at different velocities from those measured with spectroscopy; in such a case, the expansion parallax always gives a lower limit to the true distance. This was indeed observed in the symbiotic star Hen 2-147 (Santander-García et al. 2007), where the distance derived from the expansion parallax is a factor of two lower than the distance determined from the period–luminosity relation for its Mira star component. Santander-García et al. (2007) argue for the presence of a shock front in the case of Hen 2-147. In the case of V445 Puppis, it is less clear that the edges in Figure 2 arise from a strong shock front. Our optical spectra show a distinct lack of lines related to shock ionization, e.g., both [N ii] and [S ii] 6717/6731 Å are absent, as seen in Figure 5.

A direct comparison with the Schönberner et al. (2005) models is not appropriate given the order-of-magnitude difference in outflow velocities. We do not—a priori—expect a large distance correction factor for V445 Puppis. The deprojected velocities of the polar blobs are already on the high end of speeds observed in nova ejecta; a significantly greater distance would imply velocities more typical of SNe rather than classical novae. Moreover, the absence of strong shock lines, the consistency of the He i recombination line profiles (kinematic analysis), linked with the dominance of the He i recombination line in the K_s passband (spatio analysis), suggests that the outflow velocity matches the angular expansion closely. Nonetheless, to estimate by what amount the expansion parallax could underestimate the true distance in V445 Puppis, detailed (magneto)hydrodynamic simulations (see, e.g., Dennis et al. 2009) with realistic input parameters as identified here—and the possible presence of pre-existing CS material—must be obtained.

With the distance known, the nature of the helium nova progenitor can be constrained from pre-outburst observations. Unfortunately, not much is known of V445 Puppis prior to outburst. Apart from optical (V = 14.5 mag, from VSNET; see Ashok & Banerjee 2003) and near-infrared (2MASS: K_s = 11.52 mag) flux measurements (see the lower panel of Figure 1), neither spectrum nor orbital period has been obtained prior to outburst. We verified the plate archives at the Harvard-Smithsonian Center for Astrophysics for prior outbursts (or sustained long periods of obscuration indicative of a missed outburst). Despite good time coverage and the fact that V445 Puppis could be identified on many plates at approximately constant brightness (based on visual comparison with nearby stars in the field), we could find no signatures of a previous outburst in the 1897–1955 time frame.

V445 Puppis is located at the low Galactic latitude of b = −2.19°, which at a distance of 8.2 ± 0.5 kpc translates to a height below the Galactic plane of 313 ± 19 pc. At that distance, and that far below the Galactic Plane, one can use the IRAS/DIRBE Galactic reddening maps (Schlegel et al. 1998) to gauge the Galactic reddening toward V445 Puppis. In Figure 7, we show the ratio of the measured reddening (at a given distance) to the total line-of-sight Galactic reddening for open clusters in the Milky Way (Kharchenko et al. 2005) in the Galactic latitude range 1° ⩽ |b| ⩽ 4° and for E(B − V)_Schlegel ⩽ 2.5 as a function of height above/below the plane and distance, respectively. The reddening maps of Schlegel et al. (1998) are not well calibrated at low Galactic latitude; from colors of galaxies discovered at low Galactic latitude behind the southern Milky Way, it appears that the reddening values from Schlegel et al. (1998) are too high (Schröder et al. 2007) and that the true value is 87% of the Schlegel et al. (1998) reddening, e.g., E(B − V)^cal_Schlegel = 0.87E(B − V)_Schlegel.

Zoom In Zoom Out Reset image size

Comparing the location of V445 Puppis to open clusters within 10 deg of V445 Puppis (big filled circles in Figure 7), we deduce that Galactic foreground reddening toward V445 Puppis is probably in the range of E(B − V) = 0.51–0.68 = 0.75–1.0 E(B − V)^cal_Schlegel, as marked by the arrow in Figure 7. The lower limit given here could be a conservative one due to a selection effect; given the relative poor angular resolution of the reddening maps, the open clusters which set the lower ratio limits (those observed at large distances and large height above/below the Galactic plane) are likely to be observed through small-scale patches of relatively lower extinction compared to their surrounding. Finally, taking the equivalent widths of the two interstellar Na D components (Iijima & Nakanishi 2008), a value of E(B − V) = 0.62 is inferred using the calibration of Munari & Zwitter (1997). This value falls within our deduced limits.

