2. Observations and reductions

We carried out observations of stars in the region of the open cluster Westerlund1 through the V and I passbands during a photometric night on July 1995. The direct images were obtained with the 24-inch telescope of the University of Toronto Southern Observatory situated at Las Campanas Observatory, Chile, equipped with a PM METACHROME CCD coated to give improved blue response. The scale on the chip is  per pixel, and consequently the sky area covered by a single frame is about 44.

Table 1 (click here) summarizes the observational data, while V and I images of Westerlund 1 are provided in Figs. 1 (click here)a and 1b, respectively. Notice that the I image is much deeper even when obtained with an exposure time a factor thirty shorter than that of the V image, because of the smaller reddening effects.

  
Figure 1: Image of Westerlund1 (north is up and east is to the right). The field is : a) V filter (900 s), b) I filter (30 s)

   

Date Time Filter Exposure FWHM

(UT) (UT) (secs)

July 2, 1995 V 600 1.3

July 2, 1995 V 900 1.4

July 2, 1995 I 10 1.4

July 2, 1995 I 30 1.5

July 2, 1995 I 30 1.5

Observations of twelve standard stars located in the fields SA 107, SA 108 and SA 112 (Landolt 1992) were obtained in each filter at the beginning, middle, and end of the night, respectively. In addition, a series of bias exposures, flat-field frames on the twilight sky and dome were taken.

The observations have been reduced using the facilities at the Observatorio Astronómico of the National University of Córdoba. In order to clean the images by removing their instrumental signatures, we have applied the averaged bias and flat-field corrections with IRAF standard routines. The average twilight and dome flat-fields were also used to check possible effects due to non-perpendicularity of the optical path on the chip; no correction was necessary.

To obtain instrumental v and i magnitudes, we have used the DAOPHOT (Stetson 1991) package in the standard way on the IRAF environment. Aperture photometry of standard stars was performed with an aperture radius equal to 16 pixels (). The following relations between the instrumental (lower case letters) and standard (capital letters) colours and magnitudes were adopted:

where X_v, X_i are the airmasses corresponding to the observations in the V and I filters, respectively. The rms error affecting the calibration in Eq. (1) is 0.009 mag, while the corresponding one in Eq. (2) is 0.011 mag. The inversion of these expressions allowed us to transform the instrumental colour and magnitudes of the stars measured in the field of Westerlund1 to the standard VI Johnson-Cousins system.

At this stage, we have obtained three independent V,(V-I) table sets available upon request, together with the V frame coordinates (X and Y) and the instrumental DAOPHOT.ALLSTARS rms errors. Then, we performed a cross-correlation identification between all V,(V-I) tables averaging V and (V-I) values after applying appropriate offsets to the coordinates. The results are an improved set of colour and magnitude values as compared to those based on a single measure. Consequently, for stars with two or three measures, we could minimize possible anomalous values caused by the presence of contamination (e.g., cosmic rays). This procedure also allowed us to estimate the photometric internal errors (). Typically  mag and  mag for V<15 mag, increasing to 0.04 and 0.05 at , respectively.

The integrated spectroscopic observations were carried out with the 2.15 m telescope at the Complejo Astronómico El Leoncito (CASLEO, Argentina) during a run in May 1995. We employed a CCD camera containing a Tektronics chip of  pixels attached to a REOSC spectrograph, the size of each pixel being . The slit was oriented in the east-west direction and the observations were performed by scanning the slit across the objects in the north-south direction, in order to get a proper sampling of cluster stars. The total field along the slit was , allowing us to sample background regions. We used a grating of 300 grooves mm^-1 in two different set-ups, namely "blue nights" and "red nights": (a) during the blue nights we obtained spectra covering a range from 3500 to 7000 Å, with an average dispersion in the observed region of or 3.46 Å/pixel. The slit width was resulting a mean resolution of 14 Å, according to the FWHM of the He-Ar comparison lamps. (b) In the red nights the range was 5800-9200 Å with a similar dispersion (3.36 Å/pixel). The slit width was resulting a mean resolution of . An OG 550 filter was employed to eliminate the second order contamination.

A series of exposure of 15 minutes each was employed for Westerlund1 and 2 giving a total of 60 minutes in each cluster for the blue and red spectra. In addition to the observations of the cluster fields, we measured spectrophotometric standard stars in order to derive flux calibrations. In the blue range we used LTT 4364, EG 274 and LTT 7379 (Stone & Baldwin 1983). In the red range we added HD160233 (Gutiérrez-Moreno et al. 1988), which is a hot dwarf star useful also to correct for telluric absorption bands. We also took frames of He-Ar comparison lamps between or after object observations, bias, dome and twilight sky and tungsten lamp flat-fields.

The reductions were carried out with the IRAF system in the standard way at the Observatorio Astronómico of the National University of Córdoba. In summary, we subtracted the bias and used flat-field frames - previously combined - to correct the frames for high and low spatial frequency variations. Then, we performed the background sky subtraction using pixel rows from the same frame, after having cleaned the background sky regions from cosmic rays. We controlled that no significant background sky residuals were present on the resulting spectra. The cluster spectra were extracted and wavelength calibrated by fitting observed He-Ar comparison lamp spectra with template spectra. The rms errors involved in these calibrations are 0.72 Å (0.21 pixel) for the blue nights and 0.39 Å (0.12 pixel) for the red ones. Finally, we applied extinction corrections and flux calibrations derived from the observed standard stars to the cluster spectra. In addition, cosmic rays on the cluster spectra were eliminated. Finally, we eliminated the telluric absorption bands in the near-IR, following the procedures outlined in Bica & Alloin (1987).

In Figs. 2 (click here)a and 2b we present the calibrated integrated spectra of Westerlund1 and 2, respectively. Notice the very steep spectrum in the near-IR and essentially null flux shortward of 5000 Å for Westerlund1, revealing a dramatic reddening effect.

  
Figure 2: Observed integrated spectra in absolute units: a) Westerlund1, b) Westerlund2