The IR spectral range is also particularly suitable to study the mass loss
process in giant stars. Mass loss is a crucial parameter in any stellar
evolution modelling.
The late evolutionary stages of low- and intermediate-mass
giant stars are strongly influenced by mass loss processes.
Yet, our lack of empirical estimates on mass loss in low-mass
RGB and AGB stars remains one of the most serious stumbling
blocks for a comprehensive understanding of stellar evolution.
A pilot survey of GCs performed with ISOCAM in the 10 
m spectral
region (Origlia et al. 2002) has placed the whole problem into a new
perspective that we can explore in deeper detail now, thanks to the
powerful capabilities of the NASA Spitzer Space Telescope.
Follow-up Spitzer observations have been obtained, aimed at studying
mass loss along the
entire RGB in 17 globular clusters with different metallicities.
Data analysis has been completed, mass loss rates and duty cycles
have been derived for most of the clusters.
First results on 47 Tuc have been published in Origlia et al. (2007).
The results for the other clusters are ready for publication.
A relevant fraction of giant stars show an excess of mid-IR light
above that expected from their photospheric emission.
This is plausibly due to dust formation in mass flowing from these stars.
This mass loss extends down to the level of the horizontal branch and
increases with luminosity.
The mass loss is episodic, occurring in only a fraction of stars at
a given luminosity.
Using the DUSTY code and our observations we derive mass-loss rates
for these stars.
Finally, we obtain the first empirical mass-loss formula
calibrated with observations of Population II stars.
The dependence on luminosity of our mass-loss rate is considerably
shallower than the widely used Reimers law.
This work is in collaboration with F. Ferraro, S. Fabbri (Univ. of Bologna), E. Valenti (ESO, Chile), R.T. Rood (Univ. of Virginia, USA), R.M. Rich (UCLA, USA).
The mass loss phenomenon can be studied in the brightest globular cluster
red giant stars, by comparing the CaII K and H
lines from high
resolution spectra with accurate chromospheric models.
This type of analysis is being performed on several
bright red giant stars in the stellar system Omega Cen,
selected on the basis of their luminosity, metallicity and IR excess.
Preliminary results indicate that in some cases the introduction of
an outward velocity field in the chromospheric model is needed
in order to fit adequately the chromospheric line profiles.
A paper is in preparation.
This work is in collaboration with P. Mauas (Univ. of Buenos Aires, Argentina) and S. Fabbri (PhD student, Univ. of Bologna).
With a fruitful alternative approach, we examined the issue of the second parameter in globular clusters, within the framework of multiple populations. We determined the extreme and median colours of the HB from HST and ground-based photometry available for about 100 GCs. We transformed these values into masses using models from the Pisa Evolutionary Library (PEL), accounting for the evolution. Comparing these masses with those at the RGB-tip we determined the mass loss of stars: a simple linear dependence on [Fe/H] well reproduces the data. By assuming that this is a universal mass loss law, we found that the age is the second main parameter. However, at least a third parameter is required to explain the bluest colours of some GCs; preliminary evidence suggests that this parameter is He. A paper is currently under referee's review.
This work is in collaboration with R. Gratton, S. Lucatello, and V. D'Orazi (INAF-Padova Obs.).