Taking the lower limit of the Galactic reddening toward V445 Puppis (E(B − V) = 0.51), we derive a pre-outburst color of V445 Puppis of (V − K)⁰ = 1.58 mag, assuming a standard Galactic extinction law (Cardelli et al. 1989). This is significantly redder than the likely binary progenitors—high $\dot{M}$ AM CVn systems or systems with helium star donors—which typically have (V − K)⁰ ≈ −0.6. A red giant donor (symbiotic nova) can be excluded on the basis of the pre-outburst colors.

This suggests the presence of substantial CS reddening before the 2000 November outburst, which is not unreasonable given the possibility of previous outbursts or material expelled during a common envelope phase. The CS reddening around V445 Puppis does not necessarily have to follow a standard Galactic extinction given the current carbon-rich ejecta; Bergeat et al. (1999) find some deviations from a standard reddening law for CS reddening around a number of carbon-rich R CrB stars. In the right panel of Figure 4, we compare the line profile of the 7065 Å and 2.0581 μm He i recombination lines, obtained close in time and both normalized by peak blueshifted emission. As expected, the redshifted component of the optical line (at +1140 km s⁻¹) appears dimmer compared to its near-infrared counterpart, due to dust obscuration within the shell. For a standard Galactic reddening law (Cardelli et al. 1989), we expect a ratio of peak emission of near-infrared (2.0581 μm) to optical (7065 Å) of 7. The ratio observed in V445 Puppis is 3.4, substantially less and indicative of a low ratio of total-to-selective extinction, R_V ≈ 2.5 (Fitzpatrick 1999). The shell of V445 Puppis offers an opportunity to determine the nature of the dust extinction in carbon-rich outflows. The new X-shooter instrument on the VLT will be ideal to obtain simultaneous optical to near-infrared medium-to-high resolution spectroscopy of both NE and SW shells.

To make the pre-outburst color of V445 Puppis consistent with the intrinsic colors of likely progenitor binaries, a (maximum) additional color excess of A_V − A_K = 2.2 mag requires A_V = 2.5 mag for R_V = 3.1 (standard reddening law), or A_V = 2.8 mag for R_V = 2.5. The uncertainty in A_V introduced by the different reddening laws, however, is small compared to the uncertainty in the bolometric correction needed to arrive at the pre-outburst luminosity of V445 Puppis.

We thus determine a pre-outburst extinction-corrected brightness of at least V⁰ = 12.9 mag (corrected for Galactic reddening only), but more likely V⁰ ≈ 10.1–10.4 mag (including CS reddening correction), with the range in the latter allowing for differences in the CS reddening law. The corresponding absolute magnitudes are M⁰_V = −1.7 and −4.2 to −4.5 mag, respectively. For a range of likely bolometric corrections (BC = −1 to −2.5; Roelofs et al. 2007; Heber & Schoenberner 1981), the pre-outburst luminosity of V445 Puppis is log L/L_☉ = 3.3 ±  0.3 (corrected for Galactic extinction only), or log L/L_☉ = 4.34 ±  0.36 (including CS reddening correction).

It has not escaped our attention that hydrogen-deficient carbon (HDC) stars also occupy this luminosity regime. With a larger distance—this paper—and a larger Galactic reddening correction (Iijima & Nakanishi 2008), the possibility that V445 Puppis belongs to the class of HDC stars can no longer be rejected on luminosity grounds, cf. Ashok & Banerjee (2003). The (Galactic) extinction-corrected (V − K_S)⁰ colors of V445 Puppis are not too dissimilar from the HDC star HD 182040 (Brunner et al. 1998). It is not clear, though, what can cause a luminosity increase of 6 mag and a rapidly expanding shell from a HDC star; the overall light curve appears much better described by an event near a compact, massive white dwarf.

Deviations from the simple outflow model presented in Section 3 could provide clues to the mechanism responsible for shaping the nebula. Although the dynamical model fits the shell remarkably well, the largest deviations occur nearest to the nova remnant at the earliest epochs. The nebula initially had a very narrow waist, followed by a rapid broadening of the waist close to the nova remnant.

