3. The observational data
3.1. The UV color excess: detection and photometry of three objects
The images in the filters F255W and F336W (U) were used to construct the UV-CMD which is the most suitable diagram to detect UV-bright sources in a background dominated by cool stars. Fig. 2 displays the ultraviolet CMD of the very central region of the cluster, based on the PC1 data set only.
The most striking feature in the UV-CMD is the locus defined by the "horizontal branch", which runs across the diagram and ends sharply with no extension at higher temperatures and lower luminosities but a few objects. The second remarkable feature is the sequence of blue straggler stars which has been analysed in Ferraro et al. (1997a). The bright, UV-bright candidate Post-AGB star at (1.0,14.0), will be discussed in a forthcoming paper (Ferraro et al. 1997b). The CMD is expected to be substantially complete to .
Three objects (#6359, #7880 and #7290) appear bluer than and lie significantly outside the main locus of cluster stars in the UV-CMD. They are plotted as empty triangles in Fig. 2. They are well above the limit from the sky background, and the formal errors computed for all three stars (illustrated in the same diagram) show that their color is reliable. Their detection in every frame and at every wavelength indicates that they are real features. Their position is circled and labelled in Fig. 1.
The relatively bright star #6359 has a counterpart #1105 in the catalogue of the pre-refurbishment WF-PC1 data of Guhathakurta et al. (1994, Tables 2 and 3; hereafter GYBS). The two faintest stars are new detections since they are listed neither in GYBS nor in the deep ground-based catalogue of Ferraro et al. (1997e). The stars of Fig. 2 are also plotted (Fig. 3) in the CMD to allow an easier comparison with HST or ground-based CMDs of M 3 as well as with the CMD's of other clusters.
Though in this paper we limit the discussion to the PC1 data, the same procedure to select candidate blue objects has already been applied to the WF2, WF3 and WF4 fields. All preselected UV-bright, low luminosity candidates turned out to be either spurious events (mainly cosmic rays) or too faint to be statistically reliable and they did not survive the specific requirements listed in Sect. 2. Therefore, within the completeness, the low luminosity blue stars are also very rare in the three surrounding chips covering about 12 times the PC1 area. This conclusion is also strengthened by a simple simulation showing that in the WF's (which have half the PC1 angular resolution) any star like the faintest blue PC1 stars #7290 and #7880, which both have a very close neighbour red main sequence star, would have merged into composite systems that would still have been selected as photometrically peculiar in the UV-CMD.
The already noted scarcity of such stars in the ground-based CMD (Buonanno et al. 1994) obtained in the external regions is also confirmed here for the cluster central region. In this respect, M 3 is remarkably different from M13, a very similar cluster with the same metal abundance which exhibits a very populated blue HB extension in its core (Ferraro et al. 1997c) and in the outer regions (Paltrinieri et al. 1998).
3.2. First hints on their nature
In the previous section we have found that there are very few faint blue objects in the cluster core of M 3 and that they are all confined within from the cluster centre . This peculiar location strongly suggests that they are cluster members, though background extragalactic sources or foreground stars cannot be a priori excluded.
Background sources can be quasars as inferred from the spectral identifications of optical counterparts of radio sources at 20 cm (Harris et al. 1992; Carney 1976) within from the centre. The fact that there is no spectral information on the radio source in M 3 (which in addition does not show any pulsation, Kulkarni 1997, private communication) makes difficult to exclude that the source is a quasar. To further investigate this possibility in Fig. 4 we compare the colors of the blue sources identified here with the colors expected for a redshifted quasar computed by folding a representative composite spectrum in the 900-8000 Å range (Cristiani & Vio, 1990) with the PC1 filters (including red leaks). Although the colors of a QSO may deviate from that of the composite spectrum used here, the three objects cited above do not occupy the region of the color-color diagram representative of a low redshift QSO. Though not plotted here for clarity, none of the 3064 PC1 field objects which have a UV magnitude occupy the quasar region neither, suggesting that if the radio source is actually a low redshift quasar, the presently undetected optical counterpart is likely fainter than . The apparent lack of objects with colors consistent with those of a quasar all over the PC1 field, probably rules out also the possibility of a point-like complex radio structure associated to an optical counterpart but with a significant offset, as illustrated for quasars of the cluster outskirts (Harris et al. 1992).
In Fig. 4 the three objects (which in the following we will consider as stars -either single or multiple) are located above the main sequence locus which is derived from integration of the models of stellar atmospheres (Kurucz 1992) in the WFPC2 filters bands. The models are for main-sequence stars and for a normalized logarithmic metal abundance relative to the sun of -1.5. The curve was slightly translated along the color, to fit the region of the low luminosity, presumably unevolved blue stragglers. All the three stars occupy the typical UV excess region of faint blue stars.
The local space densities and pc-3 measured by Downes (1986) for the most abundant field faint blue stars namely, O-, B-type subdwarfs and hot white dwarfs respectively, would lead to less than hot white dwarf in the PC1 field of view, within the volume of space defined by twice the scale height of 325 pc (90% of the population) derived by Green et al. (1986). The negligible contributions expected from any stellar species of the field, coupled with the evidence that the stars are towards the very cluster centre, indicate that they are probable cluster members. Table 1 summarizes the photometry for all these stars.
The two brightest UV stars (#6359 and #7880) appear to lie on the extension of the HB which is better visible in Fig. 2 of Ferraro et al. (1997c) in which the CMD of M 3 is compared with that of M13, obtained with the same set-up. Once the CMD of M 3 is shifted to match the main loci in the M13 one, these two stars perfectly merge into the M13 HB blue tail. For this reason they are very likely normal HB stars, with a very thin residual hydrogen envelope (Dorman et al. 1993). Since they are only found so close to the cluster centre, one could speculate that star interactions occurring within the very inner region of the clusters could be somehow related to the origin of the HB blue tail population (Fusi Pecci et al. 1993), at least in some clusters as discussed in Ferraro et al. (1997c).
The object #7290 is located outside the main loci and shows a substantial UV excess. Its position is similar to the two faint UV-stars recently discovered in the core of M 13 (Ferraro et al. 1997d), which have been found to be possibly connected to the X-ray source detected in that cluster. In fact, star #7290, though slightly cooler (, has nearly the same UV absolute magnitude () than the brightest UV-star in M13. Assuming a distance modulus of 15.08 (Webbink, 1985), #7290 would have an absolute visual magnitude of , consistent with the observed range for the cataclysmic variables (Warner, 1987). This star deserves a more detailed follow-up study.
© European Southern Observatory (ESO) 1998
Online publication: March 10, 1998