In recent subsections it was shown that several but not all false alarms can be identified with the help of information contained in online databases or photometric follow-up observations. For remaining candidates a more precise spectral typing is necessary to better evaluate the dimension of the secondary object. Thus taken high or medium resolution spectra can be compared with reference spectra (i.e. Montes et al. 1999). This allows a more precise determination of the size of the secondary object to be able to decide if the object is planet sized or not. Giant planets orbiting are expected to have an upper limit radius of about 1.5 Rjup, but younger planets can be larger, up to 2 Rjup (see Bodenheimer et al. 2003).
If the size of the secondary is estimated to be 2 Rjup or less then measurements of the radial
velocity (RV) have to be carried out to identify whether the object responsible for the detected transit signal is a planet (less than ~ 13 Mjup), a brown dwarf companion (~ 13 Mjup – 80 Mjup)
Chapter 4: Target field selection, observational and follow-up strategy
Figure 4.4.: The 2m Alfred Jensch telescope at the Thüringer Landessternwarte (TLS). This telescope is used for RV confirmation of transit candidates detected by BEST. is nearly constant in the range of 0.5 Rjup to 2.0 Rjup (see Bodenheimer et al. 2000, Pont el al.
2004).
Spectroscopic measurements both for spectral typing and RV analysis for transit candidates detected with BEST are carried out with the 2m Alfred Jensch telescope at the Thüringer Landessternwarte (TLS) Tautenburg (see Figure 4.4.). The spherical primary mirror of the telescope has a focal length of 4m. This multi-purpose telescope can be used in 3 different optical modes: Coudé mode, Nasmyth mode for spectroscopic measurements and Schmidt mode for photometric monitoring. The focal length in Coudé mode is 92m.
The white-beam Echelle spectrograph allows taking spectra with a resolution of up to 67,000. The spectrograph is temperature-stabilized and uses an Echelle grating with 31.6 lines per millimeter. Three different grisms can be used: the UV grism (340nm – 547nm), the VIS grism (463nm – 737 nm) or the IR grism (538nm – 927nm). The highest sensitivity is reached with the VIS grism. With an hour exposure a S/N of 30 per pixel (low resolution spectra) can be reached for stars as faint as magnitude 13. Sufficient RV measurements can be obtained using the temperature stabilized iodine cell to generate a very dense reference system of absorption lines superimposed on the stellar spectrum for wavelength calibration.
For most of the BEST candidates only low-resolution spectra will be available. This allows the RV identification of candidates where the transit-like signal is caused by a brown dwarf or stellar companions. For bright stars (11 mag – 12 mag or brighter depending on seeing and transparency conditions) medium resolution spectra can be obtained allowing determination of the orbital parameters and the mass of Jupiter-mass exoplanets.
More details about the instrument and the report about the first planetary discovery with this instrument can be found in Hatzes et al.(2005).
For faint candidates additional spectroscopic analysis has to be carried out. Collaboration partners at the McDonald Observatory in Texas, USA provide access to the 2.7m Harlan J. Smith telescope and the 9.2m Hobby-Eberly Telescope (HET).
The Harlan J. Smith telescope uses a cross-dispersed 2dcoudé spectrograph (Tull et al. 1995) in combination with an iodine cell for wavelength calibration and monitoring of the instrumental profile very similar to the instrumental set-up of the TLS 2m telescope for RV measurements. Due to the larger main mirror of the 2.7m telescope spectra with higher S/N can be obtained compared to spectra taken with the 2m TLS telescope for the same stars. More details about the 2.7m telescope and its instrumental set-up can be found in Hatzes et al.(2000). Nevertheless the performance does not allow obtaining high-resolution spectra for
most of the BEST transit candidates. A much larger collecting area is necessary to reach a S/N ratio of 100 or better.
A well-suited and well-located telescope for this purpose is the Hobby-Eberly Telescope (HET) at the McDonald Observatory. The HET consists of an array of 91 hexagonal-shaped mirrors (1m diameter) with a resulting effective aperture of 9.2m. The main mirror is fixed in a truss that can be rotated in the azimuth direction only. Tracking is done with fiber optics directly moved in the spherical focal plane. The truss is tilted at 35 deg zenith angle. This construction limits the accessible regions of the sky to objects at declinations between –11 deg and +72 deg, but significantly reduces the technical and financial efforts to built and maintain a telescope of this dimension.
Targets within the BEST target fields with declinations between 46.5 deg and 53.5 deg are observable with the HET telescope. The High Resolution Spectrograph (HRS; Tull et al. 1998) was designed to be able to make RV measurements with a precision of 3 m/s or better for stars as faint as V = 10 mag. A resolving power of 60,000 is used for RV measurements. Two CCDs cover a large wavelength range centered on 594nm. An iodine cell is used for wavelength calibration and analysis of the instrumental profile. For more details about the telescope and the set-up of the HRS refer to Cochran et al. (2004).
If RV measurements confirm the orbital period of the eclipsing object derived from photometric signals and the mass of the secondary object is evaluated to be in the range typical for planets then the transit of an exoplanet is confirmed. If only a few RV measurements can be taken or the RV data is too noisy for an orbit to be fitted to the RV data then some tests must be carried out to rule out a blending scenario.
The spectra can be analyzed for secondary components to rule out that eclipsing binaries are the source of the observed transit-like signals. The first step is to search for directly visible lines from the secondary component of the binary in the spectra. In high-resolution spectra double lines of grazing eclipsing binaries of nearly the same spectral type are visually easily identified. Note that this test can be carried during the analysis of the first spectrum of the primary.
If no secondary components can be detected this way then precise analyses of the shapes of the spectral lines can be provided using several spectra of the primary object. During this bisector analysis one searches for spectral line asymmetries that are caused by a secondary stellar component that cannot be detected directly in high-resolution images or in spectra (e.g. Gray 1992). This secondary component can be an eclipsing binary that is blended by the primary component. The measured spectra are fitted with a modeled spectrum to determine any variations of the line shapes with time. If variations are detected then they have to be analyzed, if they correspond to the periodicity of the photometric signal in order to identify the non-planetary character of the transit-like signal. Some of the OGLE candidates have been identified to be blends by this method. Thus Torres et al. (2004) reported that the candidate OGLE-TR-33 is an eclipsing binary blended by a brighter star located in the same line of sight.