Nova Centauri 1991: A Nova with a Leaky Shell

J. J. Johnson, T. E. Harrison, H. Osborne, D. M. Gelino (NMSU),W. Liller (Vina del Mar), and G. S. Stringfellow (UCo.)

Abstract: We present combined optical and infrared observations of V868 Cen (Nova Cen 1991). We have modeled the circumnova dust shell using the CSDUST3 radiative transfer code. We show that the dust shell became physically thicker with time, indicating nearly constant condensation of grains from outflowing material. We also show that to explain the optical and infrared photometry and the optical spectra, the dust shell(s) must be "leaky", i.e. that the central source is not fully obscured by the circumnova dust shell.

The Nova

Nova Centauri 1991 was discovered by Liller (IAUC 5230) on 2 April 1991 (JD 2448348.64). At that time, the "red" photographic magnitude was 8.7. It seems likely that the initial visual maximum was missed, since the last pre-discovery observation was obtained 15 days prior to discovery. The optical and infrared (IR) lightcurves are shown in Figures 1 and 2. The nova appeared to have a second visual maximum 27 days after discovery and further "sub" maxima later.

The early spectra showed a reddened continuum with comparatively weak emission lines. The strongest were of H-alpha, O I and Fe II (Fig. 8; also Williams et al., 1994). The spectra obtained at the time of the second maximum showed that the emission lines had developed P Cygni profiles, indicating that material had been ejected again during this second outburst.

The Observations

We began observing the nova on JD 2448375. All IR observations (except for JD 2450963, obtained at CTIO using CIRIM) were obtained at the Siding Springs 2.3m telescope. Some optical photometry was obtained using the 1m Mount Stromlo and Siding Springs Observatory, others by Liller, and the remainder were obtained from the literature. The optical spectra were obtained at the Mt. Stromlo 74", except for JD2448358 obtained by M. Gregg at the MSSSO 2.4 m. Based on analysis of the Balmer ratio, and referring to Osterbrock (1989) for the Balmer series and Cardelli, Clayton, & Mathis (1989) for the reddening law, we determined the visual extinction to be A_v = 3.37. We have used this value to deredden the spectra and the photometry. This is much less than the A_v = 5.25 value determined by Williams (1994)

The maximum in K occurred during the visual decline, as happens during dust formation episodes. This is further born out by the reddening of the infrared colors (Figure 2) during the visual decline. Oddly, after the K maximum, the infrared colors stayed nearly constant for ~ 300 days though the K magnitude declined from 4.38 to 9.30. This indicates that the dust shell temperature must not have varied significantly though the luminosity of the nova declined by a factor of 100.

Based on the estimated t₃ and the recent recalibration of the t₃-maximum magnitude relationship by Downes & Duerbeck (2000), we derived M_v = 7.0. Using this value, and the reddening we determined, we find a distance to the nova of D= 5.320 kpc.

The Modeling

Since the photometry suggested that dust formed around the nova, we modeled the spectral energy distribution (SED) of the nova using real grains, rather than assuming a single blackbody distribution. We used the radiative transfer code, CSDUST3, formulated by Egan, Leung & Spagna (1988, and references therein). The code allows us to use the properties of real dust grains, and several different geometries for the circumnova dust shell. After several trials, we discovered that only amorphous carbon grains could reproduce the observations. We have also used a spherical geometry, centered on the radiation source. However, as we shall see, there is evidence that the shell does not have an 100% covering factor and that light escapes through holes in the shell.

The input parameters are: the luminosity (L) and temperature (T) of the central source, and the radius (R) and thickness (both physical, deltaR/R, and optical depth, tau ) of the dust shell, and the size of the dust grains, a. Two of the parameters are constrained, at least initially, by the nova outburst itself. From the absolute magnitude at maximum, we can determine that the initial bolometric luminosity of the nova was L_bol ~50,000 L_o. Of course, this luminosity will change as the outburst progresses.

Additionally, we can use the ejection velocity of v_exp = 392 km/s (determined from the optical spectra), and the time since outburst to determine the radius of the shell.

The Results

Since space is limited, we cannot go into all the details of how we arrived at the final models. In brief, we ran a variety of models, adjusting L, T, tau , deltaR/R, and a. Initially we looked at thin shell models (deltaR/R = 0.01) but found that these models had limited success, especially in later stages of the outburst. We also determined initial values for T by fitting the optical photometry with a blackbody distribution. We used either the optical photometry from the date of our IR photometry, or estimated the values based on extrapolations of the optical data. The grain size was generally tightly constrained, since smaller grains than used in a given model were too hot to fit the data, and larger grains were too cool. The temperature, luminosity and optical depth were constrained primarily by the ratio of optical to infrared flux, which were important even when all we had were limits on the optical data. The shell radius was constrained by the shape of the infrared SED, with broader SEDs corresponding to thicker shells.

