1: \documentclass[useAMS,usenatbib]{mn2e}
2: \usepackage{psfig}
3: \usepackage{graphicx}
4: %\usepackage{multirow}
5: %\usepackage{subfigure}
6:
7: \title[Simple Stellar Population Models as probed by ESO~121--SC03]
8: {Simple Stellar Population Models as probed by the Large Magellanic Cloud
9: Star Cluster ESO 121--SC03}
10:
11: \author[Y. Xin et al.]{Y. Xin$^{1, 2, 3}$\thanks{E-mail:
12: y.xin@sheffield.ac.uk}, L. Deng$^{2}$, R. de Grijs$^{1, 2}$,
13: A. D. Mackey$^{4}$, and Z. Han$^{3}$\\
14: $^{1}$Department of Physics and Astronomy, The University of
15: Sheffield, Hicks Building, Hounsfield Road, Sheffield S3 7RH\\
16: $^{2}$National Astronomical Observatories, Chinese Academy of
17: Sciences, Beijing 100012, P. R. China\\
18: $^{3}$National Astronomical Observatories/Yunnan Observatory,
19: Chinese Academy of Sciences, Kunming, Yunnan 650011,
20: P. R. China\\
21: $^{4}$Institute for Astronomy, University of Edinburgh, Royal
22: Observatory, Blackford Hill, Edinburgh, EH9 3HJ}
23:
24: \begin{document}
25:
26: \date{Accepted ---. Received ---; in original form ---}
27:
28: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2007}
29:
30: \maketitle
31:
32: \label{firstpage}
33:
34: \begin{abstract}
35: The presence of blue straggler stars (BSs) in star clusters has
36: proven a challenge to conventional simple stellar population (SSP)
37: models. Conventional SSP models are based on the evolution theory of
38: single stars. Meanwhile, the typical locations of BSs in the
39: colour-magnitude diagram of a cluster are brighter and bluer than
40: the main sequence turn-off point. Such loci cannot be predicted by
41: single-star evolution theory. However, stars with such properties
42: contribute significantly to the integrated light of the cluster. In
43: this paper, we reconstruct the integrated properties of the Large
44: Magellanic Cloud cluster ESO~121--SC03, the only cluster populating
45: the well-known age gap in the cluster age distribution, based on a
46: detailed exploration of the individual cluster stars, and with
47: particular emphasis on the cluster's BSs. We find that the
48: integrated light properties of ESO~121--SC03 are dramatically
49: modified by its BS component. The integrated spectral energy
50: distribution (ISED) flux level is significantly enhanced toward
51: shorter wavelengths, and all broad-band colours become bluer. When
52: fitting the fully integrated ISED of this cluster based on
53: conventional SSP models, the best-fitting values of age and
54: metallicity are significantly underestimated compared to the true
55: cluster parameters. The age underestimate is $\sim40$ per cent if we
56: only include the BSs within the cluster's half-light radius and
57: $\sim60$ per cent if all BSs are included. The corresponding
58: underestimates of the cluster's metallicity are $\sim30$ and
59: $\sim60$ per cent, respectively. The populous star clusters in the
60: Magellanic Clouds are ideal objects to explore the potential
61: importance of BSs for the integrated light properties of more
62: distant unresolved star clusters in a statistically robust manner,
63: since they cover a large range in age and metallicity.
64:
65: \end{abstract}
66:
67: \begin{keywords}
68: blue stragglers -- globular clusters: individual: ESO~121--SC03 --
69: Magellanic Clouds -- galaxies: star clusters.
70: \end{keywords}
71:
72: \section{Introduction}
73: Even using improved observational technologies, the number of
74: galaxies with resolved stellar contents is limited. Comparison of
75: the observed integrated spectral properties with models of
76: ``simple'' stellar populations (SSPs; single-age, single-metallicity
77: stellar populations), i.e. using the so-called population synthesis
78: technique, is therefore a practical approach to studying the
79: formation and evolution processes in unresolved galaxies and their
80: components. As the basic building blocks of the evolutionary
81: population synthesis (EPS) method, conventional SSP models are
82: generally constructed based on the evolution theory of single stars.
83: (We specify the conventional SSP models throughout this paper as the
84: theoretical SSP models based on single-star evolution theory.) Thus
85: far, accuracy assessments of the existing suites of conventional SSP
86: models have not yet adequately considered a potentially significant
87: problem revealed by observations of star clusters.
88:
89: By assuming that all the original member stars within a cluster are
90: born at the same time from the same progenitor molecular cloud, so
91: that they have the same age and metallicity, star clusters are
92: generally considered as the closest counterparts in the real world
93: to idealised SSPs. The major differences between a star cluster and
94: a conventional model SSP of the same age and metallicity can be
95: illustrated by fitting the observed colour-magnitude diagram (CMD)
96: of the cluster with a theoretical single-star isochrone. Most of the
97: member stars can be well fitted by the isochrone in terms of their
98: positions in the CMD. The number distributions along the isochrone
99: can be retrieved via the adopted stellar initial mass function
100: (IMF), and the number of low-mass member stars can be conserved in
101: the presence of evaporation due to dynamical evolution. Then, the
102: most prominent difference should be attributed to those member stars
103: ``straggling'' away from the isochrone. Yellow and red stragglers
104: (Protegies~Zwart et al. 1997a,b) are not considered in the work,
105: because they are comparatively rare and less luminous than blue
106: stragglers (BSs). The underlopers and faint stars bluer than the
107: main sequence (MS) are not bright enough to be important. Therefore,
108: special attention has only been focussed on the BSs in our previous
109: work (Deng et al. 1999; Xin \& Deng 2005, hereafter XD05; Xin, Deng
110: \& Han 2007, hereafter XDH07), in view of their potential
111: non-negligible influence on the integrated properties of star
112: clusters.
113:
114: BSs have been widely observed in stellar systems of all scales and
115: complexities, as shown by, e.g. Ahumada \& Lapasset (2007) for
116: Galactic open clusters (OCs), Piotto et al. (2004) for Galactic
117: globular clusters (GCs), Carney et al. (2005) for the Galactic field
118: stellar population, Johnson et al. (1999) for Large Magellanic Cloud
119: (LMC) clusters, Alcaino et al. (2003) for Small Magellanic Cloud
120: (SMC) clusters, and Mapelli et al. (2007) for dwarf spheroidal
121: galaxies. Considering their general locations in the CMD, i.e.,
122: brighter and bluer than the MS turn-off point, BSs are remarkably
123: hotter than the most luminous ``normal'' MS stars. Typically for old
124: Population I star clusters, when the red clump giants (RCGs) are
125: populated instead of the blue horizontal branch (HB), BSs can
126: significantly enhance the cluster' integrated spectrum at
127: ultraviolet (UV) and blue wavelengths -- the modifications may lead to
128: the star cluster being predicted as younger or of lower metallicity
129: based on conventional SSP models, and therefore cause uncertainties
130: when simply applying the conventional SSP models in EPS studies
131: (XD05; XDH07). Meanwhile, the most likely formation mechanisms of
132: BSs are all related to stellar interactions, including mass transfer
133: in close binaries (e.g., Tian et al. 2006; and references therein)
134: and stellar collisions in high-density regions (e.g., Sills et al.
135: 2005; and references therein). This provides the theoretical support for
136: this empirical work, since the consequences of the products of
137: stellar interactions are not included in the conventional SSP
138: models.
139:
140: In order to empirically correct the conventional SSP models for the
141: presence of luminous BSs, we construct the integrated spectrum of an
142: SSP (i.e., a realistic star cluster) by analysing the individual
143: stars, after careful consideration of their membership probability.
144: We assume that the stellar population of a cluster is composed of
145: two components: (i) all member stars that can be well fitted by a
146: single-star isochrone represent a conventional SSP for the cluster's
147: real age and metallicity (e.g., SSP models from Bruzual \& Charlot
148: 2003; hereafter BC03); and (ii) BSs are responsible for the
149: modifications to the conventional SSPs, and their contributions are
150: considered by including the individual spectra of BSs. By combining
151: these two components we may obtain a better approximation to the
152: true SSP corresponding to the cluster of interest (see for details
153: XD05 and XDH07).
154:
155: In our previous work (XD05; XDH07), a sample of 100 Galactic OCs
156: spanning an age range $\in[0.1, 10]$ Gyr and a range in metallicity,
157: $Z\in[0.005, 0.035]$ (where solar metallicity is Z$_\odot = 0.020$),
158: was used to detect BS contribution semi-empirically. The general
159: results from our Galactic OC studies show that the integrated
160: spectral properties of the sample clusters are dramatically modified
161: by their BS components. A preliminary assessment of the
162: uncertainties inherent to the conventional SSP models was made.
