next up previous contents index
Next: HST Observations of Up: StarsStellar Populations, Previous: The Morphology of

Science with the Hubble Space Telescope -- II
Book Editors: P. Benvenuti, F. D. Macchetto, and E. J. Schreier
Electronic Editor: H. Payne

Globular Clusters, and the Like

Alvio Renzini
European Southern Observatory, Garching, D-85748, Germany

On leave from the University of Bologna, Department of Astronomy



Some of the most recent HST results on globular clusters are briefly reviewed. These include: 1) deep studies of the faint end of the cluster main sequence; 2) the detection of cluster white dwarfs, and their use as standard candles in the accurate determination of a cluster distance, that allows reducing the cluster age uncertainty below , with a most likely value between 14 and 15Gyr; 3) the determination of the age of metal rich clusters in the galactic bulge, that together with NTT observations demonstrates that the galactic bulge and halo formed at about the same time (i.e., 14--15Gyr ago), and sets an upper limit of to any intermediate age component; 4) the first results concerning the horizontal branch luminosity-metallicity relation for globular clusters in M31; and 5) the absence of hot horizontal branch stars in metal rich globulars and the quest for the ultraviolet emission in elliptical galaxies at low and moderate redshift.

Keywords: globular clusters,galactic bulge,elliptical galaxies


Globular clusters (GC) have been obvious HST targets since when HST was first conceived. For given exposure time, the S/N ratio of faint point-like sources scales as (the telescope diameter over its resolving power), and therefore HST is virtually equivalent to a 25m telescope operating with a 1 arcsec seeing. This applies to isolated sources, but HST can work in very crowded fields, where seeing would further dramatically degrade the performance of ground-based telescopes. So, HST is likely to remain the instrument par excellence for GC imaging, even in the 8--10m telescope era.

Given the telescope performance, obvious GC targets for HST include:

Stars in the cluster central regions ()

The bottom of the main sequence (MS)

White dwarfs (WD)

GCs in crowded fields

Giants in M31 GCs

and results pertaining to each of these categories have indeed been presented at this meeting. I have been asked by the organizers to summarize for the plenary session the highlights of the recent HST results on GCs that have been presented at the various parallel sessions. Since these proceedings will contain the corresponding papers, I will keep this review very succinct, while expanding somewhat on HST work on which I am directly involved.

Globular Cluster Cores

The centers of GCs are among the most crowded fields one can imagine, yet they contain special objects that are produced only in this extreme environment. Moreover, knowing the dynamical state of GC cores is essential to understand their dynamical evolution. The high resolution of the HST, especially of the PC, allows accurate astrometry---hence proper motions---to be obtained for stars in the core of GCs, which gives important information on the cluster potential well and its dynamical status. In turn, this can be used in conjunction with other indicators (e.g., millisecond pulsars). In this mood, Meylan et al. (1996) have imaged the central regions of 47 Tuc, using for this purpose the ultraviolet F300W filter. At this wavelength, turnoff (TO) stars are as bright as stars at the tip of the red giant branch (RGB), and very clean cluster imaging is possible, without the annoying diffraction spikes and blooming columns that plague deep HST images of GCs at longer wavelengths. While still waiting for the second epoch observations of 47 Tuc, Meylan et al. have serendipitously discovered a very fast variable star (possibly some kind of cataclysmic): an object that has brightened over two magnitudes in less than one hour. Objects of this kind are likely to be binary, and most often inhabit the very crowded central regions of GCs, thus escaping detections by ground-based observations.

The Bottom of the Main Sequence

Determining the MS luminosity function (LF) to very faint limits, hence the present-day mass function, is of great importance for our understanding of globular cluster formation and dynamical evolution, and to set stringent constraints on the IMF. De Marchi has reported about the extensive work in this direction by the group to which he belongs (Paresce, De Marchi & Romaniello 1995, De Marchi & Paresce 1995a,1995b, and these proceedings). Deep I-band LFs have been obtained for a representative sample of clusters. In spite of these clusters having had very different dynamical histories, their LFs (as sampled near the half-light radius of the clusters) look pretty similar, with a sharp peak at , thereby declining towards fainter luminosities. The peak corresponds to . At from the center of NGC 7078 the LF is instead still rising, which suggests that low mass stars have been expelled from the core by dynamical processes.

