1: %\documentclass[manuscript]{aastex}
2: \documentclass[12pt,preprint]{aastex}
3: \usepackage{natbib,graphicx,latexsym}
4: \newcommand{\lsim}{\
5: \raise-2.truept\hbox{\rlap{\hbox{$\sim$}}\raise5.truept\hbox{$<$}\ }}
6: \newcommand{\gsim}{\
7: \raise-2.truept\hbox{\rlap{\hbox{$\sim$}}\raise5.truept\hbox{$>$}\ }}
8: \newcommand{\etal}{et~al.~}
9: \begin{document}
10:
11:
12: \title{The globular cluster system in NGC\,5866: optical observations
13: from HST Advanced Camera for Surveys\altaffilmark{1}}
14:
15: \author{Cantiello, Michele\altaffilmark{2,3}}
16: \author{Blakeslee, John P.\altaffilmark{2}}
17: \author{Raimondo, Gabriella\altaffilmark{3}}
18: %\author{Brocato Enzo\altaffilmark{3}}
19: %\author{Capaccioli Massimo\altaffilmark{4}}
20:
21: \altaffiltext{1}{Based on observations made with the NASA/ESA Hubble
22: Space Telescope, obtained from the Data Archive at the Space
23: Telescope Science Institute, which is operated by the Association of
24: Universities for Research in Astronomy, Inc., under NASA contract NAS
25: 5-26555.}
26: \altaffiltext{2}{Department of Physics and Astronomy, Washington State University,
27: Pullman, WA 99164; email: cantiello, jblakes@wsu.edu}
28: \altaffiltext{3}{INAF--Osservatorio Astronomico di
29: Teramo, Via M. Maggini, I-64100 Teramo, Italy; email: cantiello, raimondo@oa-teramo.inaf.it}
30:
31:
32: \begin{abstract}
33: We perform a detailed study of the Globular Cluster (GC) system in the
34: galaxy NGC\,5866 based on F435W, F555W, and F625W ($\sim$ B, V, and R)
35: HST Advanced Camera for Surveys images.
36:
37: Adopting color, size and shape selection criteria, the final list of
38: GC candidates comprises 109 objects, with small estimated
39: contamination from background galaxies, and foreground stars.
40:
41: The color distribution of the final GC sample has a bimodal
42: form. Adopting color to metallicity transformations derived from the
43: Teramo--SPoT simple stellar population model, we estimate a
44: metallicity [Fe/H]$\sim -1.5$, and $-$0.6 dex for the blue and red
45: peaks, respectively. A similar result is found if the empirical
46: color-metallicity relations derived from Galactic GCs data are used.
47:
48: The two subpopulations show some of the features commonly observed in
49: the GC system of other galaxies, like a ``blue tilt'',
50: higher central concentrations of the red subsystem, and larger
51: half--light radii at larger galactocentric distances. However, we do
52: not find evidence of a substantial difference between the average sizes of
53: red and blue clusters.
54:
55: Our analysis of the GC Luminosity Function indicates a V-band
56: Turn-Over Magnitude V$_0^{TOM}$=23.46$\pm$0.06, or
57: M$_{V,0}^{TOM}\sim-7.29 \pm 0.10$ mag, using the distance modulus
58: derived from the average of SBF and the PNLF distances. The absolute
59: Turn-Over Magnitude obtained agrees well with calibrations from
60: literature. The specific frequency is measured to be $S_N=1.4 \pm
61: 0.3$, typical for galaxies of this type.
62:
63:
64: \end{abstract}
65:
66: \keywords{globular clusters: general --- galaxies: individual,
67: NGC\,5866 --- galaxies: star clusters --- galaxies: distances}
68:
69:
70: \section{Introduction}
71:
72: The study of the Globular Cluster (GC) system in galaxies is one of
73: the fundamental keystones to understand the physical processes at the
74: base of the formation and evolution of galaxies. The investigation of
75: GC system properties involves aspects of several different fields in
76: astronomy: stellar evolution, hierarchical assembly, the determination
77: of extra--galactic distances, etc.
78:
79: The first applications of the GCs as an investigation tool came in the
80: earliest years of extra--galactic astronomy. Single bright GC were
81: recognized as possible distance indicator in the 1950s \citep{baum55},
82: but it took few decades to develop the technique as it is applied
83: now. The systematic study of the GC Luminosity Function (GCLF), in
84: fact, has shown that it has a universal nearly gaussian shape. It is
85: now recognized that the Turn Over Magnitude (TOM) of the GCLF is a
86: reliable distance indicator for most galaxies, although some
87: exceptions exist \citep{f00,richtler03,jordan07}.
88:
89: Another, more recent discovery in the field of GC astronomy, is that
90: the presence of a color bimodality is quite common in galaxies. The
91: color bimodality has been mostly interpreted as a bimodal metallicity
92: distribution \citep{harris01,brodie06}, leading to develop several
93: possible GC formation scenarios, useful to constrain the history of
94: galaxy formation. More recently, \citet{kundu07} using near-IR
95: photometry, have shown that the GC bimodality in the giant ellipticals
96: M\,87 and NGC\,4472 might represent true metallicity
97: subpopulations. However, whether or not the bimodal color distribution
98: is linked to the presence of two metallicity subpopulations is still a
99: matter of debate, and there are claims that the optical properties of
100: a bimodal GC system are not necessarily due to a bimodal metallicity
101: distribution. For example, it has been shown that a bimodal color
102: distributions might be generated by the combination of a unimodal
103: metallicity distribution and a non-linear color-metallicity relation
104: of the appropriate shape \citep{richtler06,yoon06}.
105:
106: Furthermore, the advent of space-based astronomy has provided new
107: information in this field. Most of all, the high resolution optical
108: imaging of HST makes it possible to examine the shape properties,
109: i.e. size and ellipticity, for an increasing number of extra--galactic
110: GCs, up to distances as far as the Virgo Cluster \citep{jordan07} and
111: beyond. The examination of the characteristic sizes, and radial
112: distribution of the blue and red GC subpopulations, enables the study
113: of new phenomena related to the formation history of the GC system
114: \citep{larsen03,jordan04}.
