next previous
Up: Spectropolarimetry of magnetic

2. Observations and reduction

The observations reported here were performed with the ESO 3.6 m telescope and its Cassegrain Echelle Spectrograph (CASPEC) on the following dates: May 21-25, 1991; June 7-9, 1992; August 10-13, 1992; February 7-8, 1993; and July 24-27, 1993. The Zeeman analyzer built into CASPEC was inserted in the light path to separate incoming light of right and left circular polarization (RCP and LCP, resp.), and the spectra corresponding to both polarizations were recorded simultaneously, interleaved as described in Paper I.

The configuration of the Zeeman analyzer used in 1991 and 1992 is identical to the one that had been employed between 1985 and 1988 to record the spectra from which the magnetic results presented in Papers I to V had been derived. A detailed description of this configuration has been given in Paper I. It appears useful to summarize it here for understandability of the differences introduced in the instrumental setup after 1992. Namely, the analyzer was originally a self-contained unit mounted behind the entrance slit of the spectrograph on the wheel also bearing the "rear slit viewer'' used to center the target in the slit at acquisition time. This wheel is remotely controlled via the instrument computer, so that the analyzer can be called into the light beam at exposure definition time through the observation interface. Within the analyzer, the light beam passes consecutively through:

-
a first set of lenses, to convert the diverging beam into a parallel beam;
-
an achromatic quarter-wave plate, to convert incoming opposite circular polarizations into mutually orthogonal linear polarizations;
-
a Wollaston prism, to separate these two orthogonal linear polarizations;
-
a second set of lenses, to bring back the two emergent light beams to the original focal ratio of the incoming beam. In principle, this design guarantees that the focus of CASPEC is unaffected by the presence of the Zeeman analyzer in the light beam. In practice, a small focus offset is introduced by the Zeeman analyzer, which is easily accounted for when defining an exposure by introducing the appropriate correction of the (remotely controlled) collimator position through the observation software.

The two light beams coming out of the Zeeman analyzer have polarization directions resp. parallel and orthogonal to the grooves of the echelle grating, so that they are reflected by the latter with different efficiencies. This results in differences in the continuum level between the spectra corresponding to incoming light of RCP and LCP. This difference is compensated for at reduction time by renormalizing the spectra of both polarizations to the same continuum level of 1. This is acceptable because we are interested only in the polarization of spectral lines in objects whose continuum has no significant circular polarization.

From 1993, the Zeeman analyzer of CASPEC has been modified as followed (the February 1993 observations reported here were the first test of this modified configuration):

-
within the self-contained unit borne by the "rear slit viewer wheel'', the quarter-wave plate and the Wollaston prism have been interchanged;
-
an additional achromatic quarter-wave plate has been installed in a filter wheel located between the spectrograph entrance slit and the "rear slit viewer wheel'', whose original purpose is to hold interferential filters used to select a single echelle order in view of recording long-slit (not cross-dispersed) spectra of extended objects. As the other functions of CASPEC, this "long-slit filter wheel'' is remotely controlled through the observation software.

In this new configuration, the stellar light first goes through the quarter-wave plate located in the "long-slit filter wheel''. As a result, incoming circular polarizations are converted to mutually orthogonal linear polarizations. These linear polarizations are separated by the Wollaston prism of the unit borne by the "rear slit viewer wheel''. The two resulting light beams are converted back to circular polarization by the quarter-wave plate following the Wollaston prism within the analyzer unit. This allows them to be transmitted with the same efficiency by the subsequent elements of the optical train.

There are a number of additional differences of instrumental setup between the various observing runs reported in this paper, and with respect to the homogeneous set of observations discussed in Papers I to V. They will now be described.

In May 1991 and in June 1992, we used the same camera ("short camera'', f/1.6) and echelle grating (52 lines/mm) as between 1985 and 1988. But the CCD RCA #3 (tex2html_wrap_inline2625 pixels of tex2html_wrap_inline2627 tex2html_wrap_inline2629m2) that we had previously employed was replaced by the CCD Tektronix #16 (tex2html_wrap_inline2633 pixels of tex2html_wrap_inline2635 tex2html_wrap_inline2629m2). As a result, a slightly higher spectral resolution was achieved (tex2html_wrap_inline2641), and a significantly longer spectral range was recorded overall (from 5400 Å to 6800 Å), although at the expense of somewhat larger wavelength gaps between consecutive echelle orders.

In August 1992, the "long camera'' (f/3, Pasquini 1993) of CASPEC was for the first time used in combination with the Zeeman analyzer (still in its old configuration). The larger interorder separation then achieved on the CCD allowed the use of the 31.6 lines/mm echelle grating without overlapping of opposite polarizations of consecutive orders (see Paper I). The efficiency of this grating in the spectral region of interest is higher than that of the 52 lines/mm grating, which compensates to some extent for the increased slit losses (for full exploitation of the resolution achievable with the long camera, tex2html_wrap_inline2645, the slit width was set to 105, compared to 21 with the short camera). Also, all other things being equal, the use of the 31.6 lines/mm grating would reduce the interorder gaps in the wavelength coverage, compared to the 52 lines/mm. However, with the small size of CCD #16 (which was mounted for the considered run), the increase in the geometrical scale between the short and the long camera more than cancels this advantage, since a large fraction of the whole cross-dispersed echelle pattern falls outside the detector boundaries. As a result, the spectra range only from 5710 Å to 6450 Å, with gaps of 17 Å to 27 Å between consecutive orders.

