3. Determination of the stellar velocity
3.1. Radial velocity
The single photospheric line Fe I 6546.25 Å, well detected above the noise level, was chosen to obtain a first estimation of the stellar radial velocity. This line was fitted with a Gaussian function in all the individual spectra. From the average Doppler shift of the line a heliocentric velocity -14 km s2 km s-1 was derived. The error, obtained as a standard deviation, indicates significant changes, larger than expected for a single star and for this spectral resolution.
Radial velocities obtained from Gaussian fits to photospheric lines can be uncertain due to asymmetries and/or blending with weaker lines. Therefore, a cross-correlation analysis was performed on a high signal-to-noise spectrum of the object, obtained from an average of several consecutive spectra. Gl820A, a non-active slowly rotating dwarf with a similar spectral type (K5) to that of BD+, was used as a template. The heliocentric radial velocity, computed from the shift of the cross-correlation peak, is =-18.135.08 km s-1. Significant variations were found in both nights. Intrinsic differences between the star and the template may also contribute to the error. J94 obtained a similar result after cross-correlation with several radial velocity standards. The mean heliocentric velocity reported in their work was -17.4 km s2.3 km s-1. For the remainder of the paper, we adopted the most precise of our measurements, -142 km s-1, derived from the Gaussian fitting procedure described above.
In order to examine in detail radial velocity variations the same cross-correlation technique was applied to each individual spectrum in different orders. Results are shown in Fig. 1. It is clear that radial velocity variations are phase dependent and appear to be sinusoidal at the rotational period. The amplitude of the modulation is larger on the night of August 5 (filled circles ). We disregard the possibility that the observed variations are due to an instrumental effect since calibration errors are not significant. Cross-correlation of arc lamp exposures taken throughout the night shows that wavelength calibration shifts could not be larger than 3 km s-1. Even these have been removed to first order by using calibration arcs close to the time of observation.
3.2. Rotational velocity
The rotational velocity was also determined from cross-correlation analysis of the spectra with respect to the spectrum of Gl 820A. A relationship between the width of the cross-correlation peak and vsini was established by broadening artificially the template spectrum to a set of velocities. The synthetic rotational spectra were then cross-correlated with the unbroadened template spectrum. Least-squares linear fits were done to the pairs width-velocity function, providing the calibration needed to convert the measured width to rotational velocity. This method gives vsin682 km s-1. The error is a propagation of uncertainties in the calibration fit. This result is consistent with that of J94, vsin691 km s-1, obtained by using a similar procedure.
© European Southern Observatory (ESO) 1999
Online publication: December 4, 1998