115:
116: In this paper we take advantage of archival HST images of the galaxy
117: NGC\,5866, taken with the Advanced Camera for Surveys (ACS) in the
118: Wide Filed Camera mode, to carry out the first detailed
119: multi-wavelength study of the GC system in this galaxy.
120:
121: NGC\,5866 is a lenticular galaxy, with a sharp dust lane nearly edge
122: on. Being a disk galaxy, the GC system is expected to be less
123: populated with respect to ellipticals
124: \citep[e.g.][]{harris81,brodie06}. However, the quality of these ACS
125: data allows us to analyze in some detail the clusters properties.
126:
127: The weighted average of distances moduli obtained using the SBF
128: \citep[][using the new zeropoint from Jensen et al, 2003]{tonry01},
129: and PNLF \citep{ciardullo02} methods, provide a distance modulus
130: $\mu_0=30.75 \pm 0.08$, or 14.1$\pm$0.5 Mpc. At this distance, the GCs
131: of the galaxy are near the ACS resolution limit, allowing us to carry out
132: the size and shape analysis of the GC system.
133:
134: The analysis of GC properties will be presented in the next sections.
135: The observations and data reduction procedure are introduced in
136: Section 2. Section 3 deals with the GC photometry and selection
137: criteria. The color bimodality, color-magnitude relations, and the
138: spatial distribution of the total GC system, as well as the two GC
139: subpopulations, are discussed in Section 4. In the same Section we
140: also make a comparison with stellar synthesis models. Finally, a study
141: of GCLF and specific frequency is also presented in Section
142: 4. We summarize and conclude the paper in Section 5.
143:
144:
145:
146: \section{Data reduction}
147:
148: The ACS/WFC camera observations of NGC\,5866 were taken in February
149: 2006 with the F435W, F555W and F625W filters ($\sim$ B, V, and R). The
150: total exposure times for the three different passbands were: 3900s,
151: 2800s and 2200s, respectively.
152:
153: Standard reduction was carried out on raw images using the APSIS image
154: processing software \citep{blake03}. After cosmic--ray rejection,
155: alignment, ``drizzling'', and final image combination, we adopted an
156: iterative procedure to determine the sky background and the galaxy
157: best fit models, independently for the three passbands.
158:
159: Since the galaxy fills the whole field of view of the ACS, the sky
160: background determined in the outer corners of the CCD image is not
161: accurate. We obtain the sky values extrapolating from the fit of a de
162: Vaucouleurs $r^{1/4}$ profile.
163:
164: Briefly, after masking out a few bright/saturated foreground stars, we
165: adopt a provisional sky value from the corner with the lowest median
166: number of counts. This value is subtracted to the image before
167: starting the galaxy fitting procedure. Then, we fit the galaxy
168: isophotes using the IRAF/STSDAS task ELLIPSE\footnote{IRAF is
169: distributed by the National Optical Astronomy Observatories, which are
170: operated by the Association of Universities for Research in Astronomy,
171: Inc., under cooperative agreement with the National Science
172: Foundation.}, which is based on the method described by
173: \citet{jedrzejewski87}. Once the preliminary galaxy model is
174: subtracted from the sky--subtracted frame, a wealth of faint sources
175: appears together with the dust disk. A mask of these sources is
176: obtained using a sigma detection method, and combined with the
177: previous mask. The new mask is then fed to ELLIPSE to refit the
178: galaxy's isophotes.
179:
180: After the geometric profile of the isophotes has been determined, we
181: measure the galaxy flux in elliptical isophotes. At last, to improve
182: the estimation of the sky, we fit the surface brightness profiles with
183: a de Vaucouleurs $r^{1/4}$ profile plus a constant sky offset. The
184: sky offset fitted is then adopted as the new sky value, and the whole
185: procedure of galaxy fitting, source masking, surface brightness
186: profile fitting and sky estimation is repeated, until convergence.
187: This procedure is adopted for all three filters.
188:
189: Once the sky and galaxy model have been subtracted from the frame,
190: some large--scale deviations are still present in the frame. We remove
191: these deviations using the background map obtained running SExtractor
192: \citep{bertin96} on the sky and galaxy subtracted frame. Part of the
193: procedure adopted here is the same developed by \citet{cantiello05} to
194: process ACS images for SBF analysis. We refer the reader to that paper
195: for more details on the whole procedure of sky+galaxy+large scale
196: residuals subtraction.
197:
198: It is worth noting that in \citet{cantiello05} we have made some
199: checks on the stability of the sky estimation against the fitting
200: profile chosen. There, it was shown that, adopting a more general
201: Sersic $r^{1/n}$ profile, the sky values obtained agree within
202: uncertainties with those obtained adopting a de Vaucouleurs
203: profile. In addition, in \citet{cantiello07}, we compared the sky
204: values obtained applying our method with the sky obtained by
205: \citet{sikkema06} using a different approach, over a sample of 6
206: peculiar galaxies. As a result we found that the differences between
207: the two sky background determinations is on average $\lsim$3\%.
208:
209: Figure \ref{image} shows the I-band ACS image of
210: NGC\,5866\footnote{Also visit the web site
211: http://heritage.stsci.edu/2006/24/index.html for a nice color picture
212: of the galaxy.} (left panel), and a zoom of the sky+galaxy+large scale
213: residuals subtracted frame (right panel, ``residual'' frame
214: hereafter), the nearly edge--on dust disk in the galaxy is easily
215: recognizable in this second panel.
216:
217: \section{Globular Cluster selection and photometry}
218:
219: The residual images were used for the detection and photometry of
220: point-like and extended sources present in the frames. The source
221: photometry was obtained independently on all three frames using
222: SExtractor. After source detection, the output catalogs were matched
223: using a 0.1$''$ radius.
224:
225: To derive the aperture correction, standard growth curve analysis
226: \citep{stetson90} was performed on selected, isolated compact sources.