The parameters of the models are listed in the table below and specifics of the data are discussed on the next page. The figures show the best representative models

JD 2448377: These data were obtained at the peak of the second visual maximum, 29 days after discovery. We first attempted to fit a reddened stellar distribution to the data, but there was a clear IR excess The best models had the temperature of the central source, T_eff, between 5500 and 6500 K, with a low optical depth and small grains. T_eff is the temperature that the dust grains "see".

JD 2448428: We did not have simultaneous optical photometry for these IR data, so we used the V magnitudes from before and after this date to set limits. Even these generous limits severely constrained our models. The main surprise is that the pseudo-photosphere had cooled to T_eff~ 3000 K. We were unable to find any models with T_eff > 3300 K that could simultaneously fit the IR photometry while staying within the V limits. Both blackbody and CSDUST3 models had this problem. In fact, a simple 2 blackbody model was able to fit the data, though the CSDUST3 models were also consistent with this cooler temperature. The hotter blackbody was reddened (presumably by the second cooler dust shell) by A_v = 0.5. The luminosity is the same as JD8377. We required a cool central source surrounded by a moderate optical depth, cooler dust shell in order to fit the data. Half the light at K is due to the photosphere and half due to the dust shell. There is evidence for further ejection of material between 8377 and 8428 (Williams et al. 1994), which would explain not only the cooler photosphere but also the existence of a thicker dust shell.

JD 2448498: These data were obtained near the infrared maximum, which was also the time of the reddest IR colors. The optical colors were not the reddest at this time, however, indicating that some optical light must be escaping. As with the previous model, we used the V and R magnitudes from before and after this date to constrain the model. The temperature has increased to 4500 K, indicating a changing of the pseudo-photosphere. Hotter temperatures were unable to fit even the rough limits of V and R we had. The luminosity has decreased somewhat. The main change has been the increase in the optical depth of the shell, from tau < 1 to tau = 2.8. The grains have increased in size as well, probably due to the aforementioned second ejection of material.

JD 2448669: These data were obtained almost a year after discovery and 171 days after the previous observation. The most surprising development to the model was that the width of the shell has increased. The shell has increased to nearly half the total radius (i.e., the shell begins halfway between the star and the outer rim of the shell). We were unable to continue using the narrow shell we had successfully used before. If there had been two or more ejection episodes, as suggested by Williams (1994), then the increased width of the shell is easily explained. Different shells travelling at different velocities would naturally lead to a thicker shell. In addition, the grains have decreased in size to 0.2 m. The shell may have been replenished with small grains or the large grains may have been sputtered away (unlikely, given the 9000 K temperature that the dust grains "see"), or, most likely, the shell with the large grains has cooled to the point where we cannot detect the shell in the near-infrared bands. The luminosity of the central source has decreased by a factor of 10, as expected for a year after outburst, but there could be significant light leaking, not intercepted by the circumnova dust shell. Evidence to support this will be discussed below.

The Optical Spectra

In Figs. 8 and 9, we present spectra of the nova obtained on JDs 2448358, 2448462, 2448484, and 2448742. The spectrum from JD8358 is also shown without being dereddened in order to demonstrate the magnitude of the interstellar reddening. Besides hydrogen, the most prominent emission lines are due to He and O, along with Fe II and N lines. As the outburst progressed, H-alpha steadily became stronger while O I weakened, until on the last date, O I is no longer visible, though forbidden O and N lines have appeared.

In spite of the thick dust shells evident in the infrared photometry, the emission lines indicate that there are regions of hot gas in the ejecta. The existence of the forbidden oxygen and nitrogen lines on JD8742, at a time when the IR luminosity was slightly greater than the optical, shows that there are high-energy photons not being absorbed by the dust shell. Williams (1994) argued that the oxygen line ratios can only be explained by denser globules in the ejecta of novae. These globules are natural sites of dust formation, leading to the leaky shell.

In addition, the spectra from JD8462 and JD8484 have relatively cool continua. We fit blackbodies to the continuum of each spectrum and found that JD8462 showed a 3000 K continuum, while JD8484 had a 6000 K continuum (Fig. 10). Though these spectra weren't obtained simultaneously with our IR photometry, they roughly agree with the required central source temperatures from our modelling process.

Our preliminary conclusions are that Nova Cen 1991 has an extensive but leaky dust shell composed of amorphous carbon grains. The central source (the pseudo-photosphere) reached its coolest temperature on 8428, after the secondary maximum. This is consistent with rapid grain growth. The large extent of the dust shell at later dates indicates that the ejection of material was an ongoing process and not just a one-time event. Continual grain formation would lead to an apparent isothermal dust shell as has been observed in other novae e.g. Nova Vul 1978, Ney and Hatfield (1978).