Quite similar results have been obtained by Cool, Piotto, & King (1996, reported by Piotto at the meeting) for the clusters NGC 6397, NGC 7078, and NGC 7099: the LF peak at is confirmed, as well as the difference in LF between the core and the outer parts of the clusters. The authors emphasize that the snaky shape of the MS---with at least three inflection points---should help constraining the mass-luminosity relation, that is still uncertain for low mass stars.

Figure: (a): The HST , instrumental color-magnitude diagram of the globular cluster NGC 6752. Reddening corrections have been applied. (b): The absolute, instrumental color-magnitude diagram of the local, calibrating WDs with known trig parallaxes. The WD type is also indicated.

Figure: The instrumental color-magnitude diagram of the cluster and local WDs, with the latter ones having been shifted in magnitude to match the cluster sequence. This operation delivers the distance modulus of the cluster: . The straight line is an eye fit to the cluster WD sequence.

Figure: The luminosity difference between the HB and the turnoff is plotted for a representative set of halo globular clusters (filled circles) and for NGC 6528 and NGC 6553 (filled squares). For this display [Fe/H]=0.0 and 0.1 has been adopted for the metallicity of the two clusters. The lines correspond to ages of 18, 15, and 12Gyr (upper, middle, and lower lines, respectively), with the following assumptions: 1) helium abundance ; 2) no enhancement of -elements (i.e., solar proportions); 3) HB luminosity-metallicity relations = 1.17+0.39[Fe/H] (solid lines), and = 0.73+0.15[Fe/H] (dotted line). See Ortolani et al. (1995).

A quite successful attempt at reproducing the shape of the MS with theoretical models was presented in a poster by Brocato et al. (these proceedings). Models and cluster data are indistinguishable at least down to , then the models are slightly bluer at fainter magnitudes. These models very effectively illustrate the dominant role played by the mass-luminosity relation in determining the shape of the cluster LF. For the V-band luminosity drops by about one magnitude as the mass decreases by . Instead, Between 0.2 and the luminosity drops by 3 magnitudes, and keeps dropping even more precipitously as the hydrogen burning limit is approached. No doubt, such a big drop in luminosity (starting indeed at) must play a key role in producing the rarefaction of the MS and the drop in the LF below this limit. Of course, what is most interesting is the cluster mass function, but enduring uncertainties in the M-L relation for still hamper a secure determination.

White Dwarfs, Their Use as Standard Candles, and Accurate Globular Cluster Ages

The detection of WDs in galactic GCs was listed among the top priorities of HST already in the Patras Book, back in 1982. The promise has been fully maintained indeed: as soon as the repaired HST has taken deep images of the nearby GCs the WDs have been found in the expected color-magnitude location and in the expected number (De Marchi, Paresce, & Romaniello 1995, Richer et al. 1995, Cool, Piotto, & King 1996). While this hardly came as a surprise, it is somewhat reassuring that indeed low mass stars die as theory predict they should.