To produce very narrow waists in PNe, collimated fast winds (CFWs) are needed in systems with high-density gas in an equatorial plane close to the source of the CFW (Soker & Rappaport 2000). It is unclear what collimated the fast wind in V445 Puppis. At the moment it seems likely that an equatorial disk/torus already existed pre-outburst, given the large pre-outburst CS reddening deduced in Section 4.1. Once the present optically thick dust disk clears (Figure 2—presumably additional material was ejected in the equatorial plane during the 2000 November outburst) and a relatively unobscured view of the nova remnant is possible, alternative origins of the collimated outflow could be investigated, e.g., the presence of a rapidly rotating magnetic white dwarf.

The shell of V445 Puppis is unique for a classical nova. Although many novae show bipolar shells, e.g., the classical nova HR Del (Harman & O'Brien 2003) and recurrent nova RS Oph (Bode et al. 2007; Sokoloski et al. 2008), the shell in V445 Puppis reflects the most collimated outflow seen in any nova. The knots are moving at jet-like velocities of ≳ 8000 km s⁻¹. They can be a consequence of a jet activity about 350 days after the main outburst which is not long after the sudden drop off in the V band is seen (Ashok & Banerjee 2003) and coincides with the radio flare (Rupen et al. 2001). In the symbiotic nova CH Cygni, a drop in V-band magnitude is associated with the onset of a radio jet ejection, see, e.g., Karovska et al. (2007).

At a distance of 8.2 kpc, the maximum brightness at outburst (V = 8.6 mag) is consistent with the Eddington luminosity of a massive white dwarf. Moreover, the pre-outburst luminosity of V445 Puppis of log L/L_☉ = 4.34 ± 0.36 appears to rule out an AM CVn progenitor and may require both a luminous accretion flow as well as a bright donor. For comparison, a 1.3 M_☉ white dwarf accreting at $\dot{M} \sim 10^{-6}\,M_{\odot }$ yr⁻¹ would give an expected accretion luminosity of log L/L_☉ = 3.7. Similarly, the luminosity of a moderately massive, evolved helium star is around log L/L_☉ ≈ 3.7 (Yoon & Langer 2003).

With the distance determined in this paper, a revised mass of the dust shell of 1.5 × 10⁻⁵ M_☉ is derived following Lynch et al. (2004). This is consistent with Kato & Hachisu (2003), but we note that the mass depends strongly on the assumed temperature of the shell and will ultimately be best constrained by mid- to far-infrared data of V445 Puppis recently obtained by Spitzer. This mass is likely to be a lower limit given the large optical depth of the equatorial dust disk.

The various components of V445 Puppis (the dust disk, the bipolar shell) clearly show that the immediate environment of V445 Puppis is sculpted either directly or indirectly by (repeated) outbursts. If V445 Puppis is representative of its class (of helium novae), the helium counterparts to the classical hydrogen-rich novae result in more substantial circumbinary reddening due to the carbon-rich outflow.

Validation of the white dwarf+helium star model as the appropriate binary configuration of V445 Puppis will come from the determination of its orbital period. Unfortunately, this can only be done once the optically thick dust disk surrounding the nova remnant becomes transparent. Given the low inclination of the bipolar outflow, the implied inclination of the equatorial plane which corresponds to the orbital plane of the binary is ∼86°. We thus expect this to be an eclipsing binary, which would further facilitate our ability to constrain the component masses.

There are several indirect suggestions that the white dwarf in V445 Puppis is massive: the maximum luminosity during outburst, the large velocities of the ejected blobs, and its pre-outburst (accretion) luminosity.

The lack of strong soft X-ray emission pre-outburst, as is expected from a supersoft source (Yoon & Langer 2003), could be explained by the substantial interstellar and CS absorption which currently totally obscures the underlying accreting binary. V445 Puppis has not been detected in the ROSAT all-sky survey.

Our derived luminosity is consistent with the massive white dwarf+helium star model of Yoon & Langer (2003), but see also Kato et al. (2008). It is likely that V445 Puppis belongs to a class of binaries that, in principle, could lead toward SNe of type Ia. Given that the expected lifetime of a massive white dwarf+helium star binary is short, these progenitors are associated with the relatively young (∼10⁸ yr) population of Ia SNe.

Whether V445 Puppis itself eventually leads to a SN Ia event, or if the current helium nova has pre-empted that pathway, depends critically on the mass accumulation efficiency on the surface of the white dwarf and the mass ejected during this nova outburst. This remains a topic of debate.