A Search for the Secondary Spectrum of ε Aurigae

Abstract

The observationsThe enigmatic long period (P = 27.1 yr) eclipsing binary, Aurigae, recently emerged from its 2009-2011 eclipse. We have analyzed out-of-eclipse observations (Chadima et al. 2010) obtained over the past 17 years: 306 medium-resolution, high S/N, spectroscopic observations from 6300-6700Å. Of these, 105 spectra were obtained at the Dominion Astrophysical Observatory (DAO) near Victoria, Canada, from 1994-2010, and 201 spectra were obtained at Ondřejov Observatory (OND), from 2006-2010. Analyzing these data, Chadima et al. (these proceedings) reported on a positive, but ultimately spurious, detection of a secondary spectrum. Their attempts at disentangling the binary spectra were foiled by line profile variations of the F star primary. The 6300-6700Å spectral region contains several strong stellar lines but space limitations allow us to present only the results for Si II 6347Å. We examine the centroids and higher moments of this prominent F star spectral line for any evidence of a secondary spectrum. Even if secondary contributions are blended with the F star lines, contamination by the secondary star should produce a centroid shift that is anti-correlated with the orbit of the F star primary. The analysisThe stellar line profiles observed in Aur are quite asymmetric (Figure 1a), and vary on timescales of 50-100 d (Chadima et al. 2011), complicating the detection of radial velocity (RV) variability from the companion. Spectra were transformed to the F star reference frame using the Chadima et al. (2010) orbital solution. Observations acquired after the start of spectroscopic eclipse, taken to be JD 245 4850, were discarded because of strong absorption from the companion's disk. A small number (∼10%) of low S/N observations were omitted. Also, two DAO observations with highly anomalous shifts of tens of km s −1 were rejected. This left 163 spectra: 76 DAO and 87 OND observations. These spectra were then cross-correlated with DAO or OND reference standards, using a spectral region of weak stellar lines (6375-6447Å), to remove intrinsic stellar radial velocity (RV) variations (Figure 1a). The velocity corrections found were typically small (∼ 3 km s −1 ). Then, the first four central moments µ 1 , · · · , µ 4 , were evaluated, and the line centroid µ 1 (Figure 1b), width σ = µ 0.5 2 , and skewness γ 1 = µ 3 /µ 1.5 2 = µ 3 /σ 3 were derived for each observed profile. We also evaluated mean profiles for each quintile of orbital RV, because any profile anomalies associated with the companion should be negatively correlated with the F star's orbital RV, and thus with the RV quintiles.