1 Introduction

In the last few years we have witnessed a fantastic development in an "old'' but until recently not very successful field of astrophysics: the search for extra-solar planets. Following the first success in exoplanet searches with the discovery of the planet around 51 Peg (Mayor & Queloz 1995), the number of known giant planets orbiting solar-type stars did not stop growing, being currently of 72 (including 7 multi-planetary systems).

Unexpectedly, the planets found to date do not have much in common with the ones in our own Solar System (for a review see Marcy et al. 2000; Udry et al. 2001; or Mayor & Santos 2001). One remarkable characteristic appears to be related with the parent stars themselves: stars with planetary companions are considerably metal-rich when compared with single field dwarfs (Gonzalez 1998; Santos et al. 2000; Gonzalez et al. 2001; Santos et al. 2001a, 2001b). To explain the observed difference two main explanations have been suggested. The first and more "traditional'' is based upon the fact that the more metals you have in the proto-planetary disk, the higher should be the probability of forming a planet (see e.g. Pollack et al. 1996 for the traditional paradigm of planetary formation). Thus, in this case the "excess'' of metallicity is seen as primordial to the cloud that gave origin to the star/planet system. "Opposing'' to this view, the high metal content observed for stars with planets has also been interpreted as a sign of the accretion of high-Z material by the star sometime after it reached the main-sequence (e.g. Gonzalez 1998; Laughlin 1996).

Figure 1: Spectra in the Be II line region (dots) for two of the objects observed, and three spectral synthesis with different Be abundances, corresponding to the best fit (solid line) and to changes of $\pm$0.2 dex, respectively.

Figure 2: Spectrum in the Be II line region (dots) for 14 Her (HD 145675), and two spectral synthesis with different Be abundances, corresponding to no Be (solid line) and to $\log{N(\rm Be)}=0.5$. Error-bars represent the photon noise error. A conservative upper limit of 0.5 for the Be abundance was considered.

Although recent results seem to favor the former scenario as the key process leading to the observed metal richness of stars with planets (Santos et al. 2001a, 2001b; Pinsonneault et al. 2001), signs of accretion of planetary material have also been found for some planet hosts (e.g. Israelian et al. 2001a; Laws & Gonzalez 2001). The question is then turned to know how frequent those phenomena happen, and to how much these could have affected the observed metal contents.

One possible and interesting approach to this problem may pass by the study of one particularly important element: beryllium (Be). Together with lithium (Li) and boron (B), Be is a very important tracer of the internal stellar structure and kinematics. Be is mainly produced by spallation reactions in the interstellar medium, while it is burned in the hot stellar furnaces (e.g. Reeves 1994). While most works of light element abundances are based on Li (the abundances of this element are easier to measure), Be studies have one major advantage when compared with Li. Since it is burned at much higher temperatures, Be is depleted at lower rates than Li, and thus we can expect to measure Be abundances in stars which have no detectable Li in their atmospheres (like intermediate-age late G- or K-type dwarfs). In fact, for about 50% of the known planet host stars no Li was detected (Israelian et al. 2001b). Furthermore, Li studies have shown the presence of a significant scatter for late-type stars of similar temperature. This has also been observed in open clusters where all stars have the same age, and appears to be related to the clusters's age, rotational velocities, pre-Main Sequence history, etc. (see, e.g., García López et al. 1994; Randich et al. 1998; Jones et al. 1999), a fact that may complicate or even preclude a comparison.

Given all these points, Be studies of planetary host stars can indeed be particularly important and telling. For example, if pollution has played some important role in determining the high-metal content of planet host stars, we would expect to find a similar or even higher increase in the Be contents. This is basically due to the fact that planetary material is relatively poor in H and He when compared to the star (e.g. Anders & Grevesse 1989). Furthermore, and unlike for iron, Be may be already a bit depleted in the stellar surface. Thus, the injection of planetary material into this latter could even be responsible for a more important abundance change in the Be abundance than in the iron content. If pollution has indeed played an important role, the net result of the fall of planetary material into the central "sun'' would thus be that, for a given temperature interval, we should find that planet hosts are (in principle) more Be-rich that non-planet host stars. In other words, the analysis of Be abundances represents an independent way of testing the pollution scenario.

García López & Péres de Taoro (1998) carried out the first Be measurements in stars hosting planets: 16 Cyg A and B, and 55 Cnc, followed by Deliyannis et al. (2000). In order to continue to address this problem, we present here a study of Be abundances in a set of 29 stars with planets, and a smaller set of 6 stars without known planetary companions. In Sects. 2 and 3 we present the observations and analysis of the data and in Sect. 4 we discuss the results in the context of the planetary host stars chemical abundances, but also in terms of the Be depletion processes. We conclude in Sect. 5.