163: Using either spectra or broad-band colours, the resulting ages
164: and/or metallicities will be underestimated significantly:
165: conservatively, the age underestimates are $\sim50$ per cent, as are
166: the likely underestimates of the metallicity (although they are less
167: accurately constrained because of our lower resolution in
168: metallicity).
169:
170: The results from the Galactic OCs clearly showed the non-negligible
171: contributions of BSs to the integrated properties of SSPs. However,
172: the numbers and distributions of BSs in OCs are affected by
173: significant stochastic uncertainties (Ahumada \& Lapasset 1995,
174: 2007). To make the results and conclusions more reliable and
175: convincing, we expand our working sample by including star clusters
176: in the Magellanic Clouds (MCs). MC clusters hold significant
177: advantages over Galactic OCs: (i) they span a wider range in both
178: age and metallicity, and (ii) they are significantly more massive
179: than their counterparts in the Galaxy, and therefore provide better
180: statistics on an individual cluster basis (e.g., Mackey \& Gilmore
181: 2003a,b; de~Grijs \& Anders 2006).
182:
183: In this paper, the LMC cluster ESO~121--SC03 is analyzed using its
184: integrated spectral properties. This serves as the first example to
185: detect BS contributions to the conventional SSP models using a
186: massive intermediate-age cluster in a low-metallicity environment,
187: based on a novel approach and better statistics than we were
188: afforded by our use of Galactic OCs. ESO~121--SC03
189: ($\alpha_{2000}=06^{\rm h}02^{\rm m}01^{\rm s}.36$,
190: $\delta_{2000}=-60^\circ31'22''.6$) is a distant northern LMC
191: cluster, lying at a projected angular separation of $\sim10^\circ$
192: from the LMC centre. It is described as a ``unique'' LMC cluster by
193: Mackey, Payne \& Gilmore (2006; hereafter MPG06), because it is the
194: only known cluster to lie in the LMC age gap. A significant number
195: of previous studies (e.g., Mateo et al. 1986; Bica et al. 1998;
196: MPG06) claim an absolute age of ESO~121--SC03 in the range of $8-10$
197: Gyr. Mateo et al. (1986) obtained [Fe/H]~$=-0.9\pm0.2$ combined with
198: a reddening of $E(B-V)=0.03$ mag. MPG06 derive
199: [Fe/H]~$=-0.97\pm0.01$ and $E(V-I)=0.04\pm0.02$ for the cluster.
200: MPG06 also mark a region in the CMD used to define BS candidates in
201: the cluster, but they do not study the BS population in detail.
202:
203: We present our photometric data reduction steps in Section 2. In the
204: subsequent sections we present the details of the model construction
205: and our main results. The CMD and the best-fitting isochrone of the
206: cluster are discussed in Section 3. The identification of BSs in the
207: CMD is described in Section 4. A discussion of the contributions of
208: BSs to the integrated spectral properties of the cluster is presented
209: in Section 5. In Section 6, the synthetic integrated spectral energy
210: distribution (ISED) of the cluster is fitted with conventional SSP
211: models in order to assess the uncertainties. Finally, a summary and
212: the conclusions of this study are presented in Section 7.
213:
214: \section{Data reduction}
215:
216: We take advantage of the accurate photometric data taken with the
217: Advanced Camera for Surveys (ACS) Wide Field Channel (WFC) on board
218: the {\sl Hubble Space Telescope (HST)} as part of the {\sl HST} Cycle
219: 12 snapshot survey of MC star clusters (programme 9891, PI
220: G. Gilmore). The ACS WFC consists of two 2048$\times$4096 pixel CCDs
221: with a scale of $\sim 0.05$ arcsec pixel$^{-1}$, separated by a gap of
222: $\sim 50$ pixels, and approximately covers a field of view of 202
223: $\times$ 202 arcsec$^2$. The frame was taken in each of two filters,
224: F555W and F814W. Exposure times were 300s and 200s, respectively.
225: Further instrumental and observational details can be found in series
226: of publications based on this programme, e.g., Mackey \& Gilmore
227: (2004), MPG06, and Mackey \& Broby~Nielsen (2007).
228:
229: We retrieved the FITS files from the STScI data archive. The
230: observations were reduced using the STScI reduction pipeline, i.e.,
231: they have had bias and dark current frames subtracted and are divided
232: by flat-field images. The photometry was performed with the DOLPHOT
233: software (Dolphin 2000), specifically the ACS module. Before we
234: performed the photometry, we first prepared the images using the
235: DOLPHOT tasks \textit{acsmask} and \textit{splitgroups}. These two
236: packages mask out all bad pixels in the images and then split the
237: multi-image STScI FITS files into a single FITS file per chip,
238: respectively. We then used the main DOLPHOT routine to make
239: photometric measurements on the pre-processed images, using the F814W
240: drizzled frame as the position reference. All running parameters were
241: set to the recommended values in the DOLPHOT manual. The output
242: photometry was obtained in the VEGAMAG system and corrected for
243: charge-transfer efficiency degradation. Photometric calibrations and
244: transformations were done following Sirianni et al. (2005).
245:
246: To obtain a high-quality CMD of the cluster, we used three parameters
247: from DOLPHOT to filter the photometric results. The ``sharpness''
248: measures how broad the intensity profile of a detected object is
249: relative to the point-spread function (PSF). It is zero for a
250: perfectly fit star, positive for an object that is too sharp (i.e.,
251: perhaps a cosmic ray), and negative for an object that is too broad
252: (i.e., perhaps a blend, cluster, or background galaxy). The
253: ``crowding'' parameter is given in magnitudes, and measures how much
254: brighter the object would have been measured had nearby objects not
255: been fit simultaneously. For an isolated star the ``crowding'' value
256: is zero. The $\chi^2$ parameter represents the quality of the PSF
257: fitting. In this paper, we selected only objects with $-0.3 \leq$
258: sharpness $\leq 0.3$, crowding $\leq 0.5$ mag, and $\chi^2 \leq 0.25$
259: in both frames. Meanwhile, we only kept objects classified as good
260: star (object type 1) and star errors of types 1--7 by DOLPHOT, which
261: are referred to as ``usable'' in the DOLPHOT manual.
262:
263: After following through this procedure, 2824 objects were detected in
264: the field of ESO~121-SC03. In order to evaluate the completeness of
265: the data, DOLPHOT was employed again, in artificial-star mode. We
266: generated $\sim 10^6$ fake stars, which were created with the DOLPHOT
267: task \textit{acsfakelist}. The limits of the artificial stars were set
268: to 16.0--28.0 mag in magnitudes and $-0.50-2.00$ mag in colours in the
269: CMD. The fake stars were binned in four-dimensional space: 200 pixels
270: in {\it x}- and {\it y}- positions, respectively, 0.2 mag in
271: brightness and 0.25 mag in colour. The completeness function of the
272: cluster is shown in Fig.~\ref{fig1}, integrated over position and
273: colour, as a function of $m_{\rm F555W}$. Completeness declines for
274: $m_{\rm F555W} \leq 18$ mag because we do not have any stars of that
275: brightness in the CMD.
276:
277: \begin{figure}
278: \includegraphics[width=8cm]{fig1.ps}
279: \caption{Completeness function of ESO~121--SC03. The function has been
280: integrated over position and colour as a function of $m_{\rm F555W}$.}
281: \label{fig1}
282: \end{figure}
283:
284: \section{Colour-magnitude diagram}
285:
286: The final, cleaned CMD of ESO~121--SC03 in $m_{\rm F555W}$ and
287: $m_{\rm F814W}$ magnitudes is presented in Fig.~\ref{fig2}. It is
288: clearly shown that the CMD is well defined and contains little field
289: contamination. The perfectly represented evolutionary stages, e.g.,
290: MS, red-giant branch (RGB), and RCG, indicate the high accuracy of
291: the photometry and the validity of the reduction procedure. The CMD
292: reaches more than 3 magnitudes below the Main sequence turn-off
293: (MSTO). All these aspects help us to perform a detailed and highly
294: accurate statistical study of the cluster.
295:
296: \begin{figure}
297: \includegraphics[width=8cm]{fig2.ps}
298: \caption{Cleaned CMD of ESO~121--SC03. The CMD contains 2824
299: detections. The photometric data is plotted in the VEGAMAG
300: system.}\label{fig2}
301: \end{figure}
302:
303: As described in Section 1, the synthetic ISED of a cluster is
304: constructed by assuming that the entire stellar population of the
305: cluster can be decomposed into two components: one accounts for all
306: member stars that are well fitted by a given isochrone (of single
307: stars) for the appropriate cluster age and metallicity; the other
308: includes only the cluster's BS population, which is therefore referred
309: to as the BS component. All other member stars deviating from the
310: fitted isochrone can be neglected because of their low luminosity and
311: relatively small number. More details justifying this approach can be
312: found in XD05.
313:
314: To follow the same procedure for ESO~121--SC03, we first need to
315: find the isochrone that best matches the observed features of the
316: cluster's CMD, and use this to derive accurate physical parameters,
317: including the cluster's age, metallicity, colour excess and distance
318: modulus. These parameters are very important for our construction of
319: the ISED of the SSP component and for defining the BS population.
320: The photometry in the VEGAMAG system is transformed into
321: Johnson--Cousins \textit{V} and \textit{I} magnitudes following the
322: prescription of Sirianni et al. (2005), and subsequently the new CMD
323: in $V,(V-I)$ is used to find the best-fitting isochrone. The fit
324: quality and corresponding details are displayed in Fig.~\ref{fig3}.
325:
326: As the first step to identify the best-fitting isochrone, we attempted
327: to identify the MSTO in the CMD by counting the number of stars in
328: bins of magnitude (0.01 mag) and colour (0.001 mag) in the turn-off
329: region; the intersection of two bins with maximum stellar numbers, one
330: in magnitude and the other in colour, is regarded as the cluster's
331: MSTO. The resulting MSTO locus is at $(V-I)_{\rm TO} = 0.606\pm0.005$
332: mag and $V_{\rm TO} = 21.94\pm0.01$ mag. Next, the same procedure is
333: adopted to identify the main sequence ridge line (MSRL) in the CMD:
334: the numbers of stars are measured in magnitude bins of 0.5 mag, from
335: the MSTO to 3 magnitudes below the MSTO. In Fig.~\ref{fig3}, the
336: calculated MSTO is marked as the solid bullet, and the MSRL is marked
337: by solid triangles.
338:
339: We used the Padova2000 theoretical isochrones (Girardi et al. 2000) to
340: fit the observed CMD. The original Padova2000 isochrones were
341: interpolated to [Fe/H]~$=-0.97$ dex (the value given by MPG06). In
342: total nine points, including the six MSRL points, the bottom of the
343: RGB, the mean location of the RCG, and the point on the RGB at the
344: level of the intermediate magnitude between the points of the RGB
345: bottom and the mean RCG locus, were used as fiducial points to
346: evaluate the fit quality. Next, we calculated the standard deviation
347: ($\sigma$) for the offsets between the nine fiducial points in the CMD
348: and the isochrone, in order to obtain the minimum $\sigma$ value. The
349: isochrone fitting also offered us estimates for the apparent distance
350: modulus, $(m-M)_V$, and the foreground extinction suffered by the
351: cluster.
352:
353: Given the above results and discussion, we adopted the metallicity of
354: [Fe/H]~$=-0.97$ dex, and we found a best-fitting isochrone of
355: age~=~8.9$^{+1.1}_{-0.3}$ Gyr, with $(m-M)_V=18.30\pm0.06$ mag and
356: $E(V-I)=0.03\pm0.01$ mag. These values will be used in the remainder
357: of this paper.
358:
359: Our measurements for ESO~121--SC03 are in good agreement with previous
360: results. We derived that the age of the cluster is $8.9^{+1.1}_{-0.3}$
361: Gyr, which is entirely consistent with the age estimates of MPG06, who
362: provided an age range of $8.3-9.8$ Gyr for ESO~121--SC03. Similarly,
363: Mateo et al. (1986) found that this cluster is $10\pm2$ Gyr or $8\pm2$
364: Gyr old if the LMC distance modulus is 18.2 or 18.7 mag,
365: respectively. Bica et al. (1998) gave an age of 8.5 Gyr for the
366: cluster from Washington photometry. Besides the similar age value,
367: our result confirm the conclusion of MPG06 and Mateo et al. (1986)
368: that ESO~121--SC03 is the only known cluster lying within the LMC age
369: gap. The LMC has a unique cluster formation history where nearly all
370: of its star clusters were formed either $\sim13~Gyr$ ago or less than
371: $\sim3~Gyr$ ago (Da~Costa 1991; Geisler et al. 1997 and references
372: therein; see also Parmentier \& de Grijs 2007). As suggested by
373: numerical simulations (e.g., Bekki et al. 2004), the origin of the age
374: gap is associated with the dynamical interaction between the LMC and
375: the SMC about $3-4$ Gyr ago. However, no theoretical studies have
376: explained the existence of ESO~121--SC03. Bica et al. (1998)
377: considered that this cluster may have originated from a recent
378: accretion of dwarf galaxies, but Dirsch et al. (2000) concluded that
379: the accretion of ESO~121--SC03 is not necessary, since they found that
380: this cluster has a similar age-metallicity relation as the LMC field
381: stars. In this respect, ESO~121--SC03 may reveal important information
382: about the formation and evolution history of the LMC.
383:
384: The $(m-M)_V$ and $E(V-I)$ values for ESO~121--SC03 were determined
385: based on the offsets in, respectively, magnitude and colour required
386: to align the MSTO of the isochrone with that of the observed
387: CMD. Thus, it is worth considering the consistency between our results
388: and those from other work based on different methods. There are few
389: estimates of the colour excess for ESO~121--SC03. Both Mateo et
390: al. (1986) and Bica et al. (1998) adopted $E(B-V)=0.03$ mag from the
391: Burstein \& Heiles (1982) maps, which approximates to $E(V-I)\sim0.04$
392: mag via the empirical relation $E(V-I)=1.31 E(B-V)$ (e.g., Mackey \&
393: Gilmore 2003a,b). MPG06 obtained metallicity and reddening values of
394: [Fe/H]~$=-0.97\pm0.10$ dex and $E(V-I)=0.04\pm0.02$ mag using the
395: method of Sarajedini (1994). All of these values are consistent with
396: our isochrone-fitting result of $E(V-I)=0.03\pm0.01$ mag (using
397: [Fe/H]~$=-0.97$ dex), which proves that there are no significant
398: errors in our photometric transformation from the ACS/WFC system to
399: the Johnson-Cousins $V$ and $I$ system.
400:
401: We obtained $(m-M)_V=18.30\pm0.06$ mag in this paper, which is larger
402: than the value from MPG06 ($18.11\pm0.09$ mag) and smaller than the
403: standard LMC distance modulus ($18.50\pm0.09$ mag, see e.g., Gratton
404: et al. 2003). In terms of the linear distance, ESO~121--SC03 in our
405: work is $\sim$ 10 per~cent closer to us than the centre of the LMC,
406: but $\sim$ 10 per~cent more distant than suggested by the distance
407: modulus of MPG06. Adopting the optical centre of the LMC at
408: $\alpha=05^{\rm h}20^{\rm m}56^{\rm s}$ and $\delta=-69^\circ28'41''$
409: (Bica et al. 1996), ESO~121--SC03 lies at a projected angular
410: separation of $\sim10^\circ$. Adopting the distance modulus in our
411: work, the linear distance between the centre of the LMC and
412: ESO~121--SC03 is $\sim9.5$ kpc. This value is smaller than that in
413: MPG06 (11.5 kpc), but still agrees with the conclusion in MPG06 that
414: ESO~121--SC03 is one of the most remote known LMC star clusters.
415:
416: \begin{figure}
417: \includegraphics[width=8cm]{fig3.ps}
418: \caption{Cleaned CMD for ESO~121--SC03. The photometry has been
419: transformed from the VEGAMAG system to Johnson--Cousins $V$ and $I$
420: magnitudes. The best-fitting Padova2000 isochrone is overplotted.
421: The corresponding fit parameters are included in the figure legend.
422: In the CMD, we define a region that we will use for the
423: identification of the cluster's BS population. The ZAMS is the
424: Padova1994 isochrone at log(age yr$^{-1}$) = 6.60 and
425: [Fe/H]~$=-0.97$ dex. The solid bullet is the cluster's MSTO. The
426: solid triangles represent the MSRL of the cluster. Pentagons are
427: blue stragglers. See the text (Sections 3 and 4) for
428: details.}\label{fig3}
429: \end{figure}
430:
431: \section{The Blue Straggler Population}
432:
433: As shown in Fig.~\ref{fig3}, ESO~121--SC03 is a ``clean'' cluster and
434: includes an obvious population of BS candidates, which is very
435: advantageous for our main aim of analysing the BS contributions to the
436: ISED of the cluster. However, we still need to be very careful in the
437: selection of BS members, since we do not have access to classical
438: membership probability assessment methods (i.e., proper motions and/or
439: radial velocities) that can reliably remove field stars from the
440: CMD. In this paper, we will try to reduce as much as possible any
441: contamination to the BS population of the cluster due to field stars
442: and improper classifications due to possible member stars in the MSTO
443: region. Two measurements were adopted to avoid exaggerating the BS
444: population in the cluster.
445:
446: First, to properly account for the photometric and systematic errors,
447: we defined a BS region in the CMD using three boundary limits: $V_B,
448: (V-I)_B$, and the theoretical ZAMS, as shown in Fig.~\ref{fig3},
449:
450: \begin{eqnarray}
451: V_{\rm B} &=& V_{\rm TO} - 0.10 \qquad {\rm and}\nonumber \\
452: (V-I)_{\rm B} &=& (V-I)_{\rm TO} - 0.141 \quad .
453: \end{eqnarray}
454:
455: The offset of 0.10 mag in $V_{\rm B}$ relative to the MSTO of the
456: cluster was adopted as an empirical estimate of the photometric and
457: systematic errors for {\sl HST}/ACS WFC observations (Sirianni et
458: al. 2005); the offset of 0.141 mag in $(V-I)_{\rm B}$ was obtained
459: similarly.
460:
461: The theoretical ZAMS is represented by the MS ridge line of the
462: youngest isochrone ($\log(\mbox{age yr}^{-1})=6.60$) in the
463: Padova1994 library (Bertelli et al. 1994). We have taken this
464: approach mainly because there is no such an isochrone in the
465: Padova2000 library: the updated equation of state adopted in
466: Padova2000 is only sensitive to the evolution of stars with initial
467: masses from 0.15 to 7 M$_{\odot}$ (Girardi et al. 2000). Therefore,
468: the age coverage is $\log(\mbox{age yr}^{-1})=[7.80, 10.25]$ in
469: Padova2000 (a 7 M$_{\odot}$ star is still on the MS at
470: $\log(\mbox{age yr}^{-1})=6.60$). Meanwhile, the youngest isochrone
471: fully matches the requirement for a ZAMS in this paper, where it
472: serves as one of the boundaries for the identification of BSs. The
473: ZAMS is not involved in fitting the BS positions in the CMD -- for
474: the fitting we use only the Padova2000 isochrones.
475:
476: Secondly, the obvious central concentration of BSs has been widely
477: observed in stellar systems (e.g., Lee et al. 2003 for the Sextans
478: dwarf spheroidal galaxy; Ferraro et al. 2003 for GCs). Meanwhile,
479: the radial distributions of BSs are observed to be bimodal in some
480: clusters (i.e., strongly peaked in the cluster centre, decreasing at
481: intermediate radii and rising again at greater distances), as shown
482: by, e.g., Ferraro et al. (1997) for M3, Ferraro et al. (2004) for
483: 47~Tuc, Sabbi et al. (2004) for NGC~6752, and Warren et al. (2006)
484: for M5. This behaviour has been suggested as probably due to the
485: relative efficiency of two major BS formation mechanisms: the
486: mass-transfer scenario should mainly populate the BS population in
487: the lower-density outskirts of GCs, and the collisional scenario is
488: the most probable formation mechanism in the high-density inner
489: regions (e.g., Lanzoni et al. 2007; and references therein).
490:
491: Since there are no membership probability measurements available for
492: our cluster, the BSs identified by their loci in the CMD are all
493: considered as BS member candidates, and they are divided into two
494: cases in this paper: (i) BSs inside the half-light radius ($R_{\rm
495: HL}$) of the cluster. This provides a conservative estimate of the BS
496: contribution to the total light of the cluster, and these BSs are
497: marked with open circles in Fig.~\ref{fig3}; (ii) all BSs in the
498: cluster (pentagons in Fig.~\ref{fig3} and solid circles in
499: Fig.~\ref{fig4}). This brackets the maximum alteration to the
500: conventional SSP model for the cluster due to BSs. There are 14 BSs in
501: case (i) among a total of 25 BSs in case (ii).
502:
503: Incompleteness corrections were taken into account for the
504: identification of both the BS population and the cluster's $R_{\rm
505: HL}$. Comparing the BS selection box in Fig.~\ref{fig3} with the
506: completeness function in Fig.~\ref{fig1}, we found that the level of
507: incompleteness was not significant at the luminosity level of the
508: BSs. The integrated correction for the number of the entire BS
509: population is only one BS, but one bright and/or blue BS may cause a
510: significant difference to the modification of the integrated spectral
511: properties of the cluster. Therefore, instead of taking the risk of
512: putting an artificial BS somewhere in the BS region, we decided to
513: keep using the observed BS candidates, and we assumed that the full
514: sample of BS candidates detected in the cluster corresponded to its
515: entire BS population. After the incompleteness correction, we derived
516: the cluster's $R_{\rm HL}$, $R_{\rm HL} =21.8\pm1.0$~arcsec, shown as
517: a large circle in Fig.~\ref{fig4}.
518:
519: \begin{figure}
520: \includegraphics[width=8cm]{fig4.ps}
521: \caption{Pixel coordinates of all 2824 detected stars in
522: ESO~121--SC03. The large circle marks the cluster's half-light radius
523: ($R_{\rm HL}=21.8$ arcsec). The solid circles are the
524: BSs.}\label{fig4}
525: \end{figure}
526:
527: \section{Modifications to the integrated spectral properties}
528:
529: \subsection{SSP model reconstruction~--~the synthetic ISED}
530:
531: The synthetic ISED of the cluster including the BS contribution was
532: constructed by assuming that the stellar population of the cluster
533: is made up of two components, i.e., an SSP component and a BS
534: component. The ISEDs of these two components were derived
535: separately, and the summation of the ISEDs of these two components
536: gives the new ISED of the cluster. The resulting synthetic ISED is
537: then treated as the ISED of the true SSP corresponding to the
538: parameters of the cluster. The contributions to the ISED of the
539: cluster due to BSs are measured by comparing the integrated
540: properties between the conventional ISED (i.e., the ISED of the SSP
541: component) and the synthetic ISED.
542:
543: To build the ISED of the BS component, the spectrum of each single
544: BS must be derived from the observed photometric data. To this end,
545: we used Padova2000 isochrones of the same metallicity but younger
546: ages than the cluster's SSP to match the locus in the CMD of each BS
547: individually. The parameters of effective temperature ($T_{\rm
548: eff}$) and surface gravity ($\log g$) (roughly equivalent to
549: luminosity and mass) for each BS can be determined by interpolating
550: between the two closest isochrones representative of the BS. Based
551: on the $T_{\rm eff}$ and $\log g$ determined in this way, a
552: theoretical spectrum from the stellar spectral library of Lejeune et
553: al. (1997) was assigned to the BS. The theoretical spectra were also
554: interpolated, so that the same metallicity as that of the cluster
555: was adopted. This process was performed individually for all BSs in
556: the cluster, and the ISED of the BS component is given by adding up
557: the individual spectra of the BSs directly.
558:
559: \begin{equation}
560: F_{BS}=\sum_{i=1}^{N_{BS}}f^i_{BS} \quad ,
561: \end{equation}
562:
563: A significant amount of theoretical work has focused on BS formation
564: models. BSs from equal-mass MS star collisions (Benz \& Hills 1987)
565: may suffer from huge mass loss and undergo complete chemical
566: composition mixing -- the unclear clock will, in effect, be reset to
567: a ZAMS star. Unequal-mass MS star collisions (Benz \& Hills 1992)
568: may cause much less mass loss, but also violent mixing in the
569: atmosphere of the more massive star -- this may revert the
570: observational parameters of the more massive star back to those of a
571: helium-rich MS star. Meanwhile, Lombardi et al. (1995, 2002) argued
572: that BSs from stellar collisions would not be expected to have
573: either a significant helium enhancement in the outer envelope, nor
574: replenished fresh hydrogen fuel in their cores. Detailed spectral
575: analysis of BSs and turn-off region MS stars in M67 (e.g., Mathys
576: 1991; Shetrone \& Sandquist 2000) led to the conclusion that the Fe
577: and Ca abundances of BSs are very close to those of the MS stars in
578: the cluster. Liu et al. (in prep.) conclude that most BSs exhibit
579: similar properties to those of regular MS stars, based on a detailed
580: analysis of low-resolution spectra of all BSs in M67.
581:
582: It is clear that rotational support and the stellar chemical profile
583: are very important, because they have profound effects on stellar
584: evolution and on the remnant's position in a CMD (Sills et al.
585: 2000), to which our method of obtaining the BS spectrum is certainly
586: sensitive. However, until either a well-established theory emerges
587: that can predict the actual internal structure of all BSs in a CMD,
588: or high-resolution spectra are available for all BSs in a CMD, the
589: theoretical approaches followed in this series of papers (XD05,
590: XDH07, and the present paper) are fully adequate, as long as we
591: restrict ourselves to low-resolution spectroscopic analysis.
592:
593: The main goal of this series of papers is to remedy the conventional
594: SSP models empirically by analyzing the contributions of the
595: observed BSs quantitatively. To reach this goal, what is really
596: needed is a quantification of the relative modifications to
597: conventional SSP models when the same basic ingredients (i.e.,
598: isochrone library, spectral library, and IMF) are used to describe
599: both the conventional SSP and the BS contribution. As discussed in
600: the previous sections, the positions of BSs in the observed CMD are
601: fitted with Padova2000 isochrones, and the BS spectra are extracted
602: from the Lejeune et al. (1997) stellar spectral library. Therefore,
603: the conventional SSP models (BC03) based on the same libraries and
604: using a Salpeter (1955) IMF are adopted to represent the properties
605: of the SSP components in star clusters. We interpolated the ISEDs
606: across a grid of age and metallicity to construct the ISED of the
607: SSP component so that they match the parameters of the cluster.
608:
609: For an SSP of age $t$ and metallicity $Z$, assuming an IMF
610: $\phi(m)$, the ISED of the SSP component is
611: \begin{equation}
612: F_{\rm SSP}(\lambda,t,Z)=A\int_{m_l}^{m_u}\phi(m)f(\lambda,m,t,Z){\rm
613: d}m \quad ,
614: \end{equation}
615: where $f(\lambda,m,t,Z)$ is the flux of a single star of mass $m$,
616: age $t$, and metallicity $Z$; $m_u$ and $m_l$ are the upper and
617: lower integration limits in mass, respectively. $A$ is a
618: normalization constant, which is used to restore the flux of the SSP
619: component to its real intensity.
620:
621: The ISEDs of BC03 SSP models are all normalized to a total mass of 1
622: M$_\odot$ in stars, at $t = 0$. Therefore, we need to calibrate the
623: ISEDs of the BS and SSP components similarly. In order to do so, we
624: count the number of stars in the CMD in the magnitude interval
625: between the MSTO and the point at two magnitudes below the MSTO on
626: the MS. This number is defined as $N_2$ (Ahumada \& Lapasset 1995).
627: $N_2$ is used to calculate the normalization constant $A$ in Eq.
628: (3). By using a Salpeter IMF, $A$ can be derived as follows:
629:
630: \begin{equation}
631: N=A\int_{m_1}^{m_2}m^{-2.35} {\rm d} m=N_2 \quad ,
632: \end{equation}
633: where $m_2$ and $m_1$ are, respectively, the stellar masses at the
634: MSTO and at the point 2 magnitudes below the MSTO. To ensure the
635: integrity of the stellar number counts in this range, the photometry
636: should be sufficiently deep. The photometric observational data used
637: in this paper meets this requirement. Corrections for incompleteness
638: in this mass range are also applied. For a detailed description of
639: the model construction procedure, see XD05 and XDH07.
640:
641: Incidentally, the choice of the IMF also introduces uncertainties in
642: the SSP's integrated properties, but it is unimportant for the
643: discussion in this series of papers. The conventional SSP models
644: using a Salpeter IMF are only used as a reference to the ISEDs of
645: the SSP components, what is really important in the current context
646: is the contributions of BSs to the ISED; the properties of the BSs
647: in our cluster are determined individually, and independently of the
648: adopted IMF properties.
649:
650: \subsection{Modification to the ISED}
651:
652: According to their loci in the CMD (Fig.~\ref{fig3}), BSs are the
653: bluest cluster members, with extraordinary luminosities, and thus they
654: may cause very significant modifications to the conventional view of
655: the integrated spectral properties of a cluster. The contributions of
656: BSs to a cluster's ISED are demonstrated in Fig.~\ref{fig5} for two
657: cases: (i) considering only the BSs inside $R_{\rm HL}$; and (ii) all
658: BSs in the cluster.
659:
660: In Fig.~\ref{fig5}, the solid line is the ISED of the conventional SSP
661: model with an isochrone-fitted (regarded as ``true'') age and
662: metallicity representative of that of the cluster, i.e., the ISED of
663: the SSP component. The thin dotted line is the ISED of the BS
664: component for case (i); the thin dashed line is the ISED of the BS
665: component for case (ii). The heavy dotted line is the synthetic ISED
666: of the SSP component and the case-(i) BS component; the heavy dashed
667: line is the synthetic ISED of the SSP component and the case-(ii) BS
668: component. Although they are not dominated by BSs, both synthetic
669: ISEDs show significant enhancements across a wide range of UV, blue
670: and visual wavelengths. The synthetic ISEDs become undoubtedly hotter
671: than the ISED of the conventional SSP (solid line in
672: Fig.~\ref{fig5}). Taking the values of the flux at 5510 {\AA} as an
673: example, the energy enhancements will be, respectively, $\sim10$ and
674: $\sim 18$ per cent if we include the case-(i) and case-(ii) BS
675: components.
676:
677: \begin{figure}
678: \includegraphics[width=8cm]{fig5.ps}
679: \caption{ISED modifications. The BS contributions are presented for
680: two cases. The solid line is the ISED of the SSP component. The thin
681: dotted line is the ISED of the BS component for case (i): BSs within
682: $R_{\rm HL}$. The thin dashed line is the ISED of the BS component
683: for case (ii): all BSs in the cluster. The heavy dotted line is the
684: synthetic ISED of the SSP component and the case-(i) BS component.
685: The heavy dashed line is the synthetic ISED of the SSP component and
686: the case-(ii) BS component.}\label{fig5}
687: \end{figure}
688:
689: \subsection{Modification to the broad-band colours}
690:
691: This series of papers (this paper, XD05, and XDH07) has indicated
692: that BSs in general stellar populations should be recognized and
693: treated with special care when dealing with EPS studies of galaxies.
694:
695: In addition to direct ISED analysis and spectral-index applications,
696: photometric observations using broad-band filters are more frequently
697: used than spectrophotometry when trying to understand stellar
698: populations in distant and therefore often very faint galaxies. As
699: shown in Fig.~\ref{fig5}, the BS component enhances the ISED of the
700: conventional SSP across a wide range of wavelengths, especially in the
701: UV and blue. This tends to make the stellar population appear younger
702: and/or more metal poor according to conventional SSP models, since
703: both modifications can make the colours bluer.
704:
705: Four broad-band colours, $(U-B), (B-V), (V-R)$, and $(V-I)$, are
706: taken as probes in this paper to quantitatively measure the BS
707: contributions for ESO~121--SC03. The colours are obtained by
708: convolving the ISEDs with the corresponding filter response curves.
709: The results are listed in Table~\ref{tab1}. The colours on the first
710: line are from the ISED of the cluster's SSP component, those on the
711: second line are from the synthetic ISED of the SSP and the case-(i)
712: BS component, and those on the third line are from the synthetic
713: ISED of the SSP and the case-(ii) BS component. Based on the results
714: in Table~\ref{tab1}, we conclude that the broad-band colours are all
715: modified, to some extent, with respect to the conventional models
716: (line 1) owing to the presence of BSs. The influence of the BSs
717: becomes more pronounced toward bluer wavelengths. For $(U-B)$, the
718: colour modification can be up to $\sim 50$ per cent. For $(B-V)$ and
719: $(V-R)$, the modifications are $\sim10$ per cent. Even for $(V-I)$,
720: our reddest colour, the changes are still more than 5 per cent.
721: Although these modifications are specific to the case of ESO
722: 121-SC03, i.e., for [Fe/H]~$=-0.97$ dex and an age of 8.9 Gyr, this
723: example serves as a warning for the more general use of broad-band
724: colours to derive cluster ages and metallicities based on
725: theoretical SSP modeling.
726:
727: \begin{table}
728: \centering
729: \begin{minipage}{140mm}
730: \caption{The broad-band colours of the cluster for different
731: cases.}\label{tab1}
732: \begin{tabular}{lcccc}
733: \hline
734: Component & $(U-B)$ & $(B-V)$ & $(V-R)$ & $(V-I)$ \\
735: & (mag) & (mag) & (mag) & (mag) \\
736: \hline
737: SSP & 0.101 & 0.744 & 0.516 & 0.993 \\
738: SSP + BSs $\leq R_{\rm HL}$ & 0.091 & 0.683 & 0.487 & 0.945 \\
739: SSP + all BSs & 0.045 & 0.635 & 0.453 & 0.905 \\
740: \hline
741: \end{tabular}
742: \end{minipage}
743: \end{table}
744:
745: Based on the modifications to the ISED and the broad-band colours
746: calculated as examples in this paper, which are much greater than the
747: observational errors, we strongly caution that conventional SSP models
748: should be used carefully in evolutionary population synthesis of
749: galaxies, at least for the intermediate and old population-I stars. It
750: is difficult to trace the BS population in very young star clusters,
751: and the old population-II star clusters usually have very extended HBs
752: which, instead of the BSs, will dominate the integrated light at blue
753: wavelengths.
754:
755: \section{Underestimates of ages and/or metallicities by the
756: conventional SSP models}
757:
758: In the previous section we showed that the ISED and broad-band colours
759: of conventional SSPs (for the case of ESO~121--SC03) are modified
760: dramatically by BS components, and the modifications will introduce
761: sizeable uncertainties if conventional SSP models are used to obtain
762: the basic physical parameters of a stellar population. Again, taking
763: ESO~121--SC03 as an example, we try to quantify the uncertainties by
764: fitting the synthetic ISED of the cluster using conventional SSP
765: models. We focus on the age and metallicity, the two most important
766: parameters of a stellar population.
767:
768: As presented in Fig.~\ref{fig5}, one of the main effects of BSs is
769: to make the synthetic ISED hotter with respect to that of the
770: conventional models, i.e., the spectra get enhanced more
771: significantly towards shorter wavelengths. Technically, this hotter
772: ISED can still be well fitted (or, say, misunderstood) by
773: conventional SSP models of either younger age or lower metallicity,
774: since both options make the ISED hotter. Therefore, we regard the
775: synthetic ISED of ESO~121--SC03 as the real ISED (i.e., the ISED
776: that will be observed spectroscopically), and we fit it with an ISED
777: of a conventional SSP model. The standard deviation ($\sigma$) is
778: used to identify the best-fitting result. The differences between
779: the real cluster parameters and the best-fitting results are used to
780: discuss the uncertainties that intrinsically exist in the
781: application of the EPS method.
782:
783: We emphasize here that instead of building the actual ISED of a
784: specific cluster, what we are really interested in is to discuss the
785: potential uncertainties introduced when the BS component is
786: completely ignored. Thus, although the synthetic ISED is not exactly
787: the real ISED of the cluster, the method and corresponding results
788: are effective approaches to the problem.
789:
790: \subsection{Uncertainties in age}
791:
792: Fig.~\ref{fig6} shows the uncertainties in age: we fit the synthetic
793: cluster ISED with the conventional SSP models, keeping the
794: metallicity fixed while lowering the age values. The left-hand panel
795: shows the results for the synthetic ISED of the SSP component plus
796: the case-(ii) BS component (all BSs); the right-hand panel is the
797: same but for the SSP component plus the case-(i) BS component (BSs
798: within $R_{\rm HL}$). In the top panels, the solid line is the
799: synthetic cluster ISED, the dotted line is the best-fitting ISED
800: based on conventional SSP models, and the dashed line is the ISED of
801: the SSP component. Plotting the SSP component's ISED in the figure
802: is done to show the differences between the real ISED and the
803: conventional model ISED with the true parameters, and to show that
804: the SSP component's ISED does not correctly represent the
805: observations. The conventional ISEDs are normalized to the real ISED
806: at a wavelength of 5500 {\AA}. In the bottom panels, the residuals
807: between the synthetic ISED and the ISEDs of two conventional SSPs
808: and the $3\sigma$ regions are shown, respectively, using matching
809: line styles.
810:
811: Based on the results shown in Fig.~\ref{fig6} and assuming that the
812: observed ISED is the only data available for stellar population
813: analysis, the fitting with the conventional SSP models will result
814: in a best-fitting age of 3.4 Gyr, which is more than 60 per cent
815: younger than the real age of the cluster, 8.9 Gyr, determined from
816: isochrone fits to the CMD; even considering only the BSs within
817: $R_{\rm HL}$, the best-fitting age of 5.4 Gyr still suffers from an
818: non-negligible difference of $\sim40$ per cent. In the bottom
819: panels, the $\sigma$ values are indeed small for all cases --
820: approximately at the level of only one per cent of the cluster's
821: flux, but the $\sigma$ value according to the best-fitting
822: conventional ISED is always $\sim \frac{2}{3} - \frac{3}{4}$ smaller
823: than that from the SSP component of the cluster alone, which implies
824: that the EPS method will probably lead to an incorrect best-fitting
825: age in unresolved conditions.
826:
827: \subsection{Uncertainties in metallicity}
828:
829: In a similar way as in the previous section, Fig.~\ref{fig7} shows the
830: uncertainties in metallicity in evolutionary population synthesis
831: using the conventional SSP models. The meaning of the symbols and line
832: styles in this figure is similar to that in Fig.~\ref{fig6}. The
833: results are also similar: the metallicities (expressed in [Fe/H]) are
834: significantly underestimated (i.e., more metal poor) as derived from
835: the conventional SSP models. The underestimates will be $\sim60$ and
836: $\sim 30$ per cent, respectively, if we include the cases (ii) and (i)
837: BS components. The detailed results are listed in
838: Table~\ref{tab2}. Taking the modifications in age as an example, the
839: fit uncertainty is calculated as
840: \begin{equation}
841: \Delta_{\rm fit} = \frac{(\rm age)_{\rm real}-(\rm age)_{\rm
842: fit}}{(\rm age)_{\rm real}}.
843: \end{equation}
844:
845: As shown in Figs.~\ref{fig6} and \ref{fig7}, the conventional SSP
846: models can fit perfectly most of the features of the observed ISED,
847: if we leave the parameters free. However, such good fits do not mean
848: that we have obtained good determinations of the real physical
849: parameters in unresolved conditions. Applications of conventional
850: SSP models in EPS studies may therefore seriously suffer from the
851: uncertainties addressed in this paper. This is true at least for the
852: stellar populations corresponding to the star clusters that we have
853: thus far analyzed.
854:
855: \begin{figure}
856: \includegraphics[width=8cm]{fig6.ps}
857: \caption{Fits of the synthetic ISED of ESO~121--SC03 with conventional
858: SSP models~--~uncertainties in the age determination. The left-hand
859: panel is for the synthetic ISED of the SSP component and all BSs. The
860: right-hand panel is for the synthetic ISED of the SSP component and
861: the BSs inside $R_{\rm HL}$. In the top panels, the solid line is the
862: synthetic cluster ISED (real ISED), the dotted line is the
863: best-fitting ISED based on the conventional SSP models, and the dashed
864: line is the ISED of the SSP component. The model ISEDs are normalised
865: to the real ISED at a wavelength of 5500 {\AA}. In the bottom panels,
866: the differences between the synthetic ISED and the ISEDs of two
867: conventional SSPs and the $3\sigma$ regions are shown, respectively,
868: using matching line styles.}\label{fig6}
869: \end{figure}
870:
871: \begin{figure}
872: \includegraphics[width=8cm]{fig7.ps}
873: \caption{Fits of the synthetic ISED of ESO~121--SC03 with conventional
874: SSP models~--~uncertainties in the metallicity determination. Figure
875: coding is as for Fig. \ref{fig6}.}\label{fig7}
876: \end{figure}
877:
878: \begin{table*}
879: \centering
880: \begin{minipage}{140mm}
881: \caption{Fit Uncertainties for ESO~121--SC03}\label{tab2}
882: \begin{tabular}{cccccc}
883: \hline
884: {} & \multicolumn{2}{c}{SSP + All BSs} & &
885: \multicolumn{2}{c}{SSP + BSs ($\leq R_{\rm HL}$}) \\
886: \cline{2-3} \cline{5-6} \\
887: {Real Parameter} & Age = 8.9 Gyr & [Fe/H] $=-0.97$ & {} &
888: Age = 8.9 Gyr & [Fe/H] $=-0.97$\\
889: \hline
890: Model Fit & 3.4 & $-$1.54 & & 5.4 & $-$1.28 \\
891: Uncertainty (per cent) & 62 & 59 & & 40 & 32 \\
892: \hline
893: \end{tabular}
894: \end{minipage}
895: \end{table*}
896:
897: \section{Summary and conclusions}
898: In this paper, the LMC star cluster ESO~121--SC03 is taken as a
899: specific cluster to show in detail the technique and procedure we
900: have developed to assess the potential importance of BSs to the
901: integrated properties of unresolved star clusters: data reduction,
902: isochrone fitting, identification of the BS population, synthetic
903: ISED construction, and the concept of detecting BS contributions to
904: conventional SSP models using integrated light properties.
905:
906: We construct the synthetic ISED of the cluster by decomposing the
907: stellar population of the cluster into two components: the SSP
908: component composed of all the member stars that can be well fitted
909: by an isochrone of single stars; and the BS component defined by the
910: BSs in the cluster. The synthetic ISED of the sum of these two
911: components is regarded as the real ISED of the cluster. The
912: contributions to the ISED of the cluster owing to the BS component
913: are considered for two cases. For case (i), only BSs inside $R_{\rm
914: HL}$ are used, which provides a conservative estimate of the
915: modification caused by the presence of BSs; while for case (ii), all
916: BSs in the cluster are included, which renders the maximum likely
917: alteration to the conventional results. These two cases bracket a
918: range of changes to the conventional SSP models caused by the BS
919: population in a real cluster.
920:
921: Integrated spectral properties are used to measure the modifications
922: to the conventional SSP due to BSs. By the inclusion of BSs, the
923: synthetic ISED of ESO~121--SC03 is greatly enhanced toward shorter
924: wavelengths, and all UV and optical broad-band colours become bluer.
925: Both of these observations can make the population appear a lot
926: younger or more metal poor than would be derived from fitting
927: conventional SSP models. The modifications are quantified by fitting
928: the synthetic ISED with conventional SSP models. For the synthetic
929: cluster ISED including only BSs inside $R_{\rm HL}$, the age
930: underestimate is at the level of $\sim40$ per cent, and the
931: metallicity underestimate is $\sim30$ per cent. For the synthetic
932: ISED that includes all BSs identified in the cluster, the
933: underestimates of both age and metallicity increase to $\sim60$ per
934: cent.
935:
936: In view of the common occurrence of BSs in MC star clusters (e.g.,
937: Johnson et al. 1999; Alcaino et al. 2003), and given that our
938: results for the comparatively rich MC clusters (at least, with
939: respect to the Galactic OCs) are statistically robust, we have
940: reason to believe that the BS effect discussed in this paper for
941: ESO~121--SC03 may not be an exception. BS contributions (to the
942: conventional SSPs) measured based on other MC star clusters are
943: expected to be at a similar level as for ESO~121--SC03, and likely
944: follow the same trend as for our previous results based on Galactic
945: OCs covering a wide age range (Deng et al. 1999; XD05; XDH07): the
946: clusters' age and/or metallicity estimates will be underestimated
947: significantly, by a factor of 2--4, based on fits of SSPs to either
948: spectra or broad-band colours. In a subsequent paper, we will
949: discuss the equivalent results for the entire MC star cluster
950: sample, i.e., based on the application of the method discussed here,
951: which will significantly extend the parameter coverage and the
952: robustness of this conclusion.
953:
954: In order to constrain the ISEDs of real stellar populations in a
955: reliable way and to provide the community with a set of modified SSP
956: models, extending our work on cluster samples to the MC rich star
957: clusters is now in progress. In the context of our ultimate goal, we
958: are currently still at the stage of information collection. To
959: correct the conventional SSP models empirically, a sufficient
960: working sample of clusters covering large ranges of age and
961: metallicity, and of different dynamical environments, is definitely
962: needed. Such a sample will provide statistical results of the BS
963: contributions, and reliable information of BS populations in SSPs,
964: such as the specific frequency of BSs (BS numbers in an SSP), and
965: luminosity and colour functions of BSs (BS positions in a CMD).
966: These will, we expect, eventually lead to the final solution to this
967: problem, when we can construct the BS-corrected SSP models which can
968: then be used in EPS studies in unresolved conditions.
969:
970: \section*{Acknowledgments}
971: YX, LD and ZH would like to thank the Chinese National Science
972: Foundation for support through grants 10573022, 10778719, 10333060,
973: 10521001, 10433030, and the Ministry of Science and Technology of
974: China through grant 2007CB815406. YX acknowledges financial support
975: from the Royal Society in the form of a ``Sino-British Fellowship
976: Trust Award''. YX is also grateful to the International Astronomical
977: Union for travel support from Beijing to the UK. RdG acknowledges
978: partial funding from the Royal Society through an International
979: Joint Project grant, as well as support for a high-level UK-China
980: Science Network, funded by the Office of Science and Innovation of
981: the UK government. ADM is supported by a Marie Curie Excellence
982: Grant from the European Commission under contract
983: MCEXT-CT-2005-025869.
984:
985: This paper is based on observations made with the NASA/ESA
986: \textit{Hubble Space Telescope}, obtained at the Space Telescope
987: Science Institute, which is operated by the Association of
988: Universities for Research in Astronomy, Inc., under NASA contract NAS
989: 5-26555. These observations are associated with programme \#9891.
990:
991: \begin{thebibliography}{99}
992: \bibitem[\protect\citeauthoryear{Ahumada \& Lapasset}{1995}]{b}
993: Ahumada J.A., Lapasset E., 1995, A\&AS, 109, 375
994: \bibitem[\protect\citeauthoryear{Ahumada \& Lapasset}{2007}]{b}
995: Ahumada J.A., Lapasset E., 2007, A\&A, 463, 789
996: \bibitem[\protect\citeauthoryear{Alcaino et al.}{2003}]{b} Alcaino G.,
997: Alvarado F., Borissova J., Kurtev R., 2003, A\&A, 400, 917
998: \bibitem[\protect\citeauthoryear{Bekki et al.}{2004}]{b}
999: Bekki K., Beasley M.A., Forbes D.A., Couch W.J., 2004, ApJ, 602, 730
1000: \bibitem[\protect\citeauthoryear{Benz \& Hills}{1987}]{b} Benz W.,
1001: Hills J.G., 1987, ApJ, 323, 614
1002: \bibitem[\protect\citeauthoryear{Benz \& Hills}{1992}]{b} Benz W.,
1003: Hills J.G., 1992, ApJ, 389, 546
1004: \bibitem[\protect\citeauthoryear{Bertelli et al.}{1994}]{b} Bertelli
1005: G., Bressan A., Chiosi C., Fagotto F., Nasi E., 1994, A\&AS, 106,
1006: 275
1007: \bibitem[\protect\citeauthoryear{Bica et al.}{1996}]{b} Bica E.,
1008: Clari\'{a} J.J., Dottori H., Santos Jr.J.F.C., Piatti A.E., 1996,
1009: ApJS, 102, 57
1010: \bibitem[\protect\citeauthoryear{Bica et al.}{1998}]{b} Bica E.,
1011: Geisler D., Dottori H., Clari\'{a} J.J., Piatti A.E., Santos
1012: Jr.J.F.C., 1998, AJ, 116, 723
1013: \bibitem[\protect\citeauthoryear{Bruzual \& Charlot}{2003}]{b}
1014: Bruzual A. G., Charlot S., 2003, MNRAS, 344, 1000 (BC03)
1015: \bibitem[\protect\citeauthoryear{Burstein \& Heiles}{1982}]{b}
1016: Burstein D., Heiles C., 1982, AJ, 87, 1165
1017: \bibitem[\protect\citeauthoryear{Carney et al.}{2005}]{b} Carney B.W.,
1018: Latham D.W., Laird J.B., 2005, AJ, 129, 466
1019: \bibitem[\protect\citeauthoryear{Da~Costa}{1991}]{b}
1020: Da~Costa G.S., 1991, in IAU Symp. 148, The Magellanic Clouds, Eds.
1021: Haynes R., Milne D., (Dordrecht: Kluwer), p183
1022: \bibitem[\protect\citeauthoryear{de Grijs \& Anders}{2006}]{b} de~Grijs
1023: R., Anders P., 2006, MNRAS, 366, 295
1024: \bibitem[\protect\citeauthoryear{Deng et al.}{1999}]{b} Deng L.,
1025: Chen R., Liu X.S., Chen J.S., 1999, ApJ, 524, 824
1026: \bibitem[\protect\citeauthoryear{Dirsch et al.}{2000}]{b} Dirsch B.,
1027: Richtler T., Gieren W.P., Hilker M., 2000, A\&A, 360, 133
1028: \bibitem[\protect\citeauthoryear{Dolphin}{2000}]{b} Dolphin A.E.,
1029: 2000, PASP, 112, 1383
1030: \bibitem[\protect\citeauthoryear{Ferraro et al.}{2004}]{b} Ferraro
1031: F.R., Beccari G., Rood R.T., Bellazzini M., Sills A., Sabbi E.,
1032: 2004, ApJ, 603, 127
1033: \bibitem[\protect\citeauthoryear{Ferraro et al.}{1997}]{b} Ferraro
1034: F.R., Paltrinieri B., Fusi~Pecci F., Cacciari C., Dorman B., Rood
1035: R.T., Buonanno R., Corsi C.E., Burgarella D., Laget M., 1997, A\&A,
1036: 324, 915
1037: \bibitem[\protect\citeauthoryear{Ferraro et al.}{2003}]{b} Ferraro
1038: F.R., Sills A., Rood R.T., Paltrinieri B., Buonanno R., 2003, ApJ,
1039: 588, 464
1040: \bibitem[\protect\citeauthoryear{Geisler et al.}{1997}]{b}
1041: Geisler D., Bica E., Dottori H., Clari J.J., Piatti A.E., Santos
1042: Jr.J.F.C., 1997, AJ, 114, 1920
1043: \bibitem[\protect\citeauthoryear{Girardi et al.}{2000}]{b} Girardi
1044: L., Bressan A., Bertelli G., Chiosi C., 2000, A\&AS, 141, 371
1045: \bibitem[\protect\citeauthoryear{Gratton et al.}{2003}]{b} Gratton
1046: R.G., Bragaglia A., Carretta E., Clementini G., Desidera S.,
1047: Grundahl F., Lucatello S., 2003, A\&A, 408, 529
1048: \bibitem[\protect\citeauthoryear{Johnson et al.}{1999}]{b} Johnson J.A.,
1049: Bolte M., Stetson P.B., Hesser J.E., Somerville R.S., 1999, ApJ,
1050: 527, 199
1051: \bibitem[\protect\citeauthoryear{Lanzoni et al.}{2007}]{b} Lanzoni
1052: B., Dalessandro E., Ferraro F.R., Mancini C., Beccari G., Rood R.T.,
1053: Mapelli M., Sigurdsson S., 2007, ApJ, 663, 267
1054: \bibitem[\protect\citeauthoryear{Lee et al.}{2003}]{b} Lee M.G., et
1055: al., 2003, AJ, 126, 2840
1056: \bibitem[\protect\citeauthoryear{Lejenue et al.}{1997}]{b} Lejeune
1057: Th., Cuisinier F., Buser R., 1997, A\&AS, 125, 229
1058: \bibitem[\protect\citeauthoryear{Liu et al.}{2008}]{b} Liu G., Deng
1059: L., Chavez M.D., et al., 2008, in preparation
1060: \bibitem[\protect\citeauthoryear{Lombardi et al.}{1995}]{b} Lombardi
1061: J.C.Jr., Rasio F.A., Shapiro S.L., 1995, ApJ, 445L, 117
1062: \bibitem[\protect\citeauthoryear{Lombardi et al.}{2002}]{b} Lombardi
1063: J.C.Jr., Warren J.S., Rasio F.A., Sills A., Warren A.R., 2002, ApJ,
1064: 568, 939
1065: \bibitem[\protect\citeauthoryear{Mackey \& Broby~Nielsen}{2007}]{b}
1066: Mackey A.D., Broby~Nielsen P., 2007, MNRAS, 379, 151
1067: \bibitem[\protect\citeauthoryear{Mackey \& Gilmore}{2004}]{b} Mackey
1068: A.D., Gilmore G.F., 2004, MNRAS, 352, 153
1069: \bibitem[\protect\citeauthoryear{Mackey, Payne \& Gilmore}{2006}]{b} Mackey
1070: A.D., Payne M.J., Gilmore G.F., 2006, MNRAS, 369, 921 (MPG06)
1071: \bibitem[\protect\citeauthoryear{Mackey \& Gilmore}{2003a}]{b}
1072: Mackey A.D., Gilmore G.F., 2003a, MNRAS, 338, 85
1073: \bibitem[\protect\citeauthoryear{Mackey \& Gilmore}{2003b}]{b}
1074: Mackey A.D., Gilmore G.F., 2003b, MNRAS, 338, 120
1075: \bibitem[\protect\citeauthoryear{Mapelli et al.}{2007}]{b} Mapelli M.,
1076: Ripamonti E., Tolstoy E., Sigurdsson S., Irwin M.J., Battaglia G.,
1077: 2007, MNRAS, 380, 1127
1078: \bibitem[\protect\citeauthoryear{Mateo et al.}{1986}]{b} Mateo M.,
1079: Hodge P., Schommer R.A., 1986, ApJ, 311, 113
1080: \bibitem[\protect\citeauthoryear{Mathys}{1991}]{b} Mathys G., 1991,
1081: A\&A, 245, 467
1082: \bibitem[\protect\citeauthoryear{Parmentier \& de~Grijs}{2007}]{b}
1083: Parmentier G., de~Grijs R., 2007, MNRAS, accepted (arXiv:0710.3477)
1084: \bibitem[\protect\citeauthoryear{Piotto et al.}{2004}]{b} Piotto G., De
1085: Angeli F., King I.R., Djorgovski S.G., Bono G., Cassisi S., Meylan
1086: G., Recio-Blanco A., Rich R.M., Davies M.B., 2004, ApJ, 604, L109
1087: \bibitem[\protect\citeauthoryear{Portegies~Zwart et al.}{1997b}]{b}
1088: Protegies~Zwart S.F., Hut P., McMillan S.L.W., Verbunt F., 1997b,
1089: A\&A, 328, 143
1090: \bibitem[\protect\citeauthoryear{Portegies~Zwart et al.}{1997a}]{b}
1091: Protegies~Zwart S.F., Hut P., Verbunt F., 1997a, A\&A, 328, 130
1092: \bibitem[\protect\citeauthoryear{Sabbi et al.}{2004}]{b} Sabbi E.,
1093: Ferraro F.R., Sills A., Rood R.T., 2004, ApJ, 617, 1296
1094: \bibitem[\protect\citeauthoryear{Salpeter}{1955}]{b} Salpeter E.E.,
1095: 1955, ApJ, 121, 161
1096: \bibitem[\protect\citeauthoryear{Sarajedini}{1994}]{b} Sarajedini
1097: A., 1994, AJ, 107, 618
1098: \bibitem[\protect\citeauthoryear{Shetrone \& Sandquist}{2000}]{b}
1099: Shetrone M.D., Sandquist E.L., 2000, AJ, 120, 1913
1100: \bibitem[\protect\citeauthoryear{Sills et al.}{2005}]{b} Sills A., Adams
1101: T., Davies M.B., 2005, MNRAS, 358, 716
1102: \bibitem[\protect\citeauthoryear{Sills et al.}{2000}]{b} Sills A.,
1103: Pinsonneault M.H., Terndrup D.M., 2000, ApJ, 534, 335
1104: \bibitem[\protect\citeauthoryear{Sirianni et al.}{2005}]{b} Sirianni
1105: M., et al., 2005, PASP, 117, 1049
1106: \bibitem[\protect\citeauthoryear{Tian et al.}{2006}]{b} Tian B., Deng L.,
1107: Han Z.W., Zhang X.B., 2006, A\&A, 455, 247
1108: \bibitem[\protect\citeauthoryear{Warren et al.}{2006}]{b} Warren
1109: S.R., Sandquist E.L., Bolte M., 2006, ApJ, 648, 1026
1110: \bibitem[\protect\citeauthoryear{Xin \& Deng}{2005}]{b} Xin Y., Deng
1111: L., 2005, ApJ, 619, 824 (XD05)
1112: \bibitem[\protect\citeauthoryear{Xin, Deng \& Han}{2007}]{b} Xin Y., Deng L., Han
1113: Z.W., 2007, ApJ, 660, 319 (XDH07)
1114:
1115: \end{thebibliography}
1116:
1117: \bsp
1118:
1119: \label{lastpage}
1120:
1121: \end{document}
1122: