A Representative Sample of Be Stars V
Total Page:16
File Type:pdf, Size:1020Kb
A&A 556, A81 (2013) Astronomy DOI: 10.1051/0004-6361/201220419 & c ESO 2013 Astrophysics A representative sample of Be stars V. Hα variability R. M. Barnsley and I. A. Steele Astrophysics Research Institute, Liverpool John Moores University, Twelve Quays House, Birkenhead, CH41 1LD, UK e-mail: [email protected] Received 20 September 2012 / Accepted 9 May 2013 ABSTRACT Aims. We attempt to determine if a dependency on spectral subtype or v sin i exists for stars undergoing phase-changes between B and Be states, as well as for those stars exhibiting variability in Hα emission. Methods. We analysed the changes in Hα line strength for a sample of 55 Be stars of varying spectral types and luminosity classes using five epochs of observations taken over a ten year period between 1998 and 2010. Results. We find i) that the typical timescale between which full phase transitions occur is most likely of the order of centuries, although no dependency on spectral subtype or v sin i could be determined due to the low frequency of phase-changing events observed in our sample; ii) that stars with earlier spectral types and larger values of v sin i show a greater degree of variability in Hα emission over the timescales probed in this study; and iii) a trend of increasing variability between the shortest and longest baselines for stars of later spectral types and with smaller values of v sin i. Key words. stars: emission-line, Be – circumstellar matter 1. Introduction definition, ones that must have shown emission in the past) had lower rotational velocities and earlier spectral types, an implica- The emission lines displayed by Be stars are known to be tran- tion being that they may be more prone to B/Be phase changes sient (Kogure 1990), with emission varying over timescales (Steele & Clark 2001). Using optical spectrographic follow up ranging from less than a day (Percy et al. 2002) to years taken ten years after the sample was first observed, the validity (Miroshnichenko et al. 2002). This emission variability is re- of these statements will be investigated further in this paper. flected in the first working definition given by Collins (1987) who defined a Be star as “a non-supergiant B star whose spec- trum has, or had at some time, one or more Balmer lines in 2. Observations and data reduction emission”. More recently, this definition has had to be refined to exclude The targets observed were taken from Steele et al. (1999, here- emission-line stars whose disk formation mechanisms were ac- after S99). The catalogue names of the targets and some of their cretion, e.g. Herbig AeBe stars. Consequently, a new subgroup fundamental properties (viz. spectral class, luminosity class, of these emission-line stars, termed “classical Be stars”, was cre- V band mag., v sin i)areshowninTable1. ated. Classical Be stars are characterised by their rapid rotation In S99, the stars were reclassified under the MK scheme and decretion disks (see Porter & Rivinius 2003). Martayan et al. (Morgan & Keenan 1973) using spectra taken by the (2011) proposed a refinement of Collins’ definition to “...astar Intermediate Dispersion Spectrograph (IDS) on the Isaac with innate or acquired very fast rotation which combined to Newton Telescope (INT), rather than relying on values taken other mechanisms ...leads to episodic matter ejections creating from the literature. Rotational velocities were also determined a circumstellar decretion disk or envelope” to not only reflect the in S99 by fitting Gaussian profiles to 4 HeI lines at 4026 Å, larger range of spectral types over which the phenomena have 4143 Å, 4387 Å and 4471 Å and applying the full width half been observed to occur (Conti & Leep 1974; Rauw et al. 2007) maximum – v sin i correlation of Slettebak et al. (1975). It should albeit with lesser frequency (Negueruela et al. 2004), but also be noted that this method of inferring a v sin i has not taken into to constrain the method whereby the disk is formed. It is these account the effect of gravity darkening, which has been shown classical Be stars that are the subject of the following study. to introduce a redundancy between line profile width and v sin i In 1999, a multiwavelength survey of a representative sample at the largest rotation speeds (Townsend et al. 2004; Frémat of 58 Be stars was undertaken (Steele et al. 1999; Clark & Steele et al. 2005), leading to underestimates of the true rotation speed 2000; Steele & Clark 2001; Howells et al. 2001). The sample is for the fastest rotators. Residual emission within the HeI lines “representative” in that it attempted to include objects of each may have also introduced a bias towards lower v sin i for earlier