227: Extinction corrections \citep[values from][]{sfd98} and aperture
228: corrections were applied before transforming the ACS magnitudes into
229: the standard B, V and R magnitudes, using the prescriptions by
230: \citet{sirianni05}. The transformations from ACS to standard BVRI
231: system magnitudes were obtained adopting the synthetic transformations
232: from \citeauthor{sirianni05} We have checked the compatibility of
233: these magnitudes with the ones derived adopting the empirical
234: transformations from \citeauthor{sirianni05} As a result, we have
235: found that the differences between the magnitudes obtained with the
236: two set of transformations agree on average within 0.02 mag, at least
237: in the range of colors of GC candidates which will be adopted later
238: on.
239:
240: We have chosen to select GC candidates using an approach similar to
241: \citet[][S06 hereafter]{spitler06}, based on the color, shape and size
242: of the objects. At the distance of NGC\,5866, the typical GC
243: half-light radius, $r_{hl}\sim$3 pc, is at the limit of the angular
244: resolution of the ACS camera (0.05 arcsec/pixel). Under such
245: condition, the software {\it ishape} \citep{larsen99} can be used to
246: estimate the intrinsic shape parameters of slightly resolved objects.
247: To run the $ishape$ software an oversampled PSF of the original image
248: is needed. Based on their colors, we have selected a sample of star
249: candidates from each frame and derived the PSF using the photometry
250: package DAOPHOT, in the IRAF environment. The objects selected for PSF
251: analysis were a posteriori confirmed to be stars.
252:
253: In order to save computational time, we have removed from the initial
254: catalog of matched sources all the sources easily recognizable as
255: background galaxies. The new catalog was then fed to $ishape$, with
256: the PSFs, to derive the shape parameters of the listed objects in the
257: three different passbands. One of the main parameters to run $ishape$
258: is the convolution kernel, which, once convolved with the PSF, is used
259: to fit the ellipticity, orientation, FWHM, etc. of the object. After
260: various tests, we found that the best results were obtained adopting
261: the analytic King profile, with concentration parameter c=30.
262:
263: The final shape parameters for each object were then adopted as the
264: average between the three estimations from the B, V and R
265: frames. Those objects with half--light radius in one band more than
266: 3--$ \sigma$ different from the V-band $r_{hl}$ value, were rejected
267: from the catalog. Similarly to S06, we have repeated such rejection
268: procedure twice on the whole catalog of objects.
269:
270: Next, we have rejected those object with $ishape$ ellipticities $\geq$
271: 0.5 ($<$ 10 objects at all). As a final selection criteria we have
272: used the (B-R)$_0$ and (B-V)$_0$ color of the remaining sources. Only
273: those objects whose color fell within the intervals 0.9$\leq (B-R)_0
274: \leq$ 1.7, and $0.5 \leq (B-V)_0 \leq 1.1$ were taken as globular
275: cluster candidates. After these selections, the final catalog of GC
276: candidates consisted of 109 objects. The list of GC candidates, and
277: their photometric, spatial and shape properties are reported in Table
278: \ref{tab_gc}.
279:
280: Before using these data, it is useful to estimate the amount of
281: contamination from background galaxies and foreground stars.
282:
283: Concerning the contamination due to background galaxies, given the
284: similarity of NGC\,5866 data with those of the Sombrero galaxy by S06,
285: we adopt the conclusions from these authors. S06 have reduced and
286: analyzed B and V images from Hubble Deep Field--North of the GOODS
287: survey, and found that the number of background objects matching the
288: GC selection criteria is $\sim$ 4\% the final Sombrero GC sample. When
289: this number is scaled to the imaging area of NGC\,5866 ($\sim$1/6 of
290: the Sombrero), and to the number of objects classified as GC, we find
291: that in our case the contamination due to background galaxies is
292: $\sim$ 6\% the final number of GCs.
293:
294: The number of galactic foreground stars expected in the color ranges
295: adopted for GC selection have been calculated using the
296: \citet{bahcall80} model. A number of approximately $\sim$ 8 stars is
297: predicted to fall in the range of colors adopted for the GC
298: selection. We find a similar, but somewhat lower number ($\sim$ 5-6
299: stars) using the models from \citet{robin03}. These numbers have to be
300: compared with the number of objects actually matching with the adopted
301: color ranges, and rejected because of the small half-light radius,
302: that is 4 objects at all. Given the small number statistics, we
303: consider the number of contaminating stars from models consistent with
304: the observed one, thus we can safely assume that the final sample of
305: GC in Table \ref{tab_gc} is likely free of contamination from
306: foreground stars\footnote{If the expected and the observed number of
307: stars are taken as they are, i.e. without taking into account the
308: uncertainties, we are lead to conclude that $\sim$ 3\% of the final
309: GC sample is composed by contaminating stars.}.
310:
311: \section{Discussion}
312: \subsection{Color histograms}
313:
314: Figure \ref{colhisto} shows the (B-R)$_0$, (B-V)$_0$ and (V-R)$_0$
315: color histograms for the 109 GC candidates. To test the bimodality of
316: the color distribution of the GC sample we use the KMM code
317: \citep{mclachlan88,ashman94}. This code evaluates the likelihood
318: whether or not the color distribution of sample is better represented
319: by two gaussians or one single gaussian, the positions of the peaks
320: are evaluated as well.
321:
322: Running the KMM code on the (B-R)$_0$ color, returns a likelihood of
323: $\sim$96\% that the GC color distribution is bimodal, with
324: peaks at (B-R)$_0$=1.12 (blue peak), and (B-R)$_0$=1.35 (red
325: peak). The GCs are evenly assigned to each group.
326:
327: An independent KMM run on the (B-V)$_0$ color, returned a slightly
328: lower but significant bimodality probability (93 \%), and peaks at
329: (B-V)$_0$=0.67, (B-V)$_0$=0.84 mag.
330:
331: The narrow range of GCs (V-R)$_0$ color, together with the small
332: separation between the two possible peaks ($ < 0.1$ mag for the $\sim$
333: 600 GCs of the Sombrero galaxy), and the relatively small number of GC
334: candidates, hampers any meaningful solution by KMM for this color.