To take full advantage of HST ability to detect GCWDs a special experiment was designed to use the WD cooling sequence as a standard candle, thus improving upon the determination of the distance to nearby globular clusters (Renzini et al. 1996). Fig. 1a shows the WDs detected with WFPC2 (W4 chip) at about 3 core radii from the center of NGC 6752, one of the two nearest, low-reddening GCs. Fig. 1b shows the absolute color-magnitude diagram (also obtained with WFPC2) for a sample of WDs in the solar neighborhood with well determined trig parallaxes, and with mass close to that of GCWDs (--). A comparison of the two panels shows that the two objects lying below the main WD sequence are most likely WDs of the DB variety, as also indicated by the U-band (F336W) photometry of the same objects. The two objects lying above the main WD sequence are most likely WDs in a blend with a faint red main sequence star, as suggested by the I-band (F814W) photometry. Fig. 2 shows that a vertical shift by 13.05 mag is required to bring the local WDs into coincidence with the WDs in NGC 6752, with the formal accuracy of the fit being just a few 0.01 mag. This determination improves considerably upon the accuracy with which we know the distance modulus of a globular cluster, that realistic estimates indicate to be -- mag when using other standard candles (RR Lyrae and subdwarfs).

This WD method of distance determination is similar in approach to the traditional subdwarf method (fitting the cluster MS to local, trig parallax subdwarfs), but offers several advantages over it. Indeed, WDs have metal-free atmospheres and, therefore, the cluster and local WD progenitor metallicities do not need to be specified in order to determine the distance modulus. Other methods instead need both metallicities, which basically limits the accuracy of the distance determination. Moreover, WDs are locally much more abundant than subdwarfs, which allows further improvements in the future.

An accurate distance to the clusters is essential for determining accurate GC ages. As a rule of thumb, the relative error in age is equal to the the error in distance modulus, i.e., a 0.25 mag error in the modulus generates a error in age (Renzini 1991). Other observational uncertainties (e.g., the GC chemical composition) convey substantially smaller age uncertainties. Thus, both tightening the Hubble constant (cf. Tammann's and Freedman's contributions at this meeting) and GC ages require an effort towards obtaining more accurate distances.

Armed with for NGC 6752, Renzini et al. (1996) determine the age of this cluster to be 15.3Gyr when helium diffusion is not taken into account, and 14.0Gyr when such effect is included. This assumes an -element enhancement [/Fe]=0.6. The result is in very good agreement with virtually any other determination of metal-poor GC ages, but we claim to have substantially reduced the uncertainty in this determination, from to below . The implications one can derive from coupling this age to current determination of are rather obvious, and will not be discussed further.

The Metal Rich Globular Clusters in the Bulge, and the Age of the Bulge

The case of bulge GCs offers an example of HST photometry for clusters that lie in very crowded fields, for which even the outer parts are difficult to study from the ground at TO luminosities. Ortolani et al. (1995) have obtained sufficiently deep HST photometry of two metal rich clusters of the bulge ([Fe/H]), namely NGC 6528 and NGC 6553. The color-magnitude diagrams of the two clusters are remarkably identical, from the MS all the way to the tip of the RGB. Moreover, their luminosity difference between the horizontal branch (HB) and the TO is the same, or even slightly larger, than that of the more metal poor cluster 47 Tuc ([Fe/H]=-0.7). This implies that the two bulge clusters are nearly as old as the inner halo cluster 47 Tuc, and demonstrates that the bulge underwent rapid chemical enrichment to solar abundance and beyond, very early in the evolution of our Galaxy. Moreover, Ortolani et al. (1995) show that the MS luminosity function of the two clusters is indistinguishable to that of the stars in Baade's Window (the low-reddening bulge field at from the Galaxy center), that they have obtained with the ESO NTT with superb seeing (). From this Ortolani et al. (1995) infer that the whole bulge formed quickly, some 15Gyr ago, and set an upper limit of by number to any intermediate age population in the bulge.

Our Milky Way is a spiral galaxy in a very loose group that is located rather away from major density peaks in the distribution of galaxies. Nevertheless, her whole spheroidal component looks one Hubble time old, from the outer halo globular clusters all the way to the inner bulge. Along with the well known homogeneity of elliptical galaxies, this appears to require a cosmological framework allowing a very quick build up of spheroids---no matter whether in clusters or in the field---early in the evolution of the universe. The successful detection (finally!) of forming galaxies at (cf. Giavalisco's contribution at this meeting) seems to provide the direct evidence of such events.

