Schiavon 2012 AJ 143 121

The Astronomical Journal, 143:121 (18pp), 2012 May doi:10.
1088/0004-6256/143/5/121

C 2012. The American Astronomical Society. All rights reserved. Printed in the U.S.A.
ULTRAVIOLET PROPERTIES OF GALACTIC GLOBULAR CLUSTERS

WITH GALEX. I. THE COLOR–MAGNITUDE DIAGRAMS∗
Ricardo P. Schiavon1 , Emanuele Dalessandro2 , Sangmo T. Sohn3 , Robert T. Rood4,10 , Robert W. O’Connell4 ,
Francesco R. Ferraro2 , Barbara Lanzoni2 , Giacomo Beccari5 , Soo-Chang Rey6 , Jaehyon Rhee1,7 , R. Michael Rich8 ,
Suk-Jin Yoon9 , and Young-Wook Lee9
1 Gemini Observatory, 670 N. A’Ohoku Place, Hilo, HI 96720, USA; [email protected]
2 Dipartimento di Astronomia, Università degli Studi di Bologna, via Ranzani 1, I-40127 Bologna, Italy; [email protected],
[email protected], [email protected]
3 Space Telescope Science Institute, 3700 San Matin Dr., Baltimore, MD 21218, USA; [email protected]
4 Astronomy Department, University of Virginia, P.O. Box 400325, Charlottesville, VA 22904, USA; [email protected]
5 ESO—European Southern Observatory, Karl-Swarzschild Str. 2, D-85748 Garching bei Munchen, Germany; [email protected]
6 Department of Astronomy and Space Science, Chungnam National University, Daejeon 305-764, Republic of Korea; [email protected]
7 Department of Physics, Purdue University, 525 Northwestern Avenue, West Lafayette, IN 47907, USA; [email protected]
8 Department of Physics and Astronomy, University of California, Los Angeles, USA; [email protected]
9 Center for Galaxy Evolution Research and Department of Astronomy, Yonsei University, Seoul 120-749, Republic of Korea;
[email protected], [email protected]
Received 2011 October 3; accepted 2012 January 22; published 2012 April 16
ABSTRACT
We present Galaxy Evolution Explorer (GALEX) data for 44 Galactic globular clusters (GCs) obtained during three
GALEX observing cycles between 2004 and 2008. This is the largest homogeneous data set on the UV photometric
properties of Galactic GCs ever collected. The sample selection and photometric analysis are discussed, and
color–magnitude diagrams (CMDs) are presented. The blue and intermediate-blue horizontal branch is the dominant
feature of the UV CMDs of old Galactic GCs. Our sample is large enough to display the remarkable variety of
horizontal branch shapes found in old stellar populations. Other stellar types that are obviously detected are blue
stragglers and post-core-He burning stars. The main features of UV CMDs of Galactic GCs are briefly discussed.
We establish the locus of post-core-He burning stars in the UV CMD and present a catalog of candidate asymptotic
giant branch (AGB), AGB-manqué, post early-AGB, and post-AGB stars within our cluster sample.
Key words: globular clusters: general – Hertzsprung-Russell and C-M diagrams – stars: evolution
1. INTRODUCTION for most of the “excess” UV emission observed in old stellar

populations (e.g., Greggio & Renzini 1990, 1999; Dorman et al.
It is fair to say that the last frontier of our growing under- 1995; O’Connell 1999; Brown et al. 2001), our understanding
standing of the physics of old stellar populations resides in of the physics underlying the structure and evolution of such
the ultraviolet (UV). The behavior of old stellar populations in stars is still plagued by theoretical uncertainties. Undeniably,
the UV has puzzled astronomers for almost four decades now, uncertainties are partly due to the absence of an accurate,
and in spite of major recent progress, there are still important comprehensive, statistically representative, homogeneous data
gaps in our understanding of the nature of the stars that domi- set presenting the colors and magnitudes of the stars responsible
nate the integrated light of old stellar populations in the UV— for the UV emission in Galactic GCs—in spite of painstaking
particularly the far-UV (FUV, e.g., Ferraro et al. 1998; observational efforts by a number of groups (for reviews see,
O’Connell 1999; Moehler 2001; Catelan 2009). These include e.g., O’Connell 1999; Moehler 2001). A database of that
the extreme horizontal branch (EHB) and Blue-Hook (BHk) kind would also have important applications for studies of
stars, at the hot and visually faint end of the horizontal branch extragalactic stellar populations, as it could be used to unveil
(HB), and the short-lived but more luminous supra-HB and correlations between features in the color–magnitude diagrams
post-asymptotic-giant-branch (PAGB) stars. Another popula- (CMDs) of stellar populations and their integrated properties.
tion whose nature is still not entirely well understood is that Such correlations can help understand the nature of distant
of blue straggler stars, which at the characteristic ages of Galac- systems, for which only integrated properties are available. In
tic globular clusters (GCs) are not hot enough to contribute particular, direct comparisons between integrated UV properties
substantially to the integrated FUV light, but are an important of Galactic and extragalactic GCs (e.g., Sohn et al. 2006; Rey
source of near-UV light (NUV, e.g., Ferraro et al. 2001, 2003), et al. 2007, 2009) can lend insights on the stellar population
and are in some cases detectable in integrated light longward of content of those systems (Dalessandro et al. 2012, hereafter
3400 Å (e.g., Trager et al. 2005; Schiavon 2007). Paper II).
The UV properties of old stellar populations have been a With this motivation in mind, we decided to use the Galaxy
subject of intense scrutiny ever since the discovery of the Evolution Explorer (GALEX) to undertake the largest ever
“UV-upturn” of early-type galaxies (Code 1969). While it has systematic and homogeneous census of the UV properties of
become clear in the past decade or so that EHBs are responsible Galactic GCs. Data were collected for 44 clusters in three
GALEX cycles, from which UV CMDs and integrated colors
∗ The authors dedicate this paper to the memory of co-author Bob Rood, a
were obtained. This paper discusses the sample selection and
pioneer in the theory of the evolution of low-mass stars, and a friend, who the photometric analysis of the data. Some of the data have
sadly passed away on 2011 November 2.
10 Deceased. been used in combination with Hubble Space Telescope (HST)
1
The Astronomical Journal, 143:121 (18pp), 2012 May Schiavon et al.
Figure 1. False color picture of one of the fields targeted in our survey. North
is up, east is left, and the vertical size is ∼37 . There is a population of very
blue stars covering the entire field, with a higher density toward the SE. These
are main-sequence stars belonging to the Small Magellanic Cloud. The cluster
located on the upper right of the picture is NGC 362. Blue horizontal branch
stars in NGC 362 appear as white colored objects within a few arcminutes from
the cluster center.
Figure 2. Spatial distribution of Galactic GCs in our sample. Note that

and ground-based observations for multi-band photometric NGC 2419, located at a distance of 80 kpc from the Galactic center, is outside
investigations of the stellar populations of NGC 1904 (Lanzoni the plot limits.
et al. 2007) and M 2 (Dalessandro et al. 2009), from the
innermost regions to the extreme outskirts of those clusters. lists of time-ordered photon positions and pulse heights, and
Paper II describes the derivation of integrated magnitudes for these are pipeline processed on the ground for image reconstruc-
this cluster sample and presents an analysis of correlations tion. The resulting images have pixel scales of 1.5 arcsec pixel−1
between integrated magnitudes and colors and global cluster in both FUV and NUV. Both detectors can be damaged by
properties. Paper III (R. T. Rood et al. 2012, in preparation) high global and per/pixel count rates, which prevents targeting
introduces a new classification scheme of the HBs of Galactic very (UV-) bright stars and the low Galactic latitude regions,
GCs, based on their UV morphologies. due to their high UV background. For more details, see
This paper is laid out as follows. Section 2 describes the Morrissey et al. (2005), Morrissey et al. (2007), or the GALEX
sample selection, observations, data reduction, and analysis. In instrument overview at http://galexgi.gsfc.nasa.gov/docs/galex/
Section 3 the CMDs are presented. A description of our new Documents/MissionOverview.html.
catalog of post-He-core-burning star candidates is presented in
Section 4. Our conclusions are summarized in Section 5.
2.1. Sample and Observations
2. OBSERVATIONS AND DATA ANALYSIS The data presented in this paper were primarily collected
under GALEX GI programs 056 and 099 (PI: R. Schiavon) in
GALEX is a 50 cm orbiting UV telescope launched in 2003 Cycles 1 and 4, respectively. The target selection was performed
April. GALEX has a circular field of view (FOV) of ∼1.◦ 2 di- with an eye toward spanning a wide range in metallicity and
ameter and, in imaging mode, a dichroic beam splitter allows it HB morphology. Limitations, however, were imposed by target
to collect data in two simultaneous channels, in FUV and NUV magnitudes and the safety of GALEX’s UV detectors, so that
bands, corresponding to λ = 1350–1780 and 1770–2730 Å very distant, heavily extinguished, and low Galactic latitude
(λeff = 1516 and 2267 Å) and with a spatial resolution of ∼4. 5 clusters could not be included in the sample. As a result, a
and 5. 5, respectively. GALEX detectors consist of two stacks of number of interesting clusters, particularly metal-rich ones at
three large format microplate channels and associated electron- low Galactic latitudes, were not observed, because the high
ics inserted in sealed tubes. The NUV and FUV detectors differ UV background in these regions could potentially harm the
mostly in terms of the photocathode material (CsI in the case of GALEX detectors. Clusters with very UV-bright stars within
FUV and Cs2 Te in the case of NUV) and the windows (MgF2 the GALEX FOV also could not be observed (most notably ω
for FUV, fused silica for NUV). The GALEX detectors record Cen and NGC 6752) and in some cases the pointing had to be
2
47 Tuc - FUV NUV

N
2’
Figure 3. FUV and NUV images of 47 Tuc, the reddest cluster in our sample. Note the vast difference in the crowding of the two images, which renders NUV
photometry impossible in the cluster core, at the spatial resolution of GALEX. The FUV light of the cluster is due to a few dozen sources, with roughly half of it being
due to a single very bright star (47 Tuc BS; O’Connell et al. 1997). Photometry in the FUV is accurate even in the central cluster regions.
NGC 288 - FUV N NUV

E
2’
Figure 4. FUV and NUV images of NGC 288, with a lower overall surface brightness than 47 Tuc, yet with a larger population of FUV sources—though not large
enough to present problems for FUV photometry in the central regions. In the NUV, crowding is much more important, yet photometry at the resolution of GALEX is
still achievable, though with lower precision.
NGC 5272 - FUV N NUV
2’
Figure 5. FUV and NUV images of NGC 5272. Note that stellar density is high enough that even FUV photometry is slightly more uncertain than in the cases of
NGC 288 and 47 Tuc.
adjusted in order to exclude such stars from the FOV. The target including clusters of known extragalactic origin, such as Arp 2
list for Cycle 1 totaled 25 Galactic GCs and was composed and Terzan 8 (e.g., Law & Majewski 2010). We also took
primarily of clusters for which EHB stars could be detected in advantage of the relaxation of the UV-brightness constraints
both FUV and NUV bands in a single GALEX orbit, with typical dictated by detector-safety considerations in order to obtain data
exposure times of 1500 s. for metal-rich Galactic GCs at relatively low Galactic latitude,
For Cycle 4, our strategy entailed deeper exposures on a such as NGC 6342 and NGC 6356. For the latter, as well as for
smaller sample of 15 clusters, with a focus on expanding very distant Galactic GCs (NGC 2419, Terzan 8, Arp 2, IC 4499)
coverage toward higher metallicity and younger age, while we originally had little hope of obtaining good quality CMDs,
3
NGC 7089 - FUV N NUV
2’
Figure 6. FUV and NUV images of NGC 7089, one of the densest clusters in our sample. In this extreme case, even FUV photometry is hampered in the core cluster
regions.
NGC 5053 - FUV FUV - PSF subctracted
Figure 7. Quality of PSF modeling in a good case: NGC 5053 in the FUV. Due to low density, PSF residuals are negligible and good quality photometry is achieved
for all cluster stars in the FUV.
NGC 288 - NUV NUV - PSF subtracted
Figure 8. Quality of PSF modeling in a typical case: NGC 288 in the NUV. Crowding in the central regions is relatively high, at the resolution of GALEX, photometry
is relatively inaccurate for stars within ∼0. 5 from the cluster center. The PSF-subtracted image shows a diffuse residual, associated with the detection of NUV light
from unresolved turnoff stars.
and just aimed at measuring reliable integrated colors (but see As mentioned above, our original proposals for Cycles 1
discussion in Section 3). For one cluster (NGC 6273) NUV data and 4 requested between 1 and 2 orbits to be spent on each
were not collected. cluster, which would have resulted in maximum exposure times
Finally, we further include data for six out of eight Galactic of approximately 3000 s on both bands for each target. However,
GCs from Cycle 3 GI program 075 (PI: S. T. Sohn), which our program benefited from the complexities of GALEX queue
aimed at measuring reliably UV fluxes of extreme HB stars scheduling so that longer exposure times were achieved for
in the program clusters to test the helium-rich hypothesis for some clusters—in some cases, such as that of NGC 2298,
the production of EHB stars (e.g., Lee et al. 2005). We plan exposure times were an order of magnitude longer. Exposure
to present the results of this analysis in a forthcoming paper times are also in general longer in the NUV than in the FUV,
(Paper IV; S. T. Sohn et al. 2012, in preparation). For the Cycle which is due to the several events of FUV detector shutoffs
3 program we selected clusters that exhibit extended HB blue caused by overcurrents in the FUV detector. Repeated attempts
tails in their optical CMDs. of collecting FUV data led therefore to an accumulation of NUV
4
Figure 9. Results from tests aimed at assessing the depth of our photometry. The left panel shows the CMD based on allstar, while the right panel shows the CMD
obtained by “force-finding” stars in FUV on the basis of their position in NUV by using allframe. The dashed horizontal line marks the limit corresponding to the
photometric error σFUV = 0.25.
exposures. In total, our program accrued 340 ks of open shutter clusters in our sample. The effect of crowding on our CMDs is
time, or the equivalent of ∼226 GALEX orbits. A false color discussed in Section 3.
picture of the field containing one of our clusters (NGC 362) is The photometric analysis was performed on the background-
shown in Figure 1. The spatial distribution of our cluster sample subtracted intensity images output by the GALEX pipeline
is displayed in Figure 2. (Morrissey et al. 2007, hereafter M07). These are 3840 ×
3840 pixel2 images with a plate scale of 1. 5 pixel−1 , covering
2.2. Photometry and Calibration a circular area of 1.◦ 2 in diameter, flat-field corrected and with
The quality and depth of our data are illustrated in the flux normalized by the effective area and exposure time.
Figures 3–6, where NUV and FUV images of a representative Photometry was performed by following standard procedures
subset of the sample are displayed. Consistent gray scales were for point-spread function (PSF) modeling, using the crowded-
adopted when producing these images, to allow for a fair visual field photometry package DAOPHOTII (Stetson 1987) for both
assessment of the various degrees of crowdedness of Galactic FUV and NUV images. The first step consists of defining a
GCs, as seen on both GALEX bands. To the same end, image number of bright stars across the FOV for PSF modeling. For that
sizes are set such that the FOV is equal to ∼5 times the clus- purpose, we performed a very shallow search for bright point
ter’s core radii. Displayed is one of the reddest clusters in our sources with the DAOPHOT task find on each image. Magnitudes
sample (47 Tuc) together with bluer clusters, spanning a range at this stage, before PSF modeling, were based on simple
of stellar density, increasing from NGC 288 to NGC 5272, and aperture photometry obtained using the task photometry with
NGC 7089. As a general rule, one can see that stellar density an aperture radius r = 4. 5. We then selected relatively isolated
is significantly higher in the NUV than in the FUV. This is be- bright stars spread across the FOV, for PSF determination. We
cause, on one hand, the combination of higher sensitivity and avoided stars in the very central and crowded regions.
longer exposure time makes NUV images a lot deeper than their The PSF model was typically based on 200 stars in NUV and
FUV counterparts and, on the other hand, Galactic GC stars 50 in FUV images. Quadratic spatial variations of the PSF model
are predominantly brighter in NUV than in FUV. As a result, were considered. Once the PSF model was determined, we
even at the relatively low resolution of GALEX, accurate FUV reran find and photometry with appropriate threshold levels
photometry can be obtained down to the cores of most clusters (typically 3σ –6σ off the sky background) in order to generate a
in our sample (47 Tuc and NGC 288 being two cases in point), more complete and deeper list of target stars for photometry.
getting progressively difficult at increasing cluster density, up Magnitudes were then obtained for this expanded list by
to a limit where crowding becomes a problem in the cluster performing PSF fits using the allstar routine. Tests were
central regions (e.g., NGC 7089). Unlike the FUV, crowding performed where PSF photometry was carried out replacing
in the NUV is a problem in the central regions of almost all the PSFs derived in this analysis by the average PSFs supplied
5
sources. Moreover, because crowding was far more severe in