335:
336: At present, it is still a matter of debate what is the origin of the
337: bimodality. The recent review from \citet{brodie06} listed some of the
338: different scenario proposed (major mergers, dissipationless accretion,
339: hierarchical merging, etc.). In general the photometric data
340: available, and the few spectroscopic measurements seem to point out
341: that the primary difference between the two GC subpopulations is due
342: to metallicity, without the need of invoking age variations. On the
343: other hand, recently, it has been shown that a nonlinear
344: color-metallicity relation could explain a bimodal color distribution
345: without requiring a bimodal metallicity distribution
346: \citep{richtler06,yoon06}.
347:
348: If we assume the picture where the two color peaks are generated by a
349: bimodal metallicity distribution, the metallicities corresponding to
350: the peaks can be derived by assuming some [Fe/H] versus color
351: relations. Adopting the view that GCs are mostly old, and using the
352: \citet[][R05 hereafter\footnote{The R05 models are available at the
353: Teramo-SPoT group website: www.oa-teramo.inaf.it/SPoT}]{raimondo05}
354: models in the metallicity range $-1.8 \leq$ [Fe/H] $\leq$ 0.3, for
355: ages in the interval 9 $\leq$ t (Gyr) $\leq$ 14, we derive
356: $[Fe/H]=(3.67 \pm 0.13)(B-R)_0 - (5.60 \pm 0.18)$. We have verified
357: the reliability of this equation by comparing the theoretical
358: [Fe/H]$_{th.}$ values predicted for the Galactic GCs, with the
359: observational [Fe/H]$_{obs.}$ estimations. For this comparison we have
360: adopted the (B-R)$_0$ and [Fe/H]$_{obs.}$ values from the updated
361: online \citet{harris96} catalog of Galactic GCs\footnote{The catalog
362: can be found at the web address:
363: http://www.physics.mcmaster.ca/$\sim$harris/mwgc.dat}. As a result we
364: find that the median difference is $\langle
365: [Fe/H]_{th.}-[Fe/H]_{obs.}\rangle= -0.02 \pm 0.16$ dex, confirming the
366: reliability of the adopted color-metallicity relation. The good
367: agreement of the models with Galactic GC data can also be seen in
368: Figure \ref{ggc_colfeh} (left panel) where we show the fit of the
369: [Fe/H] versus (B-R)$_0$ from the R05 models, and the Galactic GCs
370: data.
371:
372: Using the above [Fe/H] versus (B-R)$_0$ equation, the NGC\,5866 peak
373: metallicities are [Fe/H]=$-1.48\pm0.23$, and $-$0.64$\pm$0.25 dex for
374: the blue and red GC peaks, respectively.
375:
376: We have repeated the above steps using (B-V)$_0$ colors instead of the
377: (B-R)$_0$. The color-metallicity relation in this case is
378: $[Fe/H]=(5.48 \pm 0.19)(B-V)_0 - (5.13 \pm 0.16)$ (Fig. \ref{ggc_colfeh},
379: right panel), which implies [Fe/H]=$-1.49\pm0.20$,
380: and $-0.54\pm0.23$ dex for the two subpopulations\footnote{Using the
381: empirical color-metallicity relations derived by S06 from Galactic GC
382: data, the GC color peaks correspond to [Fe/H]$\sim-1.5$, and
383: [Fe/H]$\sim-$0.75 dex for the blue and red subpopulations
384: respectively.}.
385:
386: These metallicity values can be considered quite average in bimodal GC
387: systems. In fact, adopting the estimations for other galaxies reported
388: by \citet{brodie06}, we find on average [Fe/H]=$-1.47\pm0.12$ and
389: $-0.48\pm0.16$ dex for the sample of normal elliptical and S0 galaxies.
390:
391: \subsection{Blue tilt}
392:
393: Inspecting the color magnitude diagram of GCs, it has been shown that
394: in some giant galaxies a color-luminosity correlation exists for the
395: blue GCs \citep{harris06}. The interpretation of this correlation
396: must be taken carefully, due to the limits of the GC selection
397: criteria. Moreover, this phenomenon has not been detected in all
398: galaxies \citep{mieske06a,brodie06}
399:
400: To explain the existence of the color-luminosity correlation two
401: different GC formation scenarios have been proposed: the
402: self-enrichment, and the pre-enrichment one. The differences between
403: these two scenarios is that in the first one the proto-GCs are
404: able to trap and absorb the metals formed by stars within the GC;
405: in the pre-enrichment scenario it is assumed that higher mass proto-GCs
406: come from intrinsically higher metallicity clouds - see
407: \citet{strader06} and references therein for more details.
408:
409: Figure \ref{cmd} shows the color magnitude diagrams for the NGC\,5866
410: GCs. For the (B-R)$_0$ and (B-V)$_0$ we independently used the
411: KMM output classification to separate the GC system into
412: subpopulations. For the (V-R)$_0$ we used the combined results from
413: the above colors to separate between the two subpopulations. It is
414: worth noting that the sharp separation limits adopted to discriminate
415: between the two sub-populations introduce a bias in the results, since
416: real GC subsystems must overlap in some color range. Keeping in mind
417: this warning, in Figure \ref{cmd} we show the linear least--squares
418: fits to the single subsystems.
419:
420: We find a slight color-luminosity correlation for the blue GCs: the
421: slope of the (B-R)$_0$ color versus R magnitude relation is
422: $-0.012\pm0.009$. The blue tilt is more evident if (B-V)$_0$ and
423: (V-R)$_0$ data are taken into account\footnote{For this Figure we have
424: determined the blue and red peaks of the (V-R)$_0$ color using:
425: (V-R)$_{0,blue}$=(B-R)$_{0,blue}$-(B-V)$_{0,blue}$=0.45 mag, and
426: (V-R)$_{0,red}$=(B-R)$_{0,red}$-(B-V)$_{0,red}$=0.51. These numbers
427: agree with the observed (V-R)$_0$ peaks in the Sombrero (S06).}, since
428: with slopes $-0.017\pm0.009$, and $-0.013 \pm0.003$, respectively. The
429: slopes of the color-luminosity relations of the red sub-population are
430: in all cases consistent with zero.
