FASE 2: Revisión de normatividad sobre espacio público

can study the internal processes in detail and thereby constrain the physics of stars both on microscopic, atomic, scales and the convective and turbulent motions on macroscopic scales.

Obviously, studies that trace the surface evolution of elements through globular cluster stars in different evolutionary phases, would be biased, or maybe even rendered useless, if the basic assumption of a common initial abundance of all stars is falsified. It is thus important to carefully investigate this aspect. Especially, the abundance spread among stars with similar stellar parameters can lend support to, or refute the hypothesis of a common initial abundance. Such investigations of globular cluster stars have shown a rather complex picture: On the one hand, the star-to-star abundance scatter is always very small in commonly analysed elements like Ca, Ti, and Fe (a notable exception to the

rule is ω cen), seemingly consistent with the simplistic view. On the other hand, light

species like N, O and Na show large star-to-star scatter, incompatible with observational errors, and thus clearly disfavour a common initial abundance (see Sect. 7.1 and references therein). Rather, it is believed that present-day globular clusters host more than one stellar generation, with slightly different initial chemical composition of light elements.

In our investigation of signatures of in-situ depletion of Li in dwarfs, subgiants and giants (Chapter 6), we use Na abundances to distinguish between the different generations and assess possible bias in the interpretations. In Chapter 7 we focus instead on the detailed chemical compositions of a large number of elements in a sample of giant stars, in order to investigate closer the origin of multiple population and constrain chemical evolution of the cluster as a whole.

5.4

High-resolution spectroscopy with FLAMES

To perform our spectroscopic investigations of the globular cluster NGC 6397, we have used the fibre-fed, multi-object instrument FLAMES (Pasquini et al., 2002), mounted on VLT-UT2, at ESO Paranal, Chile. This instrument is ideal for medium-to-high resolution

studies of dense stellar environments, since it is capable of collecting data for>100 targets

simultaneously, located within a total field-of-view of 250. This section will describe the

most important steps in the observational procedure.

5.4.1

The observational setup

We use the UVES+MEDUSA combined mode of FLAMES. In this mode, eight optical

fibres with 100 aperture are fed to the red arm of the high-resolution ultra-violet and visual

echelle spectrograph (UVES, Dekker et al. 2000). The standard settings of UVES cover

a large wavelength region (470–700 nm or 700–1000 nm) with a resolving power of R =

47 000. At the same time, 135 individual fibres of 1.200 aperture are fed to the GIRAFFE

instrument, with a typical wavelength coverage of 30 nm and a resolving power of R =

20 000, in the settings that we used. The fibres are all either allocated to individual stars or placed at random, empty positions around the field-of-view. The latter are so called sky-

fibres and intended to sample the light only from background sources, mainly the reflected moonlight.

The targeted stars are selected from a photometric catalogue of the cluster. The pho- tometry should be as precise and well calibrated as possible, to allow the fibres to be placed centrally on the target, thus maximising the number of photons collected by the CCD detector (CCD stands for charged coupled device). The GIRAFFE CCD consists

of roughly 2000×4000 pixels, with longer of the two being the dispersion direction. The

echelle orders, one for each fibre and target, lie barely separated from each other along the other direction.

5.4.2

Data reduction and processing

The purpose of the data reduction procedure is to collapse the data from the two-dimensional CCD format, to a one-dimensional, calibrated spectra for each target. The calibration steps involve transforming the data from pixel to wavelength space, correcting for the varying sensitivity over the detector, and subtracting background contribution from the instrument itself (see Fig. 5.4). The manipulation of the science exposure and a number of calibration frames are performed in the following order:

• Subtraction of bias, which is a background signal dependent on electronics of the

system. This signal is determined by performing a number of read-outs from zero- time exposures.

• Subtraction of dark current, which is a time-dependent signal stemming from the

non-zero temperature of the detector. This is measured by performing pseudo-

observations without exposing the CCD to a light source.

• Determining the order architecture for the given setup, i.e. a model of how the light

is distributed on the detector. This is achieved by exposing the CCD to a bright, continuous, light source with known properties (i.e. a lamp or a white wall) and create so called flat fields. The order pattern is masked with fitted polynomials, whose parameters are saved for later use.

• Determining the dispersion solution, i.e. the correlation between pixel and wave-

length scale. This is achieved by exposing the CCD with the light from a lamp con- taining excited noble gas particles (thorium and argon). The detected signal shows narrow emission lines with known wavelengths, and can thus be used for calibration.

• The science frame is divided by a normalised flat field frame, which corrects the data

for various instrumental artifacts such as the inhomogeneous illumination, dust on the optical system, varying sensitivity over the detector etc.

• The order architecture is used to extract the two-dimensional data from the science

frame, and convert them to one dimension, simply by summing up the signal across each individual order.