Extragalactic Globulars

Ortolani et al. (1995) study of bulge clusters leaves unanswered one fundamental question: is the bulge younger or older than the halo? i.e., did the galactic spheroid form outside-in or inside-out? The answer to this question depends on which calibration is adopted for the HB luminosity vs. metallicity relation, i.e., on the slope (a) of the relation [Fe/H]+b, being the absolute magnitude of HB stars. If the inside-out scenario is favored, while the outside-in option is favored if . The case is illustrated in Fig. 3.

A direct calibration of the the HB luminosity-metallicity relation can be accomplished for galactic globulars by applying to other clusters the WD method described in section 4, and precisely with this purpose in mind we have recently acquired---but not yet reduced---the corresponding HST data for the cluster 47 Tuc. A complementary approach consists in looking to a family of GCs all at the same distance, and get the relation in a very straightforward way. The ideal case is offered by GCs in M31.

Figure: The flux sampled by the F555W and F814W filters, and the upper limit to the flux trough the F218W filter for the brightest elliptical galaxy in the WFPC2 field of view in the cluster A895, as a function of the rest frame wavelength. The spectral energy distribution of NGC 4649 (one of the local ellipticals with the strongest UV rising branch) is also shown.

Rich et al. (1995) have demonstrated on the M31 globular G1 that sufficiently accurate photometry can be obtained well below the HB with WFPC2, the instrument of election for this kind of purpose. Fusi Pecci et al. (these proceedings) have combined WFPC2 and FOC M31 data for 7 GCs to derive . Seemingly, Ahjar et al. (1996) have obtained WFPC2 data for 4 M31 GCs, and derive . These determinations appear to favor a flat HB luminosity-metallicity relation, though the error bars are still large, and it may be premature to conclude that spheroids form outside-in. But ongoing HST observations of GCs in both the Galaxy and M31 should soon provide a firmer answer.

Globular clusters in the Magellanic Clouds have also been extensively observed during Cycles 4 and 5, though few results have been published so far (e.g., Gilmozzi et al. 1995). Spanning a very wide range of ages, from few to yr, and with many of them being very populous, MC globulars are of fundamental importance for stellar population studies. Indeed, MC clusters provide the best available stellar population templates at ages younger than galactic GCs, and allow the best possible determination of the IMF in an extended range of masses above . The systematic HST study of a representative sample of the most populous MC globulars should be completed in the near future.

The Horizontal Branch of the Bulge GCs, and the UV Output of Ellipticals

IUE observations have shown that the UV output of nearby elliptical galaxies increases with galaxy metallicity, as measured by the Mg index (Burstein et al. 1988). Purely energetic arguments, that are not prone to the arbitrariness afflicting the UV spectral energy distribution models, indicate hot HB stars and their progeny as the most likely UV emitters in ellipticals (Greggio & Renzini 1990). Such hot HB stars can be produced by old stellar populations provided either the mass loss rate along the RGB or the helium abundance increase enough with metallicity (or some combination thereof). Very metal-rich GCs may offer a chance to check whether indeed the HB turns towards high temperatures when high metallicities are reached. The Ortolani et al. (1995) study of NGC 6528 and NGC 6553 shows that, at least up to near solar metallicity, this is not the case: even the most metal-rich GCs in the bulge have very red HBs. Should we conclude that the UV of ellipticals is not produced by hot HB stars? This may be a premature conclusion, as the bulge clusters may not be as metal rich as the most metal-rich stars in ellipticals, and their metallicity may be below the threshold for the transition to the hot HB to take place.

An alternative check of the hot HB option is offered by stellar evolution theory. For decreasing age of a stellar population the HB is indeed predicted to move from high to low temperatures. Hence, if the UV of local elliptical is due to hot HB stars, the UV emission should rabidly decrease with lookback time, i.e., with redshift. At a redshift of 0.3--0.4 most hot HB stars should have already disappeared.

To check this prediction the cluster of galaxies A895 (z=0.37) was observed with WFPC2 during Cycle 4, spending a total of 9600 s integrating through the F218W filter, and complementing this with short F555W and F814W exposures. After troublesome cosmic ray removal and calibrations, the preliminary result is shown in Fig. 4 for the brightest (hence possibly most metal rich) elliptical in the field of view (Viezzer et al. 1996). Having obtained the fluxes at the three filters wavelengths, Viezzer et al. proceed to compare them to the spectral energy distribution of the local metal-rich elliptical galaxy NGC 4649 (L. Buson, private communication). Once the normalization is made at the rest frame --6400Å\ range (sampled by the F814W filter), the rest frame Å\ flux of the galaxy in A895 appears appreciably in excess of the flux of NGC 4649, that may partly be accounted by the passive evolution effect, part may be due to the galaxy being perhaps an E+A object. No flux at all is instead detected through the F218W filter, that samples the rest frame 1500Å region, i.e., the UV rising branch of nearby ellipticals. Unfortunately, the UV sensitivity of WFPC2 is rather low, the background fairly high, and the corresponding upper limit almost coincide with the 1550Å flux of NGC 4649. One may say that indeed there is probability that the flux is lower than in NGC 4649---thus supporting the theoretical expectation of the disappearance of the UV rising branch---but deeper UV exposures are needed to really prove the case. The far better UV sensitivity of STIS should easily allow to solve the problem.


I'm grateful to my HST PI's and Co-I's, with whom it was a real pleasure to collaborate. I'm also grateful to Lucio Buson for having kindly provided the file with the SED of NGC 4649.


Ajhar, E.A., Grillmair, C.J., Lauer, T.R., Baum, W.A., Faber, S.M., Holzman, J.A., Light, R.M., Lynds, R., & O'Neil, E.J. Jr. 1996, NOAO Preprint No. 681

Burstein, D., Bertola, F., Buson, L.M., Faber, S.M., & Lauer, T.R. 1988, ApJ, 328, 440

Cool, A.M., Piotto, G., & King, I.R. 1996, preprint

De Marchi, G. & Paresce, F. 1995a, A&A, 304, 204

De Marchi, G. & Paresce, F. 1995b, A&A, 304, 211

De Marchi, G., Paresce, F., & Romaniello, M. 1995 ApJ, 440, 216

Gilmozzi, R., Kinney, E.K., Ewald, S.P., Panagia, N., & Romaniello, M. 1995, ApJ, 435, L43

Greggio, L. & Renzini, A. 1990, ApJ, 364, 35

Meylan, G., Minniti, D., Pryor, C., Phinney, S., Sams, B., & Tinney, C. 1996, in preparation

Ortolani, S., Renzini, A., Gilmozzi, R., Marconi, G., Barbuy, B., Bica, E., & Rich, R.M. 1995, Nature, 377, 701

Renzini, A. 1991, in Observational Tests of Cosmological Inflation, ed. T. Shanks et al. (Dordrecht: Kluwer), p. 131

Renzini, A., Bragaglia, A., Ferraro, F.R., Gilmozzi, R., Ortolani, S., Holberg, J.B., Liebert, J., Wesemael, F., Bohlin, R.C., et al. 1996, ApJ, submitted

Rich, R.M., Mighell, K.J., Freedman, W.L., & Neill, J.D. 1995, AJ, 111, 768

Richer, H.B., Fahlman, G.G., Ibata, R.A., Stetson, P.B., Bell, R.A., Bolte, M., Bond, H.E., Harris, W.E., Hesser, J.E., et al. 1995, ApJ, 451, L17

Viezzer, R., Gilmozzi, R., Greggio, L., Held, E., & Renzini, A. 1996, in preparation

next up previous contents index
Next: HST Observations of Up: StarsStellar Populations, Previous: The Morphology of