431:
432: Using the color metallicity relations reported in the previous section
433: to convert the tilt in terms of metallicity differences, we find that
434: the brightest blue GC have on average $\sim$ 0.4 dex higher [Fe/H]
435: respect to the faintest ones.
436:
437: It is interesting to note that the study of a sample of bright
438: ellipticals lead \citet{harris06} to conclude that color-luminosity
439: relation appears more clearly in galaxies with larger number of
440: clusters. An analogous result has been found by \citet{mieske06a} by
441: analyzing the GC properties for a sample of 79 Virgo Cluster galaxies.
442: A similar conclusion can be drawn by comparing our data with the
443: results of the Sombrero galaxy by S06. In fact, NGC\,5866 is
444: morphologically similar to the brighter Sombrero galaxy, but its GC
445: population within $\sim$1.5 galactic effective radii is $\sim$ 1/5 the
446: GC population in $\sim$1.5 effective radii of the
447: Sombrero\footnote{NGC\,5866 images have nearly the same absolute
448: magnitude completeness limit of the Sombrero images.}, and the
449: (B-R)$_0$ versus R slope for blue GCs is almost one third the one
450: derived from the Sombrero galaxy.
451:
452: \subsection{Spatial properties}
453: One of the common results concerning the projected radial density of
454: the GC subpopulations in galaxies, is that they can be fitted by a
455: power law, with the red subpopulation being more centrally
456: concentrated than the blue one. More specifically, it has been
457: shown that the outer parts of GC systems are typically well-fit by
458: power laws, while the central few kpc, hidden in these NGC\,5866 data,
459: often show the presence of cores \citep[see, e.g.,][]{ashman98}.
460:
461: Figure \ref{radist} shows the GC surface density in NGC\,5866 versus
462: the logarithm of galactocentric distance, $\log R_{gc}$. The upper
463: panel shows the total density profile, while in the lower panel the
464: profiles of the two GC subpopulations are shown separately. From this
465: figure it is readily recognizable that $i$) the red subpopulation
466: appears slightly more centrally concentrated respect to the blue one,
467: $ii$) the density profiles are well fitted by power laws, with slopes:
468: $-1.7\pm0.3$, $-0.9\pm0.3$, and $-1.3\pm0.2$ for the red, blue,
469: and total populations respectively. By comparing the slope of the
470: total density profile with data from literature we find that NGC\,5866
471: has an average behavior. In fact, using the compilation by
472: \citet{kp97}, we find that slopes in the range [$-$2.5,$-$1] are quite
473: common in galaxies of total magnitude similar to NGC\,5866, i.e.,
474: $M_{Bt}\sim -20$ mag.
475:
476:
477: The radial profiles of the (B-R)$_0$ color for the two subpopulations
478: do not show significant radial gradients (upper panels in Figure
479: \ref{colrad}). However, due to the different concentrations of the red
480: and blue GCs shown in Figure \ref{radist}, the total cluster
481: population shows a clear color gradient (Figure \ref{colrad}, lower
482: left panel). In the lower right panel of Figure \ref{colrad} we show
483: the (B-R)$_0$ color profile of the galaxy, the best fit line to the
484: total GC color profile is shown, too. As can be recognized from this
485: figure, the color profile of field stars closely resembles the color
486: profile of the total GC system, with a $\sim+0.07$ mag shift,
487: i.e. $\sim$ 0.3 dex higher metallicity.
488:
489: These observational properties can be interpreted with a scenario
490: where the metal poor blue GCs form in the early stages of the galaxy
491: life, while the red GC component and the field stars are formed later
492: on in a metal enriched potential well. Or, it could just be a
493: metallicity gradient for both the GCs and field stars, combined with
494: the nonlinear color-metallicity relation that originates the observed
495: GC bimodal color distributions \citep{yoon06,richtler06}.
496:
497: One other issue in understanding GC system formation and evolution, is
498: the relation existing between the average sizes of GC subpopulations,
499: and the variation of half-light radii with galactocentric distances.
500:
501: Some observational data have shown that the half-light radii of the
502: red subpopulation are on average 10\% to 20\% smaller than blue
503: GC. Two different explanations have been given to this phenomenon.
504: The one proposed by \citet{larsen03} involves a projection effect, due
505: to the different spatial distribution of red and blue GCs, together
506: with a strong correlation of the GC size with the galaxy radius. This
507: model predicts that the size differences should be larger in the inner
508: galactic regions, and disappear in the outskirts of the galaxy. In the
509: other scenario, proposed by \citet{jordan04}, mass segregation and
510: stellar evolution effects are the leading causes of the blue/red GC
511: size differences. In this scenario little change of relative GC sizes
512: with galactocentric distances is expected.
513:
514: By comparing the average half-light radii of the red and blue GC in
515: NGC\,5866 we do not find any significant difference between the two
516: subpopulations. Furthermore, the size versus galactocentric
517: radius comparison, shown in Figure \ref{reffrad}, reveals a slight
518: tendency of outer GC to have larger half-light radii.
519:
520: Due to the small number of GCs available, the absence of a clear size
521: difference between the red and blue GCs, and the uncertainties in the
522: half-light radius versus $R_{gc}$ correlation, these observations
523: cannot be used to clearly support either the mass segregation or the
524: projection scenarios.
525:
526:
527: \subsection{Comparison with models}
528: Figure \ref{spot} shows a comparison between the observed colors of
529: NGC\,5866 GCs and the predictions from R05. Each panel in the figure
530: shows a Simple Stellar Population (SSP) model with a defined [Fe/H],
531: and for ages 1 $\leq$ t (Gyr) $\leq$ 14.