spectral type and luminosity class and so does not serve to reflect subtypes. the spectral type or luminosity class distribution of Be stars. Five epochs of observations were used in the follow- One of the conclusions drawn from this survey was that at the ing analysis (hereafter referred to as the INT, F1, F2, F3 time of observation, non-emission line stars in the sample (i.e. by and F4 datasets) and were obtained using both the IDS and FRODOSpec (Morales-Rueda et al. 2004) on the Liverpool Table 1 is available in electronic form at http://www.aanda.org Telescope (Steele et al. 2004). Article published by EDP Sciences A81, page 1 of 7 A&A 556, A81 (2013) The IDS observations used in this analysis were made Table 2. V band mags and values of EW used as the continuum normal- on the night of 1998 August 3 using the R1200Y grating isation error. with a slit width of 1.15 arcsec, corresponding to a disper- sion of ∼0.5 Å/pixel on the EEV12 CCD. A central wave- Object name VEW(Hα) σ length of 6560 Å was chosen, giving a wavelength range be- HR 5834 5.00 0.3 tween 5800−7100Å. HD 75750 8.48 0.1 HD 100340 10.12 0.1 Reduction of INT data was performed using Figaro accord- ing to the standard prescription for long-slit spectra. Tungsten lamp flats were used to map the slit response and sky flats the (Jones et al. 2011), with any scaling error in determining the pixel-to-pixel flat field variations. Target spectra were extracted level of the continuum leading directly to a multiplicative error using simple extraction and a sky region of the same size was in the EW. To obtain an empirical estimate of this uncertainty, sets of repeated observations of B stars with different bright- also extracted and subtracted from each target spectrum. To = , . wavelength calibrate the target frames, arc exposures were taken nesses spanning the sample range (V 5 8 5 and 10) were at the start, middle and end of the night using CuAr and CuNe taken. These observations were taken sequentially on the same lamps. After extracting their spectra, the calibrations were deter- night with both FRODOSpec’s resolution configurations, using mined and copied onto the target spectra. corresponding exposure times similar to those used to observe the sample. All FRODOSpec observations were taken over a period be- α tween September 2009 and November 2010. Observations for The standard deviation of the measured H absorption EWs the F1 dataset were made between 08/09/2009 and 26/09/2009, was calculated for each set of target spectra. The V band mags. F2 between 03/11/2009 and 26/12/2009, F3 between 18/07/2010 and the measurement errors for each target are shown in Table 2. and 29/08/2010 and F4 between 10/10/2010 and 23/11/2010. These errors were associated to all measurements of EW, The first two epochs of data, F1 and F2, were taken using matching them to the nearest target brightness. In the absence the red diffraction gratings. The last two, F3 and F4, were taken of equivalent INT observations, IDS spectra were assigned the using the higher resolution red VPH gratings. FRODOSpec pro- same error figures as those used for FRODOSpec. In analysis requiring the photospheric contribution to be subtracted off,the vides wavelength coverage from 3900−5700Å (blue arm) and − errors in measuring the EW and the error in assigning a photo- 5800 9400Å (red arm) for the lower resolution configuration spheric EW were added in quadrature. = − (R 2600 and 2200 respectively) and 3900 5100Å (blue arm) Before EW measurements of the Hα linecouldbeusedto and 5900−8000Å (red arm) for the higher resolution (R = 5500 assess if an object exhibited complete disk loss/formation, it was and 5300 respectively). FRODOSpec reduction was performed required that the photospheric contribution to the EW from the using the pipeline discussed in Barnsley et al. (2012). central star, Wphot, be subtracted off the measurement of the total EW of the line, Wtot,i.e. 3. Method Wdisk = Wtot − Wphot (1) Emission in the Balmer series is commonly used to provide an where W is the EW of the emission line arising from the disk insight into the circumstellar environment surrounding Be stars disk only. (Grundstrom & Gies 2006) and results from recombination of W can be established by either using models of stellar photospheric radiation within the disk. In this paper, only the phot atmospheres (e.g. Kurucz 1979) or by measuring the EW for line strength of Hα is measured. Hα is observed in Be stars with a set of B stars (with no history of emission) over a range of a variety of profile shapes (Banerjee et al. 2000), ranging from spectral types and luminosity classes and assuming a linear fit single and double peaked emission to absorption, where no iden- to EW and spectral type for each luminosity class.