The first tests of the modified configuration of the Zeeman analyzer in February 1993 were carried out with essentially the same setup as used in May 1991 and June 1992: 52 lines/mm echelle grating and short camera. This time, the CCD was #32, a Tektronix chip having the same format as #16 but a higher efficiency in the blue-violet part of the spectrum: that is, this detector was essentially equivalent to #16 for our purpose.

CCD #32 was also used for the observations of July 1993, together with the configuration of the Zeeman analyzer introduced in Februray 1993, the 31.6 lines/mm echelle grating, and the long camera of CASPEC.

This latter configuration is closest to the present standard setup of CASPEC for observations with the Zeeman analyzer. The only difference is that CCD #32 has been replaced by #37, a Tektronix of tex2html_wrap_inline2647 pixels of tex2html_wrap_inline2649 tex2html_wrap_inline2629m2. With this setup, spectra can now be recorded at a resolving power tex2html_wrap_inline2655 over the range tex2html_wrap_inline2657, without gaps in the wavelength coverage. This configuration is mentioned here only for completeness of the record. Using it, we are currently regularly obtaining observations for a new programme, the results of which will be presented in a future paper.

The various configurations of CASPEC and of its Zeeman analyzer that we have employed for studies of stellar magnetic fields are summarized in Table 1 (click here). The dates at which each configuration was used are given in Col. 1. The two configurations of the Zeeman analyzer (ZA) itself, which have been described above, are called "old'' and "new'' in Col. 2. The echelle grating is referred to in Col. 3 by its number of lines per millimeter. Since 1991, CASPEC users have been offered the choice between two cross-dispersers (X-disp.), a "blue'' one and a "red'' one, which are optimized to give a better efficiency resp. at blue or at red wavelengths. Although the red cross-disperser should be slightly more efficient than the blue one in the spectral range covered by our observations, it is unsuitable for our purpose: when using it, the echelle orders are too packed, and opposite polarisations of consecutive orders would overlap when the Zeeman analyzer is inserted in the light beam. Accordingly, we have always used the blue cross-disperser, as indicated in Col. 4 of Table 1 (click here). Columns 5 and 6 of this table give the camera used (short or long), and the CCD (referred to by its internal ESO number).

    Table 1: Configurations of CASPEC and of its Zeeman analyzer used to study stellar magnetic fields

The data reduction steps have been described in Paper II. The procedure applied to the present observations differed from the one used for the 1985-1988 data only in two (minor) aspects:

-
while a specific FORTRAN code had been written for the reduction of our former spectra, here we used MIDAS procedures built from existing MIDAS commands. This simplification (from the point of view of the user) was made possible by the use of more powerful computers (Unix Sun workstations instead of a VMS Vax 11/750) and by the increased versatility and robustness of the MIDAS echelle package, which permitted us to use standard commands in spite of the non-standard format of the interleaved right and left circularly polarized spectra;
-
we gave up using an automatic procedure for normalization of the spectra to the continuum, which we did not always find fully satisfactory. Instead, a polynomial was fitted to the high points of each order, defined interactively by cursor entry.

    Table 2: Programme stars
References: 1: Mathys et al. (1996) (MHLLM); 2: Manfroid & Renson (1994); 3: Kurtz et al. (1996); 4: North (1990) (unpublished; cited by Catalano et al. 1991); 5: Mathys (1991) (Paper II); 6: Pedersen (1979); 7: Waelkens (1985); 8: Kurtz et al. (1992); 9: Hensberge (1993); 10: this paper; 11:  Babcock (1960); 12: Kurtz et al. (1994); 13: Mathys & Lanz (1996); 14: Kurtz (1989); 15: Catalano & Leone (1993); 16: Leroy et al. (1994); 17: Wolff (1975); 18: Kurtz & Marang (1987); 19: Preston (1970); 20: Lanz & Mathys (1991); 21: Babcock (1958); 22: Bohlender et al. (1987); 23: Landstreet (unpublished; cited by Mathys 1991); 24: Wolff (1969).


The reduced spectra have signal-to-noise ratios ranging from 100 to 300 (most of them between 150 and 200). The rms deviation of the lines of the ThAr arc spectra used for wavelength calibration with respect to the fitted dispersion law is of the order of 3 to 3.5 mÅ with the short camera, and of 1.5 to 1.8 mÅ with the long camera.

The stars for which magnetic field measurements are reported here are listed in Table 2 (click here). The presentation is identical to Table 1 (click here) of Paper II (or Table 2 (click here) of Paper III): the HD (or HDE) number and another identifier in Cols. 1 and 2, the V magnitude and the spectral type (both from the catalog of Renson et al. 1991) in Cols. 3 and 4, the adopted value of the rotation period P and the reference from which it is retrieved in Cols. 5 and 6, the heliocentric Julian date taken as the origin of the phases, the particular property (e.g. a magnetic field extremum) from which this origin is defined, and the corresponding reference in Cols. 7 to 9.


next previous
Up: Spectropolarimetry of magnetic

Copyright by the European Southern Observatory (ESO)
web@ed-phys.fr