532:
533: At first glance, these panels will give little information about the
534: age and metallicity properties of the observed GCs, due to the strong
535: overlap of models with different physical properties. This is not
536: related to the particular choice of SSP models. In order to reduce
537: model systematics, we have also considered a data to models comparison
538: using the \citet{bc03} SSP models. The new set of models essentially
539: agrees with the R05 one, and the overlap between models at different
540: [Fe/H] is still present. Moreover, since the R05 models are optimized to
541: match in detail the observational features of Color Magnitude Diagrams
542: of Galactic and Magellanic Clouds stellar clusters, in this section we
543: will take the set of stellar synthesis models from R05 as reference
544: ones.
545:
546: If no constraint is put on models, the conclusions that can be drawn
547: from the comparison shown in Figure \ref{spot} are mainly: $i$) SSP
548: models, i.e. single age and single metallicity stellar systems,
549: reproduce nicely the observational color-color properties of GCs; $ii$)
550: except one single object, no old super metal-rich (t $>$ 7 Gyr,
551: [Fe/H]$>$[Fe/H]$_{\sun}$) GC seems to be present; $iii$) the reddest
552: GCs in the sample [(B-R)$_0\sim$1.6, (B-V)$_0\sim$1] are either old,
553: t$\sim$ 14 Gyr, solar metallicity objects, or younger, t$\sim$ 5 Gyr,
554: super-solar metallicity stellar systems.
555:
556: However, if we limit the models to the range of old age models (9
557: $\leq$ t (Gyr) $\leq$ 14) some stronger constraints on the age and
558: metallicity of the GC can be obtained. To support this choice, as
559: mentioned before, we recall that old GC ages have been found in most
560: of the GC systems in galaxies both with photometric and spectroscopic
561: data \citep{brodie06}. In addition to this, as shown in Figure
562: \ref{3gxy} (upper panels), the range of (B-R)$_0$ and (B-V)$_0$
563: colors for NGC\,5866 and for Galactic GCs appears to be very similar.
564: If we assume that such color similarity is also associated with common
565: physical characteristics of the GC systems, than we can adopt for
566: NGC\,5866 an age range similar to the one of Galactic GC, that is 9
567: $\lsim$ t (Gyr) $\lsim$ 14 \citep[e.g.][]{deangeli05}.
568:
569: Based on these considerations, we can go back to the comparisons shown
570: in Figure \ref{spot}, taking into account only the models in the
571: selected age interval. With this constraint we find that the blue tail
572: of GCs has [Fe/H]$\leq-1.8$ dex. If we extrapolate the [Fe/H] versus
573: (B-R)$_0$ equation derived from R05 models using the same age
574: limitations adopted here, then the most metal poor GC in NGC\,5866 has
575: [Fe/H]$\sim-2.3$ dex. On the other hand, inspecting the color properties
576: of red clusters, the conclusion is that the reddest GCs in NGC\,5866
577: are old objects, t$\sim$14 Gyr, with nearly solar metallicity.
578:
579: Before concluding this section, we find interesting to note that if
580: the same arguments are applied to the GC system in the Sombrero galaxy
581: (data from S06, shown in the right panels of Figure \ref{3gxy}), the
582: metallicity range spanned by the GC system is $-2.2\lsim$ [Fe/H]
583: $\lsim$ 0.6. Thus, while the blue tails of the GC system appear to be
584: quite similar in the three galaxies, the red tail of the GCs in
585: the Sombrero includes also a population of very metal rich clusters.
586:
587: \subsection{Globular Cluster Luminosity Function}
588:
589: The review by \citet{richtler03} gives a fairly detailed study on the
590: use of GCs as distance indicators. By comparing GCLF distances with
591: those derived using the SBF method, \citeauthor{richtler03} proves
592: that the TOM is a good distance indicator, although some exceptions
593: exists, related to a possible contamination from intermediate age GCs,
594: or young massive clusters.
595:
596: Figure \ref{cfv} shows the V-band completeness functions for NGC\,5866
597: derived within three annular regions with radii $R_{gc} (arcsec) \leq
598: 35$, $35 < R_{gc} (arcsec) \leq 70$, and $R_{gc} (arcsec)> 70$. A 95\%
599: completeness level is reached at V$_0=23.9,~ 24.8,~25.1$ mag for the
600: three different annuli, respectively. The completeness functions have
601: been derived via standard artificial star tests, using the DAOPHOT
602: task $addstar$ to generate the artificial stars, then we ran
603: SExtractor for the detection of sources, as described in Section 2.
604:
605: Adopting SBF distances, \citet{richtler03} obtained an absolute TOM
606: M$_{V,0}=-7.35 \pm 0.25$ mag\footnote{We have corrected the TOM value
607: reported by \citet{richtler03} using the SBF zeropoint correction as
608: discussed by \citet{jensen03}.}, which at the distance of NGC\,5866
609: translates into an apparent TOM V$_0^{TOM}\sim$ 23.46 mag. This means
610: that the GCLF is fairly complete up to $\sim$0.5, 1.4, and 1.7
611: magnitudes fainter than the TOM in the three different annular regions
612: taken into account.
613:
614: Fitting a gaussian distribution to the GCLF corrected for
615: incompleteness, gives a TOM V$_0^{TOM}=23.46 \pm 0.06$ mag, with a
616: dispersion $\sigma=1.13 \pm 0.05$. Adopting the SBF+PNLF average
617: distance $\mu_0 = 30.75 \pm 0.08$, the absolute TOM is M$_{V,0}^{TOM}=
618: -7.29 \pm 0.10$ mag, in nice agreement with the value provided by
619: \citeauthor{richtler03}.
620:
621: \subsection{Specific Frequency}
622:
623: Although the field of view of the ACS camera does not sample the
624: whole NGC\,5866 area, we have obtained an estimation for the Specific
625: frequency, $S_N$, of GCs over the ACS sampling area.
