Dust Formation in the Nova V1533 Scorpii

V1533 Scorpii was discovered 2013 June 3.615 by Nishiyama & Kabashima (2013). Pre-discovery images from the OGLE-IV project indicated that brightening was already underway by June 1.367 (Poleski et al. 2013). Visible spectroscopy from multiple observers (Arai 2013; Buil 2013; Kajikawa 2013) indicated that V1533 Sco was an Fe ii type nova. Photometry compiled and reported by the AAVSO indicated that V1533 Sco took 11–12 days to decline by two visible magnitudes (t₂), a rate that places V1533 Sco at the fast edge of the "fast" speed class (Warner 1995).

Consistent with this rapid decline rate was the broad width of the emission lines, which are known to correlate with speed class (Warner 2008). Arai (2013) reported FWHM values ranging from 1800 to 2500 km s⁻¹.

Fast novae frequently do not condense dust but V1533 Sco did. This paper presents infrared spectra that bracket the dust formation period and capture the strong spectral changes that occurred during that time. Preliminary reports on these spectra were given by Rudy et al. (2013a, 2013b).

The infrared spectroscopic data presented here were obtained on the nights of 2013 June 17 and July 16 using the Spex spectrograph of the Infrared Telescope Facility (Rayner et al. 2003). The June observations covered the wavelength range 0.7–2.5 μ; the July measurements extended the wavelength coverage to 5.2 μ. The raw measurements were converted to absolute flux using the Spextools software package (Cushing et al. 2004) together with observations of the A-star calibrator HD 157486.

The panels of Figure 1 illustrate the transformation of the spectrum by dust. The two top panels allow a line-by-line comparison between the two epochs. The bottom left panel includes the MWIR data and illustrates the dominant contribution from hot dust as well as an assessment of the dust temperature. The bottom right panel provides the observational evidence for the presence of molecular hydrogen.

Zoom In Zoom Out Reset image size

While our spectra bracket dust formation, they do not tell when formation actually occurred. For this, we make use of the SMARTS near-infrared photometry (Walter et al. 2012), which began on June 15. In particular, the J − K color shows a pronounced increase beginning the night of our observations (June 17), approximately 14 days after maximum light. Three days later, the the J − K color had increased by 0.5 mag, and we use this date—17 days after maximum light—as when dust formation was well underway. Using the same J − K criterion, dust formation appeared to be largely complete by July 2 (day 29 from maximum light), when the J − K color was more than 2 mag redder than on June 15.

The spectrum of V1533 Sco prior to dust formation shows features of several low ionization species (Na i, Ca ii, Fe ii) and has strong lines of C i. The O i lines λ0.8446 and λ1.1287 that are excited by Lyβ fluorescence are already prominent but have not reached full strength. The O i line λ1.3165, produced primarily from continuum fluorescence and originating in the neutral zone, is strong. These three O i lines can be used to estimate the reddening (Rudy et al. 1991). Employing the reddening law of Schlafly & Finkbeiner (2011), gives E(B − V) = 2.5 ± 0.4, quantifying the result of Kajikawa (2013) who reported a "highly reddened continuum."

The spectrum of V1533 Sco after dust formation is characterized by the fading or disappearance of many emission lines, a general increase in excitation, and the presence of a strong thermal dust component. Among the lines that faded significantly were those of Na i, Ca ii, and Fe ii. The C i features faded significantly but several remained detectable, which is important since some theories of dust formation predict saturation of either carbon or oxygen. While some fading of the C i lines is probably due to depletion of carbon by dust, the increase in excitation can also account for this weakening. The persistence of the O i line λ1.3165 indicates that oxygen, like carbon, is not fully depleted from the neutral region.

This increase in excitation of the ionizing spectrum is typical for novae—as the ejected material expands and dissipates, the ionizing spectrum hardens. The best example of this is the change of the He i lines. Between the two epochs, He i λ1.0830 and λ2.0587 both increased by more than an order of magnitude relative to the H i lines. The O i lines λ0.8446 and λ1.1287 also increased while λ1.3165 decreased, commensurate with a reduction in the neutral zone. Repeating the reddening computation for the three O i lines in the July 16 spectrum gives E(B − V) = 2.7 ± 0.3. This is consistent with little or even no change from the reddening determined prior to dust formation. This does not indicate that little dust was created, but rather that little obscured our line-of-sight and that dust formation was patchy around V1533 Sco.

The lower left panel of Figure 1 presents the overall shape of the continuum and allows a fit to estimate the temperature of the dust. Our best match to the data is with a 870 K that contributes 80% of the flux at 2.3 μ. Unfortunately, the dust emission is featureless over our wavelength range and does not provide a signature for the composition. While carbon rich dust is generally expected from novae at this early stage, the temperature does not exclude dust that is oxygen rich instead.

The lower right panel of Figure 1 presents evidence for a possible detection of molecular hydrogen (H₂). (H₂ is the most common molecule in the universe, but has not, to our knowledge, been detected in novae.) The presence or absence of H₂ in the neutral portions of the emission line regions of novae provide information as to the conditions within those zones. The June 17 spectrum shows two weak emission lines at the approximate locations of the H₂ S(1) and Q(1) features, which, for gas in thermal equilibrium at temperatures of 4000 K or less, are expected to be the strongest features of the molecule. However, both lines are displaced slightly from their laboratory wavelengths, behavior not seen for the C i λ2.2913, which also originates in the neutral zone. There are C i features that fall near S(1), but they are expected to be significantly weaker than λ2.2913 and λ2.1029 (the latter is not detected). The Mg ii λ2.1432 has a companion feature at λ2.4048 that could be the feature labeled as Q(1). It also has a companion λ2.1369 that probably appears on the red wing of S(1) but cannot account for the S(1) feature. Thus while the match to the observations to H₂ is not very satisfactory, there is not a combination of other emission lines that matches better. The evidence for the presence of H₂ emission from V1533 Sco is suggestive but not definitive.

We thank W. Golisch, K. Wagner, and D. Griep of the IRTF assistance in acquiring the data.

This work made use of the Atomic Line List (van Hoof 2018).

Facilities: IRTF - Infrared Telescope Facility, AAVSO - , SMARTS. -