HST photometry of 47 TUC and analysis of the stellar luminosity function in Milky Way clusters

Abstract

We present V and I photometry for over 1000 stars in a region 5 arcmin from the centre of the globular cluster 47 Tuc. The field was imaged with the Wide Field and Planetary Camera 2 (WFPC-2) as part of the Hubble Space Telescope's Medium Deep Survey key project. The luminosity function (LF) continually rises in the domain 5<~<e1>M_I<~</e1>9 with a slope Delta log Phi(M)/Delta(M)~0.15 and then drops off sharply. We compare our LF with that derived by De Marchi & Paresce for a neighbouring HST field. The two independent LFs are remarkably similar in the entire range of luminosities probed (<e1>M_I<~</e1>10). Comparisons are also made to other HST LFs derived by several authors for both globular and open clusters in the Galaxy. We use the KS test to assess the significance of the differences found. The luminosity distributions obtained with HST are consistent with being derived from the same population down to M_I~9.0. Beyond that, statistically significant variations arise. Globular cluster LFs also differ according to the prominence of a plateau in the bright end (5<~<e1>M_I<~</e1>6.5). The mass functions are rather uncertain and sensitive to the mass-luminosity relation. Different approaches to deriving the 47 Tuc mass function from its LF lead to markedly different results at the low-mass end. For M>~0.4 M_, the 47 Tuc mass function is significantly different from that of omega Cen. The calibrated HST M_V(V-I) colour-magnitude diagrams (CMDs) show a trend with metallicity in the expected sense of systems that are more metal rich having redder CMDs for a fixed absolute magnitude. The main-sequence slope becomes shallower with increasing metallicity. The CMDs derived from HST are in general agreement with previous ground-based studies, especially for metal-rich stars. However, the CMDs of metal-poor subdwarfs observed from the ground are shallower than those of globular clusters observed with HST. This discrepancy may either result from calibration problems of HST data or reflect real differences between the CMDs of globular cluster and field halo stars.