626:
627: The concept of specific frequency, $S_N=N_{GC} \times 10 ^{0.4
628: (M_V+15)}$, was introduced by \citet{harris81} as a tool to quantify
629: the amount of stars in GCs with respect to the amount of field stars
630: in a galaxy. To derive $S_N$ we choose to not do any extrapolation to
631: derive the total $N_{GC}$ and the total magnitude, $m_V$, to avoid the
632: uncertainty introduced by the profile extrapolation. Instead, adopting
633: an approach similar to \citet{blake97}, we adopt the number of GCs
634: found in the ACS sampling area, after completeness correction, and the
635: total magnitude obtained from the same masked frame that we used to
636: detect GCs ($m_V = 10.9 \pm 0.1$ mag).
637:
638: As noted by \citet{blake97}, if the GC system follows the same radial
639: profile of the galaxy light in the outer regions, there should be
640: little difference between the global $S_N$ and this $S_N$ over a
641: limited area, referred to as ``metric $S_N$'' by \citeauthor{blake97}
642:
643: With these assumptions, we derive $S_N= 1.4 \pm 0.3$, which is within
644: the normal $S_N$ range for galaxies in the same luminosity class of
645: NGC\,5866 \citep[$S_N$=0.7-7.5,][]{kp97}. We have also estimated the
646: total $S_N$ extrapolating the total number of GCs. By integrating over
647: the surface density, corrected for magnitude completeness, we obtained
648: $N_{GC}^{Total} \sim400$. Adopting a total magnitude $m_V\sim9.8\pm
649: 0.1$ mag from the RC3 catalog, we find $S_N^{Total}\sim 1.6\pm 0.4$,
650: in nice agreement with the value found over the limited ACS area.
651:
652: Finally, since the GCs are evenly distributed between the two
653: subpopulations, we find that the specific frequency for the blue and
654: red subsystems is $S_N^{blue} \sim S_N^{red} \sim 0.7$.
655:
656:
657: \section{Conclusions}
658: We have used archival HST data to study the GC system of NGC\,5866
659: galaxy in the B, V and R passbands. Using standard color, size and
660: shape selection criteria, the final list of GC candidates consists of
661: 109 objects. The estimated contamination from background galaxies is
662: $\sim$ 6 \% of the sample, while essentially no contamination is
663: expected from foreground stars.
664:
665: The color distribution of the GC system has a bimodal form. Adopting
666: the color-metallicity relations from R05 models, we find that the blue
667: and red peaks correspond to a metallicity [Fe/H]$\sim-1.5$, and $-$0.6
668: dex, respectively. These metallicity peaks are quite average among the
669: galaxies with known color bimodality. Similarly, we have found
670: that the specific frequency of GCs is $S_N=1.4 \pm 0.3$, which is a
671: normal value for galaxies of the same luminosity class of NGC\,5866.
672:
673: By inspecting the color magnitude diagram of the two GC subsystems, a
674: color-luminosity correlation appears in the blue GCs, with brighter
675: objects being slightly redder than the fainter ones. Interpreting this
676: feature as a luminosity-metallicity correlation, the bright blue GCs
677: are found to be 0.4 dex more metal rich than the fainter ones.
678:
679: Inspecting the spatial properties of the GCs, we find that $i$) the
680: red subpopulation appears to be more centrally concentrated than the
681: blue one; $ii$) the radial density of the total GC population, and of
682: the two subpopulations taken separately, follows a power law radial
683: profile; and that $iii$) the GCs have larger half-light radii at
684: larger galactocentric distances. We do not find any sign of a
685: significant difference between the average sizes of the
686: subpopulations.
687:
688: Taken separately, the two GC subsystems do not show significant color
689: gradients. However, due to the different radial concentrations, the
690: total GC system shows a color gradient, very similar to the one of
691: field stars, except that the GCs are systematically $\sim$0.1 mag
692: bluer in (B-R)$_0$. If the GC to field stars color offset is simply
693: associated to a metallicity difference, then the population of field
694: stars at a given galactocentric distance is, on average, 0.3 dex more
695: metal rich of the GC population at the same radius.
696:
697: A comparison of data with the R05 models has shown that no old, very
698: metal rich (t$>$9 Gyr, [Fe/H]$>$0) cluster is present. Moreover, if
699: only models in the age range 9$\leq$ t (Gyr) $\leq$ 14 are taken into
700: account, then the metallicity range for the GC system in this galaxy
701: is found to be $-2.3\lesssim$ [Fe/H] $\lesssim 0.0$. When compared to
702: Galactic and Sombrero data, we find that NGC\,5866 and Milky Way GC
703: share similar metallicity range, while the Sombrero has a tail of very
704: metal rich GCs ([Fe/H] $\sim0.6$).
705:
706: The study of the GC luminosity function has shown that the GC sample
707: is fairly complete out to $\sim$1.2 mag below the expected TOM. A
708: gaussian fit to the GCLF provided us V$_0^{TOM}=23.46 \pm 0.06$
709: mag. Adopting a distance modulus $\mu_0=30.75 \pm 0.08$ from the
710: weighted average of SBF and PNLF distances, implies
711: M$_{V,0}^{TOM}=-7.29 \pm 0.10$ mag, in agreement with the mean value
712: from \citet{richtler03}.
713:
714:
715: \acknowledgments
716:
717: We thank the anonymous referee for helping us to improve this paper
718: with useful suggestions. This work was supported by the NASA grant
719: AR-10642, and by COFIN 2004 under the scientific project ``Stellar
720: Evolution'' (P.I.: Massimo Capaccioli).
721:
722:
723:
724:
725: \clearpage
726:
727:
728: \begin{figure}
729: \epsscale{1.}
730: \plottwo{fig1a.ps}{fig1b.ps}
731: \caption{Left panel: NGC\,5866 galaxy image from the HST ACS.
732: The rectangle shows the location of area zoomed in the right
733: panel. Right panel: A zoom of the disk of dust in NGC\,5866 as seen in
734: the residual frame.
735: \label{image}}
736: \end{figure}
737:
738:
739: \begin{figure}
740: \epsscale{1.}
741: \plotone{fig2.eps}
742: \caption{Color histograms of the final sample of GC candidates.