the NUV than in the FUV, we usually adopted different cluster-
centric distance thresholds for the two bands.
Finally, we point out that the outer ∼5 annulus of GALEX
images is affected by optical distortions that may cause false
detections and large magnitude errors (see, e.g., Rey et al. 2007).
We note that our photometry is not affected by these problems,
because typically this area is well beyond the tidal radii of the
clusters in our sample, except for the cases of 47 Tuc and M 3
(NGC 104 and 5272, respectively). For these two clusters, all
photometry within the outer ∼5 annulus was discarded.
Cross-correlation of the FUV and NUV catalogs was per-
formed using CataXcorr, developed at the Observatory of
Bologna (P. Montegriffo et al. 2003, private communication),
which has the important advantage of allowing a visual check
of the quality of the geometric roto-translation solution. The
final catalogs consist of stars detected in at least one of the two
filters. This choice has been made in order to maximize the num-
ber of sources for possible cross-match with optical catalogs.
For the reasons explained above, there is a large number of NUV
sources without an FUV counterpart. On the other hand, because
crowding is more severe in the NUV than in the FUV, there is
a (small) number of central FUV sources without reliable NUV
Figure 10. Results of tests for deviations from linearity in GALEX detectors.
A comparison is shown between “predicted” and measured magnitudes for a
magnitudes.
number of standard stars from the CALSPEC database (see text). “Predicted” For the reasons discussed above, the depth achievable in
values are magnitudes obtained through synthetic photometry on CALSPEC GALEX CMDs is set by the shallower FUV photometry. In
spectra by M07, while measured values are aperture magnitudes by M07 (curves) fact, we showed in previous works (e.g., Lanzoni et al. 2007;
and PSF magnitudes from this work (data points). Gray (black) curves and Dalessandro et al. 2009) that our NUV images are often deep
open (filled) symbols represent NUV (FUV) magnitudes. Saturation becomes
important at 14.5 mag for both channels alike, but is more intense in the FUV enough to detect stars ∼1 mag fainter than the main-sequence
channel for brighter sources. Our PSF photometry seems to be only slightly turnoff. With the aim of maximizing the number of stars with
more affected than M07’s aperture photometry, in the FUV channel only. Only magnitude measurements in both GALEX bands, we attempted
a handful of stars in our entire sample are substantially affected by detector to use the allframe routine (Stetson et al. 1989) in order to
nonlinearity.
“force-find” stars in the FUV images on the basis of their
by the GALEX team.11 No significant differences between these positions in the NUV. Results of this experiment are shown
tests and the original photometry were found for a couple of in Figure 9. The left panel shows the CMD from photometry
clusters bracketing the full range of stellar densities and number based on allstar, and on the right panel the CMD obtained
of clean stars available for PSF determination. from forcing allframe to find FUV stars at their known
Aperture corrections (typically ∼0.2 mag) were calculated NUV positions is shown. These plots suggest that this use
on each image by using 15–20 isolated and bright stars, which of allframe leads to detection of sources 3–4 mag fainter
were used to generate reliable curves of growth. Instrumental in FUV than just using allstar. However, by performing a
magnitudes were converted to the ABMAG photometric system, visual analysis on the images, it became clear that most of
using the zero points provided by M07, as follows: the additional FUV detections were not real. To further verify
this result, we performed PSF photometry at random FUV
FUV = −2.5 log(counts s−1 ) + 18.82 (1) background positions ending up with a CMD that is very similar
to the one obtained when force-finding FUV stars (gray points in
NUV = −2.5 log(counts s−1 ) + 20.08. (2) the right panel of Figure 9). For this reason we decided to adopt
only the catalogs obtained by using the allstar routine as
To illustrate the quality of the PSF modeling, we show in already described. Figure 9 shows also that the two approaches
Figures 7 and 8 typical residuals from PSF subtraction in FUV give virtually identical results when stars with σFUV > 0.25 are
and NUV images, respectively. Visual inspection shows that removed from the CMD.
in low density areas, such as the FUV image of NGC 5053 Photometric depth varies from cluster to cluster according
in Figure 7 and the NUV image of NGC 288 outside the to exposure times (see Table 1), thus being in all cases deeper
cluster core in Figure 8, stellar brightness profiles are properly in the NUV than in the FUV. In our deepest images, we reach
reproduced by the PSF models used. In contrast, residuals are FUV ∼ 24.4 and NUV ∼ 25.0. NGC 2419 is the only cluster
much worse in crowded areas such as the core of NGC 288 in our sample (with both FUV and NUV images available) for
in NUV. For the reasons explained above, at the low spatial which it has not been possible to obtain reliable photometry
resolution of GALEX, crowding often caused photometry near of individual stars. Since NGC 2419 is a cluster with a large
the cluster center to be unreliable. We therefore exclude stars population of emitters both at NUV and FUV wavelengths
located within a given cluster-centric distance, for which we felt and it is one of the most distant clusters in the Galaxy
that reliable magnitudes could not be obtained on the basis of (d = 87 kpc; Dalessandro et al. 2008), it appears extremely
PSF-fitting photometry. The threshold cluster-centric distance dense in GALEX images making photometric measurements
varies from cluster to cluster depending on the density of UV of individual stars virtually impossible at the GALEX spatial
11 See http://www.galex.caltech.edu/researcher/techdoc-ch5.html
resolution.
6
Table 1
Target List
Cluster FUV texp NUV texp R.A.C Decl.C OBS Date Cycle
(s) (s) (deg) (deg)
NGC 104 2235 4069 6.085 −72.132 2006 Jul 6 GI1
NGC 288 1606 1606 13.418 −26.245 2004 Dec 6 GI1/MIS
NGC 362 2623 3027 15.809 −70.848 2005 Oct 23 GI1
NGC 1261 1225 1225 48.064 −55.217 2004 Dec 9 GI1
NGC 1851 2797 4487 78.526 −40.047 2004 Dec 10 GI1
NGC 1904 1326 3176 81.196 −24.461 2004 Dec 14 GI1
NGC 2298 10757 22171 102.066 −35.945 2004 Dec 15 GI1
NGC 2808 987 988 137.896 −64.913 2007 Mar 11 GI3
NGC 2419 1262 3695 114.688 38.869 2008 Dec 16 GI4
NGC 4147 1678 1678 182.526 18.542 2006 Mar 29 GI1
NGC 4590 1634 5081 190.020 −26.605 2007 Mar 30 GI1
NGC 5024 1656 1656 198.230 18.169 2007 May 2 GI1
NGC 5053 1781 1782 199.112 17.698 2007 May 3 GI1
NGC 5272 1679 1680 205.547 28.375 2007 May 1 GI1
NGC 5466 1841 3532 211.364 28.535 2007 May 1 GI1
NGC 5897 1590 2936 229.352 −21.010 2007 May 6 GI1
NGC 5904 1563 1566 229.592 2.069 2007 May 12 GI3
NGC 5986 4224 4225 236.514 −37.786 2007 Jun 6 GI3
NGC 6101 2010 2010 247.039 −72.502 2008 Jul 26 GI1
NGC 6218 120 23891 251.811 −1.948 2006 Jul 2 GI1
NGC 6229 1603 5419 251.769 47.477 2007 Apr 13 GI1
NGC 6235 1875 25131 253.373 −22.585 2005 Jun 24 GI1
NGC 6254 1911 25362 254.287 −4.099 2005 Jun 23 GI1
NGC 6273 2264 ... 255.603 −26.563 2007 Jun 17 GI3
NGC 6284 5767 4225 236.514 −37.786 2007 Jun 6 GI3
NGC 6341 1911 1911 258.884 43.123 2008 May 25 GI4
NGC 6342 3101 3101 260.730 −19.451 2008 May 27 GI4
NGC 6402 5185 5184 264.500 −3.350 2007 Jun 17 GI3
NGC 6356 3369 3369 260.949 −17.642 2008 May 27 GI4
NGC 6397 1584 409 265.574 −53.770 2008 Jul 17 GI4
NGC 6535 1671 1671 270.670 −0.330 2008 May 31 GI4
NGC 6584 4799 4799 274.578 −52.228 2008 Jul 17 GI4
NGC 6809 840 840 294.994 −31.063 2008 Jul 14 GI4
NGC 6864 1882 4817 301.520 −21.921 2005 Aug 6 GI1
NGC 6981 2470 5039 313.366 −12.537 2005 Aug 5 GI1
NGC 7006 1457 4690 315.375 16.185 2006 Aug 12 GI1
NGC 7089 3143 4418 323.372 −0.823 2005 Aug 5 GI1
NGC 7099 2305 2305 325.197 −23.192 2008 Aug 4 GI4
NGC 7492 1697 3302 347.224 −15.639 2005 Aug 26 GI1
Arp 2 4027 4027 292.355 −30.770 2008 Jul 11 GI4
Pal 11 2120 15771 296.428 −7.942 2005 Jun 17 GI1
Pal 12 1510 3401 326.662 −21.251 2006 Aug 1 GI1
IC 4499 4279 4279 225.077 −82.213 2008 Jul 29 GI4
Terzan 8 3084 3084 295.438 −34.000 2008 Jul 12 GI4
2.3. Deviations from Linearity Table 2

Standard Stars Used in the Nonlinearity Tests
The GALEX detectors present deviations from linearity
when count rates exceed ∼1000 counts s−1 (see M07). This Star FUV FUVM07 FUVpredicted NUV NUVM07 NUVpredicted
affects bright source photometry, particularly in the FUV (see GD50 12.74 12.70 11.98 12.82 12.84 12.57
below). In order to correct observed magnitudes, M07 compared HZ4 14.55 14.58 14.53 14.52 14.56 14.50
aperture photometry for a sample of HST spectrophotometric HZ2 13.21 13.20 12.86 13.37 13.39 13.25
standards observed by GALEX, with synthetic photometry based G191B2B 12.26 11.47 99.99 11.71 11.65 10.17
GD108 12.49 12.52 12.39 13.08 13.19 12.77
on spectrophotometric data from the CALSPEC database.12
HZ21 13.11 12.99 12.55 13.27 13.30 13.13
Because our photometry is based on PSF-fitting instead of GD153 12.78 99.99 11.33 12.37 12.36 11.91
aperture photometry, we decided to repeat the analysis done by HZ43 12.73 12.31 10.75 11.98 11.98 11.36
M07, by performing PSF photometry on the GALEX archival LTT9491 16.09 16.16 16.09 14.60 14.64 14.58
data for HST spectrophotometric standards, in order to assess the G93 12.94 99.99 12.14 12.66 12.67 12.39
impact of deviations from linearity on our magnitudes. We used NGC 7293 12.12 10.03 10.93 12.38 99.99 11.70
13 of the spectrophotometric standards from M07 (see Table 2) LDS749B 15.63 15.66 15.57 14.75 14.78 14.71
BD33 12.87 12.35 10.51 12.75 12.66 10.47
12 See http://www.stsci.edu/instruments/observatory/cdbs/calspec.html
7
(a)
Figure 11. (a) Color–magnitude diagrams of Galactic GCs. The photometry shown in Figure 11(a) through (g) was not corrected for reddening or extinction. Note
the variety of HB morphologies. The vast majority of the objects in the CMD of 47 Tuc (NGC 104) and NGC 362 with (FUVAB –NUVAB ) 1.5 are actually
main-sequence stars from the Small Magellanic Cloud.
spanning a range of 4–5 mag both in FUV and NUV. For each NUV case, deviations from linearity are consistent between this
of these stars we obtained FUV and NUV magnitudes by using work and M07, including a star that deviates very strongly from
the same procedures described in Section 2.2, and compared our linearity (BD33, in Table 3), for which our photometry is in good
results with those from M07. agreement with that of M07. On the other hand, the data suggest
The results are displayed in Figure 10, where our measure- that nonlinearity effects are slightly stronger in our PSF-fitting
ments are plotted against synthetic magnitudes as reported by photometry than in M07’s aperture photometry, particularly in
M07. Data points for both FUV (filled circles) and NUV (open the FUV.
triangles) are plotted. The solid lines are fits from M07 to the
3. THE COLOR–MAGNITUDE DIAGRAMS
relation between their aperture magnitudes and synthetic pho-
tometry, the black (gray) line represents fits to FUV (NUV) The CMDs obtained in this work are displayed in
data. The dashed line shows the one-to-one relation. It is clear Figures 11(a)–(g). The outstanding variety of colors and mag-
from this figure that nonlinearity becomes detectable in both nitudes of UV-bright sources in Galactic GCs is immediately
bands at ∼13th mag. Deviations increase with increasing bright- obvious, even on a perfunctory perusal of these diagrams. There
ness, the effect being more severe in the FUV than in the NUV. are, nonetheless, features that are common to all diagrams, and
For the brightest FUV source, nonlinearity leads to a 2.5 mag we briefly comment on those here. In Paper III, we present a
overestimate in magnitudes, the effect being ∼1 mag weaker new classification of Galactic GCs, based on the morphology
in the NUV. Most importantly, all but a handful of the stars of their HBs in UV CMDs, and study correlations between this
for which we have photometry are safely below the limit where new HB morphology index with global cluster properties.
nonlinearity effects are detectable. We start by discussing the CMD of M 3 (NGC 5272), which
It is interesting to contrast our results with those by M07, by is reproduced in better detail in Figure 12. M 3 is a moderately
comparing our data points with their fits in Figure 10. In the metal-poor Galactic GC ([Fe/H] ∼ −1.5) with a relatively blue
8
(b)
Figure 11. (Continued). The paucity of sources in the CMD of NGC 2808 is due to the shallowness of the exposures for this cluster.
HB (HB parameter = 0.08; Lee et al. 1994; Borkova & Marsakov available on the GALEX Web site. Models were adopted for
2000). GALEX magnitudes were corrected from extinction val- surface gravities typical of HB stars (Dorman et al. 1993), so
ues estimated using the Cardelli et al. (1989) extinction curve that the scale does not apply in detail to other stellar types such
and extinction values from Harris (1996). Extinction in the UV as blue stragglers and PAGB stars.
is substantially higher than in the optical, amounting in the case The first vacuum-UV CMDs for GCs were obtained by the
of NGC 5272 to AFUV ∼ 0.08 and ANUV ∼ 0.09, as opposed to Astro/Ultraviolet Imaging Telescope (e.g., Hill et al. 1992;
AV ∼ 0.03. We note however, that, because the effective wave- Parise et al. 1994; Whitney et al. 1994), and the general fea-
length of the NUV filter coincides with a bump in the Galactic tures of those diagrams are also seen in our GALEX photometry.
extinction curve (Cardelli et al. 1989), interstellar extinction Some of those same features are also seen in HST CMDs ob-
does not redden FUV–NUV, leading instead to a slight blueing tained by Ferraro et al. (1997, 2003). Three main structures are
of that color. Finally, absolute magnitudes in Figure 12 were visible in the CMD of this cluster, as indicated in Figure 12.
obtained adopting distance moduli taken from Harris (1996). The cluster HB extends from the lower right to the upper left
Only stars located at cluster-centric distances between 120 of the diagram, ranging from 4.5 to −0.5 in (FUV–NUV), and
and 1300 are displayed in Figure 12, to minimize crowding from 8 to 2 in MFUV . It is obvious from this figure that the
effects on photometry performed within the cluster core, and “horizontal” branch is not horizontal in the UV (slightly more
to minimize field contamination beyond the cluster tidal radius. so in NUV than FUV), and its slope is mainly a result of bolo-
A Teff scale is provided on the top axis of the diagram, which metric correction effects. The HB spans a wide range in Teff ,
was obtained by interpolating values into (FUV–NUV) versus going from F stars in the blue HB, at Teff ∼ 7000 K, all the
Teff versus [M/H] tables calculated on the basis of fluxes way to O stars in the so-called extreme HB at Teff ∼ 30,000 K.
from Kurucz model atmospheres,13 adopting the filter responses A few stars are also seen at the blue end of the HB, displaced
by up to 1 mag fainter in FUV than the blue tip of the HB, at
13 See http://kurucz.harvard.edu/grids.html about MFUV ∼ 3 and FUVAB –NUVAB ∼ −0.25. Those are the
9
(c)
Figure 11. (Continued)
so-called “Blue Hook” stars, whose origin is still not well un- association between this apparent gap and the Grundahl jump
derstood (e.g., Whitney et al. 1994; Moehler et al. 2004; Busso depends on currently unavailable synthetic photometry based on
et al. 2007; Rood et al. 2008; Dalessandro et al. 2011). detailed model atmosphere calculations for the relevant stellar
A few gaps are apparent along the HB of Figure 12, one parameters and abundance patterns.
of them at (FUV–NUV)/Teff ∼ 0.9/8500 K, and two other less The remaining two gaps do not seem to have observed
prominent ones located at (FUV–NUV)/Teff ∼ 3.3/7450 K and counterparts in the CMDs of Ferraro et al. (1998), which
0.0/12,000 K. The latter gap is the one that is the most likely casts doubts on the reality of those gaps. As pointed out by
to be real. It corresponds to the “G1” gap, identified by Ferraro Catelan (2008), stochastic effects due to small samples could
et al. (1998) in HST/WFPC-2 optical CMDs of M 3 and other be to blame, since some of the previously proposed gaps did
Galactic GCs. It also coincides with the position associated not stand the test of better quality CMDs, based on more
with the Grundahl jump—a discontinuity in the HBs of GCs, robust samples. According to Catelan (2008), real features such
first pointed out by Grundahl et al. (1998), which manifests as the Grundahl jump are probably associated with chemical
itself as a brightening of the Strömgren u or the Johnson composition discontinuities along the HB, which can manifest
U-band magnitudes of stars hotter than Teff ∼ 11,500 K. The themselves through opacity effects due to specific chemical
Grundahl jump has been interpreted by Grundahl et al. (1999) species, which may operate on some photometric bands, but
as being due to a decrease of hydrogen opacity relative to not on others. The latter could conceivably explain the presence
metal opacity, associated with an increase of light element of these two gaps in our CMD, but not in those of Ferraro
opacities due to radiative levitation for Teff 11,500 K. et al. (1998), provided an opacity source can be identified that is
Inspection of Figure 8 of Grundahl et al. (1999) suggests that important in the NUV/FUV but not in the optical. Alternatively,
the differential impact of radiative levitation on FUV- and NUV- these gaps may be due to the fact that the nonlinearity of the
like photometric bands can potentially generate a gap with a (FUV–NUV)–Teff relation leads to a color stretching of the
similar size to that observed in Figure 12. However, a definitive redder part of the UV HB, which may make such gaps more
10
(d)
readily detectable in the UV than in the optical. This issue powerful mean to study blue stragglers, and in particular their
clearly deserves further investigation in future studies. spatial distribution in GCs.
Another feature of UV HB morphologies is the clump of stars Another important population in this CMD is that of post-He-
at 2.5 MFUV 3.5 and 0.3 FUV–NUV 0.8. This feature core burning stars, whose identification is difficult, given their
is actually an artifact caused by the highly nonlinear character of rarity and the uncertainties surrounding their evolutionary paths
the color–Teff relation. At (FUV–NUV) ∼ 4.0 (Teff ∼ 7000 K), in the CMD, as well as their lifetimes. There are two PAGB can-
a 0.5 mag color interval spans a few 100 K in Teff , whereas at didates in this CMD, which are approximately 1.5 mag brighter
(FUV–NUV) ∼ 0.3 (Teff ∼ 10,000 K) the same color interval than the brightest HB stars, at Teff greater than ∼20,000 K. See
spans several 1000 K, leading to the accumulation of data points discussion in Section 4.
in that area of the HB for any cluster with a substantial number The cloud of points that is located toward fainter magnitudes
of stars hotter than Teff ∼ 8500 K. and bluer colors than the HB is mostly populated by background
The next important population visible in the CMD of sources, with an average color of (FUV–NUV) ∼ 0.5 and
Figure 12 is that of blue stragglers. Their identification in this MFUV 5 (FUV 19 in Figures 11(a)–(g)). Some of those
case is easy, as they are spread along a sequence that is parallel, objects may actually belong to the cluster populations with
and 1–1.5 mag fainter than the HB (e.g., Ferraro et al. 1999; bright blue stragglers contributing on the red side and young
see also Figure 2 by Ferraro et al. 1997). A Girardi et al. (2000) white dwarfs demarcating the blue envelope. Based on WFPC2
zero-age main sequence for the metallicity of NGC 5272 is plot- data, Ferraro et al. (2001) argued for the presence of young
ted as a dashed line, in order to facilitate the identification of white dwarfs, with ages 13 million years, in the corresponding
the cluster’s blue stragglers. Only the hottest and brightest blue locus of the (mF 218W –mF 439W ) CMD of 47 Tuc. In particular,
stragglers are detected in the FUV. Lanzoni et al. (2007) and they showed that the blue envelope of that CMD population
Dalessandro et al. (2009) have recently shown that the combi- is consistent with theoretical expectations both for the colors
nation of GALEX data with wide-field optical photometry is a and number counts of young white dwarfs. However, while
11
(e)
that study refers to a small region at the center of the cluster, triangles, whereas stars at larger cluster-centric distances, within
where the background field contamination is expected to be the cluster tidal radius, are plotted with open circles. The
low, the GALEX FOV is expected to be heavily contaminated different CMD loci occupied by stars within and outside the
by background objects. In fact, inspection of high-resolution 2 radius shows that crowding produces a population of stars
images taken with the wide-field imager, attached to the ESO artificially brighter and redder than the cluster’s HB population.
2.2 m telescope (Lanzoni et al. 2007), indicated that the majority While the brighter magnitudes are a straightforward effect of
of the sources in that region of the CMD consists of distant blending, the apparent redder colors are due to the fact that
galaxies. In addition, the number of objects in this region of the blending is more severe in the NUV than in the FUV. The case
UV CMD of M 3 is consistent with the number of extragalactic of NGC 7089 is somewhat extreme, since the HB of this cluster
known objects as found in the NASA Extragalactic Database.14 is so populous that crowding is important in both FUV and
In summary, the low resolution of GALEX images and the NUV images (Figure 6). In most cases, crowding in the FUV is
relatively low resolution of the ESO 2.2 m images do not allow far less severe, and its effect on CMDs is that of producing
one to distinguish unequivocally between white dwarfs and blue a predominantly redder population, due to crowding in the
stragglers on one side, and background galaxies on the other. NUV. We also point out that because the “brightening” effect
Therefore, we decide to leave them in the plots, with the caveat associated with stellar blending should be typically of the order
that absolute magnitudes should be disregarded for most objects of 0.75 mag, it is possible that some of the very bright stars
in that region of this diagram. at FUV 14.5 in NGC 7089 may be real UV-bright cluster
The effect of crowding on the GALEX CMDs can be assessed members. See discussion in Section 4.
in Figure 13, where stars in the field of NGC 7089 are plotted.
Stars within 2 from the cluster center are shown as gray 4. UV-BRIGHT STARS
While the integrated light of old stellar populations in the
14 http://nedwww.ipac.caltech.edu/ FUV is dominated by EHB stars, post-He-core burning stars
12
(f)
also contribute a fraction of that radiation (e.g., Greggio & phase (van Winckel 2003). GCs are the one type of stellar system
Renzini 1990, 1999; O’Connell 1999). A few definitions are where the initial masses of these stars are best constrained, so
required at this point. According to standard stellar evolution that observations of post-core-He burning stars in clusters can
theory, post-HB evolution depends strongly on the mass of in principle contribute to the betterment of stellar evolution
the stellar envelope. After core-He exhaustion, stars with the models. However, stellar evolution proceeds at a very fast pace
highest envelope masses evolve into the AGB phase, undergoing after the core-He burning stage, with timescales varying between
thermal pulses and eventually losing their envelopes, evolving 104 and 106 yr. The incidence of these stars in stellar systems
toward higher temperatures at constant high luminosity as PAGB of relatively low mass, such as GCs, is therefore low, and thus
stars. Stars with lower envelope mass experience a much shier strongly affected by stochastic effects. The wide FOV of GALEX
excursion into the AGB phase and never undergo thermal pulses, and the size of our sample configure an ideal situation for the
evolving toward higher temperatures, after envelope loss, at cataloging of these rare stellar types. We describe in this section
constant, but lower, luminosities. The latter are called post the procedure we followed in order to identify PAGB and other
early-AGB (PEAGB) stars. Finally, at the extreme low end of UV-bright star candidates.
envelope mass, stars never make it to AGB phase after core- The paucity of post-core-He burning stars makes their iden-
He exhaustion, departing the blue end of the HB in a small tification solely on the basis of photometry in any given single
excursion toward higher luminosities, but never becoming as GC extremely uncertain, though an early attempt was made
bright as PEAGB stars. The latter are the so-called AGB-manqué using a UV CMD of NGC 6752 by Landsman et al. (1996).
(AGBM) stars. Because the average number of PAGB stars per cluster is of
Our knowledge of the total contribution of these stars to the order of ∼1, they form no sequence in any of the CMDs
the integrated light of old stellar populations is limited by shown in the previous section. In the absence of a sequence, dis-
uncertainties in evolutionary tracks, which are to a large extent tinguishing post-core-He burning stars from fore/background
due to difficulties in the modeling of mass loss during the AGB field contaminants in the CMD of any individual cluster is very
13
(g)
hard, and usually requires a spectroscopic follow-up. However, HB are also apparent in Figure 14. The bluest and brightest
stacking the CMDs of many clusters should boost the number in FUV have colors roughly between (FUV–NUV) = −0.5
of UV-bright stars per unit CMD area, highlighting the locus and +0.5, and extend to magnitudes as bright as MFUV ∼ −3.
occupied by stars in these evolutionary stages. Figure 14 shows The other family of stars is located toward redder colors and
a stack of the best 23 CMDs from Figures 11(a)–(g), which do fainter magnitudes, consisting of a population of stars on average
not have a very strong background contamination. The clusters 2–3 mag brighter than the HB, with (FUV–NUV) 1.5. These
included are NGC 1851, 1904, 2298, 4147, 4590, 5024, 5053, stars are mostly foreground contaminants, as discussed below.
5272, 5466, 5897, 5904, 6101, 6218, 6229, 6254, 6341, 6535, Finally, we note that there is a population of stars that are brighter
6584, 6809, 6981, 7089, 7099, and 7492. Data for each cluster than the HB by no more than 1 mag, spread through its entire
were placed on an absolute magnitude scale, adopting redden- extension. These are most likely unresolved stellar blends.
ing and distance modulus from the latest version of the Harris We first turn our attention to the main objects of interest, the
(1996) catalog, and only stars within the radial limits displayed population of stars revealed by the CMD stack just above the
in Figures 11(a)–(g) are shown in Figure 14. The spread in mag- extreme HB stars at Teff of a few tens of 104 K. This stellar se-
nitude of the stacked HB is likely caused by uncertainties in the quence is too blue and extends toward too bright magnitudes to
adopted distance moduli and in the adopted reddening values. harbor a significant fraction of blends. We note that in Figure 13
Because all the clusters are brought to the same distance, all almost all the stars considered to be due to blends produced
the typical features of the UV CMDs of old stellar populations by crowding effects are redder than (FUV–NUV) ∼ 0. More-
appear in sharp contrast in this CMD stack. For instance, the over, because Figure 14 excludes stars within central cluster
blue straggler sequence stretching below the red part of the HB, regions, crowding effects should be minimal anyway. So, we
and the supra-HB stars in the other extreme of the HB are more conclude that this sequence of hot UV-bright stars constitutes
clearly seen in the CMD stack than in most individual CMDs a real population of UV-bright stars hosted by our sample of
of Figures 11(a)–(g). Two stellar sequences brighter than the Galactic GCs. In fact, these stars indeed occupy the same locus
14
Table 3 Table 3
Post-He-core Burning Candidates (Continued)
Star ID FUV NUV R.A. Decl. Class Star ID FUV NUV R.A. Decl. Class
NGC 104-1 13.069 12.448 00:23:58.0 −72:05:30 P(E)AGB NGC 6254-117 13.990 15.071 16:57:05.2 −04:07:56 AGBM
NGC 288-5695 16.687 16.227 12:52:40.9 +26:33:53 AGBM NGC 6254-152 13.208 14.829 16:56:43.7 −04:05:41 AGBM
NGC 362-1625 14.038 14.885 1:03:11.5 −70:49:13 AGBM NGC 6254-189 14.390 15.851 16:57:06.3 −04:03:19 AGBM
NGC 362-372 13.649 14.496 1:02:16.2 −70:51:42 AGBM NGC 6254-241 13.433 14.564 16:57:14.7 −04:05:03 AGBM
NGC 362-1626 13.772 14.632 1:03:38.8 −70:49:12 AGBM NGC 6254-242 11.096 12.905 16:57:09.4 −04:04:24 P(E)AGB
NGC 362-1444 13.769 14.791 1:01:53.1 −70:54:11 AGBM NGC 6254-364 14.167 15.680 16:57:01.1 −04:04:30 AGBM
NGC 1851-44 10.895 12.355 05:14:08.6 −40:03:03 P(E)AGB NGC 6342-110 18.991 18.803 17:21:38.3 −19:34:04 P(E)AGB
NGC 2808-103 16.178 16.193 09:12:05.4 −64:52:37 P(E)AGB NGC 6397-149 14.773 15.029 17:41:30.620 −53:28:18.90 AGBM
NGC 2808-128 16.714 16.410 09:12:00.5 −64:51:26 P(E)AGB NGC 6402-31 17.581 17.508 17:37:33.2 −03:14:52 PAGB
NGC 2808-670 18.024 18.328 9:11:24.9 −64:52:45 AGBM NGC 6864-8 15.853 15.412 20:06:05.5 −21:54:59 PAGB
NGC 2808-756 18.288 18.482 9:11:50.9 −64:50:30 AGBM NGC 6864-77 18.125 17.832 20:05:19.2 −22:04:21 AGBM
NGC 2808-1146 18.255 18.944 9:12:17.2 −64:48:04 AGBM NGC 6864-102 18.669 18.126 20:07:01.8 −21:46:18 AGBM
NGC 4590-9 15.925 15.627 12:38:33.0 −26:41:15 AGBM NGC 6864-212 19.620 19.086 20:04:55.2 −21:51:57 AGBM
NGC 5024-5 15.269 15.655 13:14:00.1 +18:31:31 P(E)AGB NGC 6864-224 19.632 19.138 20:06:40.8 −22:00:23 AGBM
NGC 5024-7 15.506 15.832 13:13:10.6 +18:07:36 P(E)AGB NGC 6864-225 19.631 19.140 20:05:21.5 −21:53:06 AGBM
NGC 5024-161 17.994 18.307 13:12:39.5 +18:04:54 AGBM NGC 6864-452 19.286 20.017 20:06:10.5 −21:38:58 AGBM
NGC 5272-2 13.613 13.778 13:42:16.9 +28:26:02 P(E)AGB NGC 7006-16 16.834 17.385 21:01:35.1 +16:06:10 AGBM
NGC 5272-32 16.137 16.488 13:42:01.2 +28:23:25 AGBM NGC 7089-407 12.492 13.221 21:33:31.4 −00:49:09 P(E)AGB
NGC 5272-38 16.146 16.527 13:42:05.9 +28:19:05 AGBM NGC 7089-387 12.839 17.126 21:33:35.6 −00:51:22 P(E)AGB
NGC 5466-1 13.075 13.997 14:03:17.2 +28:39:30 PAGB NGC 7089-234 14.175 15.013 21:33:19.7 −00:47:5 AGBM
NGC 5466-36 17.391 17.793 14:05:31.6 +28:33:10 AGBM NGC 7089-194 14.788 15.834 21:33:17.9 −00:49:58 AGBM
NGC 5897-26 16.564 16.719 15:18:56.6 −21:08:40 AGBM NGC 7089-765 14.326 14.938 21:32:29.4 −00:48:31 AGBM
NGC 5897-34 17.578 17.080 15:17:23.7 −20:37:56 AGBM NGC 7089-89 14.571 15.579 21:33:30.9 −00:47:20 AGBM
NGC 5897-61 17.499 17.835 15:18:40.6 −21:16:08 AGBM NGC 7099-10 14.763 15.058 21:40:18.1 −23:13:23 AGBM
NGC 5904-1 13.318 13.361 15:18:34.2 +02:05:02 P(E)AGB NGC 7 099-68 16.545 16.651 21:41:38.7 −22:54:11 AGBM
NGC 5904-3 14.413 14.612 15:18:32.8 +01:54:54 AGBM Arp 2-25 18.227 17.757 19:27:45.3 −30:24:51 P(E)AGB
NGC 5986-634 19.140 19.443 15:46:38.1 −37:41:25 AGBM Arp 2-61 19.555 19.860 19:29:11.2 −30:17:09 AGBM
NGC 5986-701 19.165 19.580 15:46:46.5 −37:49:38. AGBM Arp 2-70 19.831 20.254 19:28:57.0 −30:27:59 AGBM
NGC 6101-4874 16.782 17.159 16:26:47.1 −72:15:24 AGBM Pal 12-25 16.423 16.498 21:47:49.2 −21:17:33 AGBM
NGC 6218-48 16.643 17.423 16:47:12.6 −01:41:24 AGBM Pal 12-59 18.094 17.556 21:46:04.3 −21:21:46 AGBM
NGC 6235-154 18.585 19.174 16:53:39.1 −22:16:17 AGBM IC 4499-485 19.811 20.409 14:56:54.2 −82:12:22 AGBM
NGC 6235-184 19.449 19.412 16:52:47.4 −22:04:14 AGBM Ter 8-38 17.800 17.270 19:41:40.8 −34:03:59 P(E)AGB
NGC 6235-186 19.578 19.437 16:52:51.1 −22:14:21 AGBM
NGC 6235-196 18.847 19.507 16:52:49.4 −22:11:50 AGBM
NGC 6235-254 19.754 19.780 16:53:18.0 −22:11:39 AGBM
as the PAGB, PEAGB, and AGBM stars identified by Brown
NGC 6235-355 19.496 20.136 16:53:43.4 −22:17:36 AGBM
NGC 6235-439 19.734 20.350 16:53:38.3 −22:01:02 AGBM
et al. (2008) in a STIS UV CMD of M 32 stars (their Figure 3).
NGC 6235-43 16.919 17.475 16:53:20.8 −22:02:40 P(E)AGB In order to gain further insight into their nature, we reproduce
NGC 6254-66 14.244 15.544 16:56:48.0 −04:04:33 AGBM the CMD stack in Figure 15, overlaying evolutionary tracks by
NGC 6254-112 13.813 15.052 16:57:02.8 −04:08:19 AGBM Brown et al. (2008) for a PEAGB and a PAGB star of ∼0.5
15
Figure 12. Reddening- and distance modulus-corrected CMD of NGC 5272

(M 3), indicating the main populations that dominate the UV light of old stellar
populations. The Teff scale on the top axis was obtained using Kurucz model Figure 14. Stacked color–magnitude diagram of 23 Galactic GCs (see
fluxes (http://kurucz.harvard.edu/grids.html) for the cluster metallicity, adopting Section 4). The UV-bright population composed by candidate post-He-core
surface gravities from Dorman et al. (1993) and is appropriate only for HB stars. burning at (MFUV –MNUV ) ∼ 0 and MFUV 2 is clearly seen in this stack.
Note the presence of gaps in the cluster horizontal branch, and an artificial clump Redder stars brighter than the HB are likely to be predominantly background
of HB stars around (FUV–NUV) ∼ 0.5, which is due to the strongly nonlinear sources. The blue stragglers can also be very clearly spotted in this diagram,
color–Teff relation. The brightest and hottest cluster blue stragglers are clearly below the redder half of the HB.
detected, 1–1.5 mag below the HB. Two candidate PAGB stars are visible at
MFUV ∼ 1 and (FUV–NUV) ∼ −0.5. A few white dwarf candidates are also
detected, but, at the GALEX resolution, it is very difficult to distinguish them
from background sources. The latter are predominantly extragalactic.
Figure 15. Stacked color–magnitude diagram from Figure 14 (gray dots), with
theoretical models from Brown et al. (2008) overlaid on the data. The thick
solid line represents the post-HB evolutionary path for a star of M = 0.77 M ,
Figure 13. Observed CMD of one of our densest clusters, NGC 7089, illustrating while the dash-dotted line is for an M = 0.515 M star. The dashed model
the effect of crowding in our photometry. Gray triangles represent sources within is the ZAHB. The solid squares represent candidate PAGB stars. Large circles
2 of the cluster center, and all others are located between that inner radial are candidate P(E)AGBs and crosses are candidate AGBM stars. Large gray
distance and the cluster tidal radius (1300 ). The main effect of crowding is triangles indicate the positions of a few well-known PAGB stars. Note that,
to displace stars toward brighter FUV magnitudes and redder colors. The color while for all the other stellar types photometry is only plotted for stars within
effect is due to the fact that crowding is more severe in the NUV than in the the cluster-centric limits shown in Figures 11(a)–(g), photometry is shown for
FUV. all PAGB stars. See discussion in Section 4.
(dash-dotted line) and 0.8 M (thick solid line), respectively. (ZAHB; dashed line) is also shown, which matches very well
The model prediction for the zero-age horizontal branch the lower envelope of our observed HB.
16
On the basis of the discussion above, we can use the UV properties of these systems. In this paper, we describe the
evolutionary tracks in Figure 15 to assign the UV-bright stars sample selection, observations, and data reduction, presenting a
in our sample to the above evolutionary classes. Candidates brief description of the main features of the UV CMDs. HB stars
for the different classes are listed in Table 3 and shown in are the most important feature of the UV CMDs, and our CMDs
Figure 15, where the data from Figure 14 are shown as gray reveal an outstanding variety in the shape of the HB in our cluster
dots. Filled circles indicate the positions of all PAGB candidates sample. Blue straggler stars are also detected in many clusters.
in our entire cluster sample, regardless of their cluster-centric We present a catalog of PAGB, PEAGB, and AGBM candidates,
distances. The large gray triangles indicate the positions of a which should be useful for studies of these rare, but UV-bright,
few PAGB stars known to exist in clusters from our sample. stellar types. We hope these data will provide better constraints
We chose not to impose a cluster-centric distance cut in our on models of stellar evolution during, and after, the HB phase.
selection of PAGB and PEAGB candidates, because they are In Paper II, we present the integrated UV photometry for this
bright enough that crowding effects on their photometry are sample, while a new classification scheme of the morphology
minimal. That is not the case of AGBM stars, though, which of the HBs of Galactic GCs in UV is presented in Paper III.
lie close enough in magnitude to the HB that their locus in GALEX provided us with an opportunity, unique in this
the CMD may be substantially contaminated by unresolved decade, to collect precious data that will be crucial to help
pairs of HB stars. Therefore, the list of AGBM candidates untangling the intricacies of the latest stages of evolution of
presented in Table 3 only includes stars within the cluster- low-mass stars, so as to allow a deeper understanding of the
centric distance thresholds displayed in Figures 11(a)–(g). We UV properties of old stellar populations. We hope that this
consider stars brighter than the ZAHB by more than 1 mag data set will enable notable progress in this field during the
in MFUV and fainter than the PEAGB tracks to be AGBM upcoming years. The photometric catalogs can be downloaded
candidates. Stars brighter than the PEAGB class are either from http://www.cosmic-lab.eu.
PEAGB or PAGB candidates, we therefore refer to these stars
as P(E)AGB. In view of the uncertainties in evolutionary tracks Based on observations made with the NASA Galaxy Evolu-
and the possible contamination of our magnitudes by stellar tion Explorer. GALEX is operated for NASA by the California
blends (for stars within the crowded areas of the clusters), we Institute of Technology under NASA contract NAS5-98034. We
refrain from attempting a distinction between the latter two thank Allen Sweigart and Tom Brown for making available the
classes in our sample. Finally, stars brighter than the PAGB theoretical evolutionary tracks employed in this paper. R.P.S.
track are considered to be PAGB candidates. Note that two acknowledges funding by GALEX grants NNG05GE50G and
of the stars identified as PAGB in previous literature (large NNX08AW42G and support from Gemini Observatory, which
gray triangles) would be classified as AGBM and P(E)AGB is operated by the Association of Universities for Research in
according to our classification scheme. We also impose a color Astronomy, Inc., on behalf of the international Gemini partner-
cut in our definition of PAGB, PEAGB, and AGBM candidates, ship of Argentina, Australia, Brazil, Canada, Chile, the United
by requiring that they have (FUV–NUV) < 0.7. Note that the Kingdom, and the United States of America. E.D. thanks the
AGBM and P(E)AGB candidates identified in Figure 15 include hospitality and support from Gemini Observatory where most
stars from all clusters in our sample, not only the 23 clusters of this work was developed, during two extended visits in 2009
included in the stacked CMD from Figure 14. and 2010, and a shorter visit in 2011. This research is also
Finally, we focus on the redder population of stars brighter part of the project COSMIC-LAB funded by the European Re-
than the HB. According to Brown et al. (2008) tracks, PAGB search Council (under contract ERC-2010-AdG-267675). The
stars spend only 25% of their time with colors redder than financial contribution of the Italian Istituto Nazionale di As-
FUVAB –NUVAB ∼ 0.7, so the fact that there are more bright trofisica (INAF, under contract PRIN-INAF 2008) and the Agen-
stars in Figure 14 on the red side of that color threshold zia Spaziale Italiana (under contract ASI/INAF/I/009/10)
than in the blue side is strongly suggestive of the presence of is also acknowledged. S.T.S. and J.R. are partially supported
back/foreground contamination. There are approximately 26 by GALEX grant GI3-0075/NNX07AP07G. S.-C.R. acknowl-
stars in Figure 14 with MFUV > 2.2 and FUVAB –NUVAB < 0.7. edges support from Basic Science Research Program (No. 2009-
Conversely, there are approximately 67 stars with brighter than 0070263) and the Center for Galaxy Evolution Research through
the HB by ∼1 mag and with 0.7 < FUVAB –NUVAB < 5. the National Research Foundation of Korea (NRF). S.-J.Y.
If the evolutionary tracks are correct, we would expect to find acknowledges support from Mid-career Researcher Program
no more than ∼9 stars in that region of the CMD. Therefore, (No. 2009-0080851) and Basic Science Research Program (No.
we suggest that the vast majority of the bright stars redder than 2009-0086824) through the NRF of Korea.
FUVAB –NUVAB ∼ 0.7 are not cluster members, likely be-
ing foreground A and F stars. That is not to say, of course, REFERENCES
that there are no cluster PAGB stars in that region of the di-
agram—in fact, they are very likely to be there, but finding Borkova, T. V., & Marsakov, V. A. 2000, Astron. Rep., 44, 665
them on the basis of GALEX data alone would be like finding Brown, T. M., Smith, E., Ferguson, H. C., et al. 2008, ApJ, 682, 319
Brown, T. M., Sweigart, A. V., Lanz, T., Landsman, W. B., & Hubeny, I.
needles in a haystack. Therefore, we impose a color cut in our 2001, ApJ, 562, 638
definition of PAGB, PEAGB, and AGBM candidates, by requir- Busso, G., Cassisi, S., Piotto, G., et al. 2007, A&A, 474, 105
ing that they have FUVAB –NUVAB < 0.7. This color cut is Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
aimed at minimizing contamination of the candidate sample by Catelan, M. 2008, Mem. Soc. Astron. Ital., 79, 388
Catelan, M. 2009, Ap&SS, 320, 261
back/foreground contaminants sources. Code, A. D. 1969, PASP, 81, 475
Dalessandro, E., Beccari, G., Lanzoni, B., et al. 2009, ApJS, 182, 509
5. CONCLUSIONS Dalessandro, E., Lanzoni, B., Ferraro, F. R., et al. 2008, ApJ, 681, 311
Dalessandro, E., Salaris, M., Ferraro, F. R., et al. 2011, MNRAS, 410, 694
We have used GALEX to image 44 Galactic GCs in the FUV Dalessandro, E., et al. 2012, AJ, submitted
and NUV, thus creating the largest homogeneous database of the Dorman, B., O’Connell, R. W., & Rood, R. T. 1995, ApJ, 442, 105
17
Dorman, B., Rood, R. T., & O’Connell, R. W. 1993, ApJ, 419, 596 Lee, Y.-W., Joo, S.-J., Han, S.-I., et al. 2005, ApJ, 621, L57
Ferraro, F. R., D’Amico, N., Possenti, A., Mignani, R. P., & Paltrinieri, B. Moehler, S. 2001, PASP, 113, 162
2001, ApJ, 561, 337 Moehler, S., Sweigart, A. V., Landsman, W. B., Hammer, N. J., & Dreizler, S.
Ferraro, F. R., Paltrinieri, B., Fusi Pecci, F., Rood, R. T., & Dorman, B. 2004, A&A, 415, 313
1998, ApJ, 500, 311 Morrissey, P., Conrow, T., Barlow, T. A., et al. 2007, ApJS, 173, 682
Ferraro, F. R., Paltrinieri, B., Fusi Pecci, F., et al. 1997, A&A, 324, 915 Morrissey, P., Schiminovich, D., Barlow, T. A., et al. 2005, ApJ, 619, L7
Ferraro, F. R., Paltrinieri, B., Rood, R. T., & Dorman, B. 1999, ApJ, 522, 983 O’Connell, R. W. 1999, ARA&A, 37, 603
Ferraro, F. R., Sills, A., Rood, R. T., Paltrinieri, B., & Buonanno, R. 2003, ApJ, O’Connell, R. W., Dorman, B., Shah, R. Y., et al. 1997, AJ, 114, 1982
588, 464 Parise, R. A., Maran, S. P., Landsman, W. B., et al. 1994, ApJ, 423, 305
Girardi, L., Bressan, A., Bertelli, G., & Chiosi, C. 2000, A&AS, 141, 371 Rey, S.-C., Rich, R. M., Sohn, S. T., et al. 2007, ApJS, 173, 643
Greggio, L., & Renzini, A. 1990, ApJ, 364, 35 Rey, S.-C., Sohn, S. T., Beasley, M. A., et al. 2009, ApJ, 700, L11
Greggio, L., & Renzini, A. 1999, Mem. Soc. Astron. Ital., 70, 691 Rood, R. T., Beccari, G., Lanzoni, B., et al. 2008, Mem. Soc. Astron. Ital., 79,
Grundahl, F., Catelan, M., Landsman, W. B., Stetson, P. B., & Andersen, M. I. 383
1999, ApJ, 524, 242 Schiavon, R. P. 2007, ApJS, 171, 146
Grundahl, F., Vandenberg, D. A., & Andersen, M. I. 1998, ApJ, 500, L179 Sohn, S. T., O’Connell, R. W., Kundu, A., et al. 2006, AJ, 131, 866
Harris, W. E. 1996, AJ, 112, 1487 Stetson, P. B. 1987, PASP, 99, 191
Hill, R. S., Hill, J. K., Landsman, W. B., et al. 1992, ApJ, 395, L17 Stetson, P. B., Hesser, J. E., Smith, G. H., Vandenberg, D. A., & Bolte, M.
Landsman, W. B., Sweigart, A. V., Bohlin, R. C., et al. 1996, ApJ, 472, L93 1989, AJ, 97, 1360
Lanzoni, B., Sanna, N., Ferraro, F. R., et al. 2007, ApJ, 663, 1040 Trager, S. C., Worthey, G., Faber, S. M., & Dressler, A. 2005, MNRAS, 362, 2
Law, D. R., & Majewski, S. R. 2010, ApJ, 718, 1128 van Winckel, H. 2003, ARA&A, 41, 391
Lee, Y.-W., Demarque, P., & Zinn, R. 1994, ApJ, 423, 248 Whitney, J. H., O’Connell, R. W., Rood, R. T., et al. 1994, AJ, 108, 1350
18

Schiavon 2012 AJ 143 121

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Schiavon 2012 AJ 143 121

Uploaded by

Copyright:

Available Formats

The Astronomical Journal, 143:121 (18pp), 2012 May doi:10.

ULTRAVIOLET PROPERTIES OF GALACTIC GLOBULAR CLUSTERS

1. INTRODUCTION for most of the “excess” UV emission observed in old stellar

Figure 2. Spatial distribution of Galactic GCs in our sample. Note that

47 Tuc - FUV NUV

NGC 288 - FUV N NUV

NGC 5272 - FUV N NUV

NGC 7089 - FUV N NUV

NGC 5053 - FUV FUV - PSF subctracted

NGC 288 - NUV NUV - PSF subtracted

sources. Moreover, because crowding was far more severe in

2.3. Deviations from Linearity Table 2

Figure 12. Reddening- and distance modulus-corrected CMD of NGC 5272

You might also like