743: \label{colhisto}}
744: \end{figure}
745:
746:
747: \begin{figure}
748: \epsscale{1.}
749: \plotone{fig3.eps}
750: \caption{Left panel: Galactic GC (B-R)$_0$ color and [Fe/H] data
751: (full dots) compared with the linear fit to the R05 models (dotted line,
752: see text). The fit to the models nicely agrees with observational
753: data. Right panel: as left panel, but for (B-V)$_0$ color.
754: \label{ggc_colfeh}}
755: \end{figure}
756:
757:
758:
759: \begin{figure}
760: \epsscale{1.}
761: \plotone{fig4.eps}
762: \caption{Color-Magnitude Diagrams of the final sample of GC. There is
763: some evidence of the color-magnitude tilt for the blue GC
764: subpopulation (empty squares). No significant tilt is found in the
765: red peak GC candidates (full squares). The dashed lines show the
766: least-squares fit to the data.
767: \label{cmd}}
768: \end{figure}
769:
770: \begin{figure}
771: \epsscale{1.}
772: \plotone{fig5.eps}
773: \caption{Upper panel: The surface density of the entire GC population.
774: The least-squares profile fit (solid line) shows that the radial
775: density profile is well reproduced by a power law, $\log \sigma_{GC}
776: = (-1.3 \pm 0.1) \log R_{gc} + (1.0 \pm 0.1)$. Lower panel: Same as
777: upper panel but the two subpopulations are shown separately. The
778: dotted (dashed) line represents a linear fit to the data of red (blue)
779: GCs. The fitted lines show that the red subpopulation is more
780: centrally concentrated than the blue one.
781: \label{radist}}
782: \end{figure}
783:
784:
785: \begin{figure}
786: \epsscale{1.}
787: \plotone{fig6.eps}
788: \caption{GC colors versus the galactocentric distance. The crosses
789: mark actual GC data. Full big squares correspond to a running average
790: of data, error bars are shown with small squares. The upper two panels
791: show that no significant radial color gradient exists if the two
792: subpopulations are taken separately into account. If the whole GC
793: system is considered, a non negligible color gradient is found, with
794: bluer GC colors in the galaxy outskirts (down left panel). The galaxy
795: color profile (thick solid line, uncertainties shown with thin lines),
796: and the least-squares line of the GC system radial profile (dashed
797: line) are shown in the lower right panel.
798: \label{colrad}}
799: \end{figure}
800:
801:
802:
803: \begin{figure}
804: \epsscale{1.}
805: \plotone{fig7.eps}
806: \caption{Cluster sizes versus the galactocentric distance. The dashed
807: lines show the least-squares fit to the data. These data show the
808: tendency of larger GC radii at larger galactocentric distances.
809: \label{reffrad}}
810: \end{figure}
811:
812:
813: \begin{figure}
814: \epsscale{1.}
815: \plotone{fig8.eps}
816: \caption{Observed GC colors compared with the R05 SSP models for ages
817: $1 \leq t (Gyr) \leq$ 14 and different metallicities (as
818: labeled). {\it [See electronic version of the Journal for a color
819: version of the figure]}
820: \label{spot}}
821: \end{figure}
822:
823:
824: \begin{figure}
825: \epsscale{1.}
826: \plotone{fig9.eps}
827: \caption{Color-color comparisons (upper panels), and (B-R)$_0$ color
828: histograms (lower panels) for Galactic, NGC\,5866 and Sombrero clusters, as
829: labeled. The least-squares fit lines are also shown. The Sombrero GC
830: system includes a rich component of red, super metal rich globular
831: clusters which is practically missing in the other two galaxies
832: system. On the other hand, the blue tails of the GC distributions are
833: quite similar for the three galaxies.
834: \label{3gxy}}
835: \end{figure}
836:
837:
838: \begin{figure}
839: \epsscale{1.}
840: \plotone{fig10.eps}
841: \caption{V-band completeness functions within three different
842: galactocentric annuli.
843: \label{cfv}}
844: \end{figure}
845:
846: \begin{figure}
847: \epsscale{1.}
848: \plotone{fig11.eps}
849: \caption{The observed GCLF histogram (corrected for
850: incompleteness) and the best fit gaussian line. The TOM obtained is
851: V$_0^{TOM} = 23.46 \pm 0.06$ mag (dashed line), dispersion $\sigma =
852: 1.13 \pm 0.05$. Adopting such TOM magnitude we derive the absolute TOM
853: $M_{V,0}^{TOM}=-7.29 \pm 0.10$ mag, in agreement with the ``universal''
854: value from \citet{richtler03}. The 95\% completentess magnitude limits
855: at different galactic annuli are also shown in the figure with
856: vertical arrows.
857: \label{gclf}}
858: \end{figure}
859:
860:
861:
862: \clearpage
863: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
864: %%%%%%%%%%%TABLES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
865: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
866:
867: \input{tab1}
868:
869: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
870: %%%%%%%%%%%BIBLIO %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
871: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
872: \clearpage
873:
874:
875: \bibliographystyle{apj}
876: \bibliography{cantiello_jun07}
877:
878: \end{document}
879:
880: %%%%%%%FIGURES REJECTED
881:
882:
883: \begin{figure}
884: \epsscale{1.}
885: \plotone{../SBF/GCLF/ishape.eps}
886: \caption{Comparison of effective radii $r_{eff}$ derived
887: in the B, V and R images. As in S06 we have
888: adopted an iterative procedure to reject the GCs with
889: estimated $r_{eff}$ 3$\sigma$ out of the computed average.
890: In both panels the bigger circle refer to the whole sample
891: of GC candidates; smaller empty circles are the GC selected after
892: the second iteration. Full dots mark the final good GC candidates.
893: \label{fig2}}
894: \end{figure}
895:
896: \begin{figure}
897: \epsscale{1.}
898: \plotone{../SBF/GCLF/bc03.eps}
899: \caption{As Fig. \ref{spot} but for the \citet{bc03} SSP models.
900: {\it See electronic version of the Journal for a color version of the
901: figure}
902: \label{bc03}}
903: \end{figure}
904: