1: %----------------------------------------------------------------------- % %
2: %Paper entitled: ``mass loss in M3'' revised version 24/07/07
3: %
4: %Local definitions by the authors
5: \def\Lsun{L$_\odot$}
6: \def\Msun{M$_\odot$}
7: \def\msun{M$_\odot$}
8: \def\Mprog{M$_{prog}$}
9: \def\Mprogl{M$_{prog1}$}
10: \def\Mprogu{M$_{prog2}$}
11: \def\Mtrans{M$_{trans}$}
12: \def\mtrans{M$_{trans}$}
13: \def\Mtransec{M$_{trans2}$}
14: \def\Mv{M$_{\rm v}$}
15: \def\Teff{T$_{\rm eff}$}
16: \def\teff{T$_{\rm eff}$}
17: \def\0BMV{(B--V)$_{\rm 0}$}
18: \def\BMV{B--V}
19: \def\Z{Z}
20: \def\Y{Y}
21: \def\V{V}
22: \def\B{B}
23: \def\Feh{[Fe/H]}
24: \def\vtot{$V_{\rm tot}$}
25: \def\Zeq{$Z_{\rm eq}$}
26: \def\Zsun{$Z_\odot$}
27: %\def\roc{Log($\rho_{\rm c})}
28: \def\Dm{$\Delta_{rm m}$}
29: \def\DM{$\Delta_{rm M}$}
30: \def\Dem{$\delta_{rm m}$}
31: \def\simgt{\lower.5ex\hbox{$\; \buildrel > \over \sim \;$}}
32: \def\simlt{\lower.5ex\hbox{$\; \buildrel < \over \sim \;$}}
33: \def\ocen{$\omega$~Cen}
34: %-----------------------------------------------------------------------
35: %\documentclass[12pt,preprint]{aastex}
36: \documentclass[preprint2]{aastex}
37: %\usepackage{graphics}
38: %\usepackage{aastexug}
39: %\usepackage{natbib}
40: %%%%\documentstyle{l-aa}
41: %
42: \begin{document}
43: %\input /home/power/tex/psfig.sty
44: % \thesaurus{05(08.08.2; %Stars: horizontal branch
45: % 10.07.2; %Globular Clusters: general
46: % 08.13.2)} % mass loss
47: %
48: \title{Is mass loss along the red giant branch of globular clusters
49: sharply peaked? The case of M3}
50:
51: %
52: \author{Vittoria Caloi\altaffilmark{1} \& Francesca D'Antona\altaffilmark{2} }
53: %
54: %
55: %\offprints{} \mail{}
56: \affil{\altaffilmark{1}
57: INAF - IASF Roma, Via Fosso del Cavaliere, I-00133 Roma, Italy;
58: vittoria.caloi@
59: iasf-roma.inaf.it
60: \affil{\altaffilmark{2}INAF - Osservatorio Astronomico di Roma, via Frascati
61: 33, 00040 Monteporzio, Italy; dantona@mporzio.astro.it}
62: }
63:
64: %\date{Received ; accepted }
65:
66: \begin{abstract}
67: There is a growing evidence that several globular clusters must contain
68: multiple stellar generations, differing in helium content. This hypothesis has
69: helped to interpret peculiar unexplained features in their horizontal branches.
70: In this framework we model the peaked distribution of the RR Lyr periods in
71: M3, that has defied explanation until now. At the same time, we try to
72: reproduce the colour distribution of M3 horizontal branch stars. We find that
73: only a very small dispersion in mass loss along the red giant branch
74: reproduces with good accuracy the observational data. The enhanced and
75: variable helium content among cluster stars is at the origin of the extension
76: in colour of the horizontal branch, while the sharply peaked mass loss is
77: necessary to reproduce the sharply peaked period distribution of RR Lyr
78: variables. The dispersion in mass loss has to be $\le$ 0.003 \Msun, to be
79: compared with the usually assumed values of $\sim$ 0.02 \Msun. This requirement
80: represents a substantial change in the interpretation of the physical
81: mechanisms regulating the evolution of globular cluster stars.
82: \end{abstract}
83:
84: \keywords{globular clusters: general ---
85: globular clusters: individual NGC 5272 ---
86: Stars:evolution --- stars:horizontal branch --- stars:mass loss}
87: %\authorrunning{Caloi and D'Antona}
88: %\titlerunning{M3, a new perspective}
89:
90: \section{Introduction}
91: Since the early '70s the population distribution along the
92: horizontal branch (HB) in globular clusters (GCs) has been interpreted
93: in terms of a variation of the mass of the hydrogen--rich envelope on
94: top of the helium core, left over by the red giant evolution. The star
95: to star difference in the amount of the hydrogen--rich material was
96: explained as due to the stochastic nature of the mass loss process
97: \citep{rood1973}. The
98: main feature to be reproduced was the distribution in {\it number}
99: of HB members among the red, variable and blue regions. This has been
100: possible for a large part of the GC system, assuming an average mass
101: loss along the red giant (RG) branch evolution of $\Delta$M $\sim$
102: 0.22 \Msun, with a $\sigma \sim$ 0.02 \Msun\ \citep[see, f.e.][]{lee1994}.
103:
104: On the other hand, it has always been clear that a not negligible number
105: of HBs could not be understood following this assumption. A special case
106: was that of the exclusively blue HBs (a consistent percentage of the
107: total GC number), for which an increase in mass loss, in age, in RG
108: rotation etc., has been invoked, generally with not great success
109: \citep{fp1993}. But
110: there are HB distributions that defy any simple explanation: bimodal and
111: trimodal distributions, extremely long blue tails, hot blue stars in
112: metal rich clusters are the most impressive examples \citep[e.g.][]{rich1997,
113: sosin1997,ferraro1998}.
114:
115: %\clearpage
116: \begin{figure}[tb]
117: \centering
118: \includegraphics[width=8cm]{f1.eps}
119: \caption{Period histogram (period in days) for M3 RR Lyrs from the data
120: by Corwin \& Carney
121: 2001, where RRc have been fundamentalized (continuous red line);
122: dotted green line: RRabs, dashed blue line: fundamentalized RRcs
123: }
124: \label{fig3}
125: \end{figure}
126:
127: \begin{figure}[tb]
128: \centering
129: \includegraphics[width=8cm]{f2.eps}
130: \caption{CM diagram of not-variable HB members M3 from the data
131: by Corwin \& Carney
132: 2001 (not a complete sample, see text), with
133: the histogram of their distribution with respect
134: to \BMV. Dashed (red) line: red HB members, dashed-dotted (blue) line:
135: blue HB members; continuous (green) line: RR Lyrs histogram}
136: \label{fig1}
137: \end{figure}
138: %\clearpage
139:
140: Recently, the presence of peculiar patterns in the chemical abundances
141: in many (if not all) GCs has provoked a renewed interest in the
142: hypothesis of helium variations among globular cluster stars
143: \citep{norris1981,jb1998,dantona2002}.
144: The presence of more than one stellar generations, with differing helium
145: contents, can explain features such as the multiple main sequences in
146: $\omega$ Cen and NGC 2808 \citep{bedin2004,norris2004,dantona2005,lee2005,piotto2005,
147: moehler2006,piotto2007},
148: the appearance of blue HB
149: stars in metal rich GCs (e.g., NGC 6441, Caloi \& D'Antona 2007),
150: as well as of very blue and very faint HB members that are not
151: understood in terms of ordinary evolutionary patterns \citep{sweigart2001}.
152:
153: So a new physical factor, the helium content, has begun to show a
154: relevance comparable to that of mass loss in shaping HB morphology and
155: population. At the same time, old unresolved riddles on HB star
156: distribution, related in particular to some properties of the RR Lyr
157: variables, have been recalled to attention. The attempts to solve these
158: problems are imposing unexpected conditions on mass loss properties and
159: on the role of varying helium content, so that we are possibly facing a
160: rather substantial change of perspective in the interpretation both of
161: the HB population and of the mass loss phenomenon during the RG ascent.
162: This paper intends to expose this new point of view and
163: some of its consequences.
164:
165: \section{M3, the horizontal branch and the RR Lyrae variables}
166:
167: The problem of the peaked distribution (see Fig. 1) of the RR Lyr
168: periods in M3 \citep{ct1981, rc1989} has been revisited by
169: \cite{catelan2004} and \cite{ccc2005}. Catelan reached the conclusion that
170: this feature cannot be understood in the framework of canonical HB
171: evolution, while Castellani et al. found an interesting way to obtain
172: the observed period distribution with current models. Their solution
173: requires a strong constraint on mass loss, that has to have a very small
174: dispersion ($\sigma$ $\sim$ 0.005 \Msun) compared to normally accepted
175: values ($\sigma$ $\sim$ 0.02 \Msun). This occurrence allows the models
176: to populate the RR Lyrae region just at the turning point of their
177: blueward loops, with little dispersion, maximizing their permanence in
178: the \teff\ interval required to provide the peak in the period
179: distribution. Anyway, these authors found two main difficulties with
180: this modelization: i) a number of red HB stars noticeably larger than
181: observed, and ii) the blue HB stars must be simulated by an {\it ad hoc}
182: population, with a mean mass, and a $\sigma$\ different from those
183: required to explain the RR Lyr period distribution.
184:
185: %\clearpage
186: \begin{figure}[tb]
187: \centering
188: \includegraphics[width=8cm]{f3.eps}
189: \caption{CM diagram of M3 from the data PH94 and CCD96
190: for the non variable stars (crosses, blue) plus the RR Lyrs
191: by Corwin \& Carney (triangles, green).
192: Also shown is the histogram of their distribution with respect
193: to \BMV (continuous line, red); dashed line (blue): non variable
194: stars, dotted (green) line: RR Lyr variables}
195: \label{fig2}
196: \end{figure}
197: %\clearpage
198:
199: In this paper we give a model for the period distribution of RR Lyrae
200: variables in M3, following the suggestion of a very small mass
201: dispersion for the mass loss along the RG branch. The extension in
202: colour of the HB is then reproduced by assuming that the GC contains a
203: second stellar generation with variable helium content, enhanced with
204: respect to the first generation. For this cluster, while explaining in
205: detail the HB morphology and the peaked period distribution of its RR
206: Lyr, the scenario of multiple star generations provides a strong support
207: to the rather surprising condition that mass loss must be sharply
208: peaked.
209:
210: \section{The observational sample}
211:
212: At present we have periods, average magnitudes and colours for almost
213: all the RR Lyrae variables in M3 \citep{benko2006, cc2001}. As
214: guide to the details of M3 HB, we took the photometry by
215: \cite{cc2001}, who give \V, \BMV\ for about 170 RR Lyr variables (and
216: periods for 201), together with \B, \V\ photometry for blue and red HB
217: members. The sample (and the corresponding histograms) shown in
218: Fig. \ref{fig1} is not complete since it shows only the stars with the
219: best photometry (see Fig. 4 in Corwin \& Carney), and we have limited the
220: number of RR Lyrae variables to 100, roughly as expected from HB star
221: counts in complete samples (see later). The importance of this sample is
222: that it guarantees a uniform photometry all over the HB, and this is
223: crucial to estimate the change in number density with colour when
224: passing from the red to the variable and to the blue regions. From
225: Fig. \ref{fig1} we see that the number density grows passing from the
226: red HB to the RR Lyrae region; then there is a dip in the population
227: between variables and blue HB, followed by another peak at \BMV\ $\sim$
228: 0.0 -- 0.1. The dip in the colour distribution reproduces the paucity of
229: stars in the RRc colour interval (specially at lower luminosity) and at
230: the border of the blue region, clearly seen in Fig. 5 in Corwin \&
231: Carney paper.
232: %ref 1
233: It is important to clarify that this dip is not affected by the well known
234: uncertainties in the definition of the "mean"colours for RR Lyr variables
235: \citep[see, f.e.][]{bono1995}. In Fig. 2 we use
236: magnitude average colours (B-V)$_{\rm mag}$ \citep{preston1961, sandage1990},
237: generally accepted as the best approximation to the colour of
238: the "static star" with the same average energy output. But any other common
239: procedure would give the same result for RR Lyrs of type c, owing to their
240: symmetric light curves \citep{bono1995}, while possible differences up to
241: 0.05 mag may be found for highly asymmetric light curves (as apply to RR Lyrs
242: of type a). So the relative positions of RRc variables and constant blue HB
243: stars can be considered as well established.
244: %ref 1
245:
246: So any model for the star distribution along the HB must
247: reproduce these two main features: the sharp peak in colour of the RR
248: Lyraes (corresponding to the sharp peak in periods), and the dip in the
249: population at the blue of the variables.
250: Unfortunately, the \cite{cc2001} HB sample is not complete. We then
251: examined the complete samples from the CCD photometry by
252: \cite{ferraro1997} (CCD96) of the external cluster regions, and the
253: photographic sample PH94 from \cite{buonanno1994}. The colours of the
254: photographic sample turned out to be very similar, in the red and blue
255: HB sections, to those in \cite{cc2001}, so that this sample guarantees a
256: smooth continuity with the colours of the RR Lyrae. To obtain a larger
257: complete sample, we include also the CCD photometric data by
258: \cite{ferraro1997}, but we ``corrected back'' this photometry the PH94
259: photometry by using Equation 7 of the Ferraro et al. paper This
260: correction is certainly not adequate for the bluest HB stars, which were
261: not well measured on the photographic plates, but this is not an
262: important point for our analysis, which is focused on the ``horizontal''
263: part of the HB. In Fig. \ref{fig2} we show this complete sample (PH94,
264: plus CCD96, plus the variables from \cite{cc2001} with the related
265: population histogram. The histogram for the variables is normalized to
266: the number in the PH94 plus CCD96 samples as follows.
267: The numbers we obtain for the red, variable and blue HB regions are 132,
268: 222 and 217, respectively. This balance among the various populations is
269: intermediate between what found for inner and outer regions by
270: \cite{catelan2001}. In fact, they found a substantial difference between
271: the population distributions for distances from the cluster center r $<$
272: 50'' (28,65,81) and for r $>$ 120'' (27,46,30). Given the uncertatinties
273: involved, we shall consider our numbers as indicative of a substantial
274: similarity between the variable and blue populations, while the red
275: region is noticeably less populated than the variable one. As
276: already pointed out by \cite{ccc2005}, and as we shall see in the
277: following, this latter point represents one of the most difficult
278: constraints with which the simulations have to comply.
279:
280: Besides, we have to reproduce the RR Lyr period distribution and their
281: mean period. The sharp peak in the period histogram (Fig. \ref{fig3})
282: descends from the colour distribution of the variables (Figs. \ref{fig1}
283: and \ref{fig2}); the mean period $\rm P_{\rm f}$ obtained
284: fundamentalizing the RRc variables (addition of 0.128 to the logarithm
285: of their periods, van Albada \& Baker 1973) varies slightly with the
286: chosen sample, and we assume $\rm P_{\rm f}$ to be comprised between
287: 0.53 and 0.54 d. The sample by Corwin \& Carney provides $\rm P_{\rm
288: f}$ = 0.537 d.
289:
290:
291: \section{The simulations}
292:
293: \subsection{The models}
294: We assume for the stars in M3 a heavy element content \Z=0.001 and use
295: our HB models descending from main sequence structures with $\rm Y_{\rm
296: MS}$ =0.24, 0.28 and 0.32, as fully described in
297: \cite{dantonacaloi2004}. These models are the basis of the syntetic HBs
298: computed in the following way. We fix a cluster age, and compute the
299: evolving RG mass according to the relation (1) from
300: \cite{dantonacaloi2004}, which is a function of both age and Y. We
301: assume that the stellar content of the cluster is divided into two main
302: populations: a fraction of stars with cosmological helium (first
303: generation), for which we assume Y=0.24, and the remaining fraction
304: having larger and variable helium content (second generation). The
305: number of the first generation stars, and the number vs. helium
306: distribution of the second generation stars is changed in order to fit
307: the whole HB. The HB mass is obtained by assuming that all RG stars lose
308: an amount of mass $\delta$M, with a gaussian dispersion $\sigma$\ around
309: a fixed value $\delta$M$_0$. Then the HB mass varies both due to the
310: mass dispersion $\sigma$\ around the average assumed mass loss, and due
311: to the dependence of the RG mass on the helium content. As the evolving
312: mass {\it decreases} with increasing helium content, the stars with
313: higher helium will populate bluer regions of the HB.
314: In previous work \citep{dantonacaloi2004} we increased slightly the mass loss
315: when increasing the helium content, following the idea that the global mass loss
316: inversely depends on the stellar gravity, and so smaller masses would lose more mass
317: \citep{lee1994}. Here we assume that the average mass lost is the same for all the
318: helium contents. In fact, full evolutionary tracks including explicit
319: consideration of mass loss, computed according to Reimers' formulation,
320: showed us that the total mass loss along the RG branch is independent of the
321: helium content (at each given age): the track location and gravity compensate
322: in such a way that the mass lost is constant at the level of 1-2$\times 10^{-3}$\Msun,
323: in most cases.
324:
325: We assume a parametric mean mass loss along the RG branch with gaussian dispersion
326: $\sigma$, and extract random both the mass loss and the HB age in
327: the interval from 10$^6$yr to 10$^8$yr, according to the chosen Y
328: distribution. We thus locate the luminosity and \Teff\ along
329: the evolution of the HB mass obtained. These values are transformed into
330: the observational plane M$_{\rm v}$ vs. \BMV\ using \cite{bessell-castelli-plez1998}.
331: We identify the variable stars as belonging to a
332: fixed \Teff\ interval and compute their period according to the
333: pulsation equation (1) by \cite{dmc2004}. The results are very similar
334: if we adopt the classic \cite{va1973} relation. The real problem is
335: given by the choice of the exact boundaries of the RR Lyr strip, that affect
336: strongly the number and mean period of the RR Lyrae variables.
337:
338: We assumed as width of the RR Lyr strip $\Delta$log$T_{\rm eff}$ = 0.08,
339: a value commonly adopted (see, e.g., Castellani et al. 2005), and made a
340: few experiments varying the choice of the red limit. We found that such
341: limit {\it had} to be taken at log($T_{\rm eff}$) = 3.80, since for
342: lower values a much larger than observed number of long period RRab's (P
343: $>$ 0.6 d) was present.
344: Lower limits for the red edge temperature can be
345: found in the literature (see the already quoted Catelan 2004 and
346: Castellani et al. 2005). In order to avoid the presence of too many long
347: period variables, Castellani et al. (2005) make use of decreased values
348: of their model luminosities. We prefer to keep the theoretical
349: values for the luminosity and adopt the correspondingly suitable value
350: for the red edge.
351:
352: One has also to consider that the estimate of the limiting temperatures
353: is by no means a settled question. As discussed, e.g., by
354: \cite{dmc2004}, the theoretical red edge of the pulsation region gets
355: hotter by $\sim$ 300 K if the efficiency of superadiabatic convection in the star
356: external layers is increased by changing the ratio
357: mixing length to pressure scale height ($l/H_{\rm p}$) from 1.5 to 2.
358: The comparison with observed periods shows that a
359: larger convection efficiency should be preferred in several GCs (see
360: Fig. 14 in Di Criscienzo et al. paper). At the same time, recent
361: observational data also suggest hot values for the instability strip red
362: boundary. Corwin et al. (1999) found for the colours of the RR Lyr
363: variables in NGC 5466 (a very metal poor cluster) the limits 0.15 $<$
364: \0BMV\ $<$ 0.38; similarly, Wehlau et al. (1999) found for NGC 7006, a
365: GC with a Z close to that of M3, the limits 0.14 $<$ \0BMV\
366: $<$0.38. Both these colour intervals correspond to \Teff\ intervals very
367: close to our choice. Besides, Silbermann \& Smith (1995), in their
368: discussion of the blue and red edges in M15, say that the red edge may
369: be near 6300 K.
370:
371: We intend to reproduce the star distribution along the HB in some
372: detail, not simply to obtain the overall distribution among red (R),
373: variable (V) and blue (B) regions. To this purpose, the parameters we
374: can play with are: i) the number of stars with cosmological Y (first
375: stellar generation); ii) the number of stars with enhanced Y and their
376: distribution as function of Y; iii) the mass loss and mass loss
377: dispersion. The total number of stars involved is 571 (see above), to be
378: distributed among helium contents from the cosmological one (Y = 0.24)
379: up.
380:
381: \subsection{Results}
382:
383: On the basis of Fig. \ref{fig1} and Fig. \ref{fig2}, it appeared
384: reasonable to assume that the red HB and variable regions are populated
385: mainly by stars with cosmological Y content, while blue HB members
386: derive from smaller mass RG branch stars with a higher Y
387: content, starting slightly above the cosmological one.
388: evolving RG mass.
389:
390: Once fixed the age and the heavy element content, for
391: a given value of Y, the star distribution on the HB is decided by the
392: amount of the mass loss on the RG branch. After few trials we could
393: check that the only way to obtain peaked distributions of the variables
394: in colour and period was to adopt a very well defined value
395: for the mass loss. For an age of 11
396: Gyr and a population with initial Y content = 0.24, a peaked
397: distribution in (\BMV) (and period, see later) was obtained with a sharp
398: mass loss value of $\Delta$(M) = 0.203 \Msun, $\sigma$(M) $\leq$
399: 0.003 \Msun. So the mass loss parameter turns out to be the most
400: delicate and the less arbitrary among the ones mentioned before. We
401: shall come back to this point.
402:
403: In Fig. \ref{fig4} we show a simulation of the HB distribution obtained
404: with the quoted parameters, compared
405: with the observed one. Some points are worth of attention.
406: The simulation reproduces the main population peak in the RR Lyr region
407: and the smaller one in the red section. It reproduces as well the dip at the
408: blue of the RR Lyraes and the peak at \BMV\ $\sim$ 0.0. The distribution
409: in number along the HB gives 159 red members, 211 variable and 201 blue members,
410: to be compared with the expected 132, 222 and 217 (see Sec. 3.1). Except
411: for a slight excess of red stars (of the order of two $\sigma$), the
412: similarity between the observed and the simulated distribution is
413: remarkable. We stress that this is the first time that a {\it detailed}
414: simulation has been attempted, since we tried to reproduce not simply
415: the number of stars in the blue, variable and red HB regions, but the
416: entire {\it colour histogram} of the HB population.
417:
418: %\clearpage
419: \begin{figure}[tb]
420: \centering
421: \includegraphics[width=8cm]{f4.eps}
422: \caption{Simulation of the HB distribution (continuous magenta
423: line) compared with
424: the observed one (dashed green line, see Fig. \ref{fig2}); the
425: dotted red line indicates the
426: contribution by members with Y=0.24, the dashed-dotted blue line
427: the contribution by members with
428: enhanced Y, according to the distribution in Y shown in the
429: lower panel. In white, the central track of the simulations.}
430: \label{fig4}
431: \end{figure}
432:
433: \begin{figure*}[tb]
434: \centering
435: \includegraphics[width=8cm]{f5a.eps}
436: \includegraphics[width=8cm]{f5b.eps}
437: \caption{Simulated (continuous red line) and observed (dotted
438: magenta line) period
439: histograms; the contributions by members with Y=0.24 is indicated
440: by the green dashed line, by members with enhanced helium by the blue
441: dashed-dotted line. The number of
442: ``theoretical'' RR Lyraes is reported. The left panel shows the
443: simulation for $\sigma=0.0015$\Msun, while the right panel shows the
444: simulation for $\sigma=0.005$\Msun. All other parameters are the same.}
445: \label{fig5}
446: \end{figure*}
447: %\clearpage
448:
449: The left panel of Figure \ref{fig5} shows the period histogram relative
450: to the same simulation as in Fig. \ref{fig4}, compared with the
451: observations (dotted line, see Fig. \ref{fig3}). The observational
452: histogram of the RR Lyr periods has been obtained with a step of 0.025
453: d, which allows to appreciate the presence of a secondary peak in the
454: distribution. We checked this feature comparing the histograms from the
455: samples by Benk{\H o} et al. (2006) and by Corwin \& Carney (2001), with
456: an increasing step from 0.01 to 0.04 d. The two distributions appear
457: very similar, and both show the presence of the
458: secondary peak for steps $\leq$ 0.03 d, beyond which only the main peak is
459: evident. Therefore, we consider this feature as well established, and an
460: intriguing event that our simulation reproduces it. The agreement at
461: the low and high limits for the period are also quite satisfactory, as
462: well as the mean period. We think that the simultaneous satisfaction of
463: so many observational requirements gives strong support to our basic
464: choices.
465:
466: Changing $\Delta$(M) from 0.203 \Msun\ to 0.201 or 0.205 \Msun, the
467: positions of the peaks in Figs. \ref{fig4} and \ref{fig5} appear
468: slighlty shifted with respect to the observed distributions; in the case
469: of a lower mass loss, one finds too many red stars ($\sim$ 170). We
470: cannot exclude these values for what concerns the CM diagram, since the
471: distribution in \BMV\ cannot be considered completely determined
472: (photometric imprecision, colour transformations). In any case, a small
473: change in the absolute value of the mass loss does not affect the main
474: point, that is, the need for a very small dispersion in mass loss, as
475: discussed in the following.
476:
477: The sharply peaked mass loss allows to have an almost unique value for
478: the mass
479: evolving through the RR Lyr region, and this is reflected in the peaked
480: period distribution. A larger spread in mass loss will give rise to a larger
481: spread in the masses evolving through the instability strip, and the sharp
482: peak will be smoothed out.
483:
484: To show how stringent is the assumption of a very small
485: $\sigma$, the right panel of Figure \ref{fig5} shows the period
486: distribution for an identical simulation, differing only for the increase in
487: $\sigma$ to 0.005 \Msun. We see that the period distribution is much
488: flatter and the peak had disappeared. We applied a Kolmogorov Smirnov
489: (KS) test to the simulated and observed periods, to derive the
490: probabilities that they are drawn from the same parent
491: distributions. The simulation shown in the left panel has a probability
492: of 89\%, while this is reduced to 5\% for the right panel.
493: In order to prove that the situation in Fig. \ref{fig5} is not
494: peculiar, we applied the KS test to 500 simulations with $\sigma$=0.0015 as well as to 500
495: ones with $\sigma$=0.005. The simulations have strictly the same inputs, apart from
496: the dispersion in mass loss. The probability that the observed
497: and simulated periods are extracted from the same distribution is $>$50\% for
498: 379 cases (75\%) when $\sigma$=0.0015, but only for 15 cases (3\%) when $\sigma$=0.005.
499: In the intermediate case of $\sigma$=0.003, we have probability $>$50\% for 245 cases
500: out of 500. In addition, remember that $\sigma$=0.005 is still a small mass loss dispersion indeed,
501: to be compared with $\sigma$=0.02 commonly assumed to reproduce the color extension of HB stars
502: in M3.
503:
504: \section{Discussion of the other simulation requirements}
505:
506: In building up a synthetic HB model with variable helium content, one
507: might wonder how flexible (arbitrary?) is the choice of the number {\it
508: vs} helium distribution. Actually, the distribution in number among the
509: red and variable members on the one side, and the blue HB stars on the
510: other, depends directly on the assumed number for the first generation
511: (that is, with cosmological helium) stars. This number has to be lower
512: than the sum of red and variable stars, because some helium-enriched
513: stars will always populate these regions, even if originating on the
514: blue side of the RR Lyr zone. In turn, these stars cannot be too many,
515: otherwise the RRc region would turn out too populated, with also a
516: decrease in the mean RR Lyr period. On the other hand, the first
517: generation cannot be too numerous, otherwise the number of red stars
518: largely exceeds the expected one.
519:
520: The details of a helium distribution satisfying these conditions (being
521: the parent of the simulation discussed in the paper) are reported in the
522: lower panel of Fig. \ref{fig4}. While minor variations in the
523: specific numbers would not be relevant, some characteristic features can
524: be identified: i) the
525: cluster population is equally divided between original (281) and enriched
526: (290) helium abundances; ii) the bulk of the helium enriched population is
527: around Y $\sim$ 0.26; iii) a tail with Y up to $\sim$ 0.30 may be necessary
528: to explain most of the blue HB members (but we do not wish to put too
529: much emphasis on this point).
530:
531: Some stars (about 50) with 0.24 $\simlt$ Y $\simlt$ 0.245 appear in the
532: helium distribution: they are necessary to avoid too large a dip in
533: population between RR Lyr variables and blue HB stars. The same position
534: on the HB would have been attained if we had assumed that these stars
535: had Y = 0.24 and a slightly larger mass loss. The effect of an increase
536: in Y from 0.24 to 0.245 is equivalent, for what concerns the resulting
537: HB star mass, to an increase in the mass lost by an evolving giant, with
538: Y = 0.24, from 0.203 to 0.210 \Msun. A small asymmetry towards a larger
539: mass loss would not be unexpected, and would be present in about
540: 15\% of the first stellar generation (50/331).
541:
542: The amount of blue HB members is given roughly by the number of helium
543: enhanced stars. As stressed before, there are no definite observational
544: values for the relative numbers of red, variable and blue stars, and in
545: any case the number of blue members can be easily accomodated varying
546: the number of helium enhanced stars. If we exclude the 50 stars with
547: 0.245$\le$ Y $\le$0.24, according to the latter hypothesis, we are left
548: with $\sim$ 40\% of stars with enhanced helium. This percentage is close
549: to what we found for other GCs (D'Antona et al. 2006). However, in this
550: case the helium distribution is peaked at the relatively small value of
551: Y $\sim$ 0.26, compatible with the mild chemical anomalies observed in
552: this cluster (Sneden et al. 2004).
553:
554: As most of the HB members in Fig. \ref{fig4} (90\%) has Y $\leq$ 0.265,
555: this ensures that the CM diagram of M3 is not
556: expected to show peculiarities in the position of the main sequence,
557: the turn--off, etc. In particular, the position of the GB bump will be
558: very close to that expected for a purely cosmological Y content.
559: %ref 4
560: In addition, the HB luminosity of the stars at the blue side of the RR Lyr
561: region is only mildly affected by the enhanced helium, which at these
562: locations is in the range 0.25$\simlt$Y$\simlt$0.26. This is shown in Figure
563: \ref{fig6}, where on the observed HB are superimposed three tracks with
564: Y=0.24, a track with Y=0.26 and one with Y=0.28.
565: Maybe a better photometry could discriminate whether actually there is a
566: helium enhancement at the blue side of the RR Lyrs.
567:
568: \begin{figure}[tb]
569: \centering
570: \includegraphics[width=8cm]{f6.eps}
571: \caption{Tracks for standard helium Y=0.24, Z=0.001 and masses M=0.6588, 0.63
572: and 0.58 \Msun\ (from right to left, full lines) are superimposed to
573: the observational HB data. The dashed (red) line is the M=0.63 \Msun\
574: track for Y=0.26, and the dot--dashed (red) line is the M=0.58 \Msun\
575: track for Y=0.28. The presence of slightly helium enhanced stars at the
576: blue side of the RR Lyrs appears consistent with the data.
577: }
578: \label{fig6}
579: \end{figure}
580: The problem of the red star number has been widely discussed in
581: Castellani et al. (2005), who found by far too many stars of this
582: kind. We are less plagued by this problem, but still the number of red
583: stars in the simulations turns out most of the times larger than
584: observed. It is not easy to think of a way out of this excess, given the
585: strict conditions imposed by the RR Lyrae colour and period
586: distribution. As mentioned before, the contribution to the variables by
587: stars coming from the blue side cannot be too large because of the dip
588: in population just at the blue of the RR Lyrae and in the RRc region
589: (see Fig. 5 in Corwin \& Carney 2001). Besides, too many variables
590: coming from the blue would lower substantially the observed RR Lyr mean
591: period. This question will be eaxamined in more detail in the following
592: Section.
593:
594: \section{Discussion and conclusions}
595:
596: \subsection{The global interpretation of the HB of M3}
597:
598: The extraordinary precision in the mass loss required by the simulations
599: deserves some comments. In Fig. \ref{fig4} it is shown the track of a
600: HB star with a mass of 0.659 \Msun, corresponding to the evolving giant
601: of 11 Gyr (Z=0.001, Y=0.24) after a mass loss of 0.203 \Msun. The two
602: peaks in the histogram of the star distribution in the red and variable
603: regions (Fig. \ref{fig4}) have a clear correspondence with the time
604: development along the evolutionary track. The peak in the red region is
605: given by the first evolutionary phases along the short redward loop and
606: the turning toward the blue, while the peak in the RR Lyr region is
607: the result of the slowing down of the evolution in colour during the
608: bend of the track at the end of the blueward loop, with the beginning of
609: the final movement to the red. The requirement of a small dispersion in
610: mass is due to the necessity of keeping this pattern clear of the
611: confusion that would arise mixing the evolutions of even a rather small
612: mass spectrum.
613:
614: Figures \ref{fig4} and \ref{fig5} show that RRc variables derive mainly
615: from helium-enhanced stars, at the colour interval around the maximum
616: blue extension of the 0.659 \Msun\ track. Here one finds also the dip in
617: the star distribution at the RR Lyr blue limit, set by the colour
618: limit of HB members with original helium content, and by the small
619: difference in helium content among the first star generation and the
620: succeeding ones. The dispersion along the blue HB, given the reduced
621: dispersion {\it in mass}, is due to the dispersion {\it in helium}, as
622: shown in the lower panel of Fig. \ref{fig4}.
623:
624: In fact, there are two main differences between our analysis and the
625: analysis by \cite{ccc2005}: i) they assume for the mass distribution
626: which provides the RR Lyr variables a $\sigma=0.005$ \Msun, while we
627: have shown that our best matches with the period distribution are
628: obtained by a $\sigma$\ much smaller -- almost a unique mass is evolving
629: in the RR Lyr region; ii) they have to fill up the blue side of the HB
630: with an {\it ad hoc} population, while we assume the model of a second
631: stellar generation with varying helium content, as we have already done
632: to interpret other complex HB distributions, such as in NGC 2808
633: \citep{dantonacaloi2004} and in NGC 6441 \citep{cd2007}. In this
634: framework, our interpretation for the strong constraint on the mass loss
635: rate along the RG is much more cogent.
636:
637: \subsection{A sharp value of mass loss also in other GCs?}
638:
639: The population distribution along the HB and the period distribution of
640: the RR Lyr variables have been obtained assuming the presence of (at
641: least) two populations with differing helium contents and, most
642: important, an extremely narrow dispersion in the mass loss. This latter
643: condition may be considered as necessarily requiring the presence of
644: the first condition, since the blue HB population cannot be
645: understood in terms of ``one mass'' evolution, as it happens for the red
646: and variable members.
647:
648: So the multiple star generations with varying Y hypothesis appears
649: capable of coping with the rather strong condition of a very sharp
650: amount of mass loss during RG evolution in M3
651: population. The variations in Y takes the place of the variations in
652: mass, to explain the HB star distribution. We have seen already in
653: other clusters similar situations. For example, in NGC 2808 (D'Antona \&
654: Caloi 2004) the red clump required a relatively small mass dispersion
655: ($\sigma \simeq$ 0.015), and the long blue HB region was obtained with a
656: distribution in Y. But the situation in M3 appears much more stringent
657: in the requirement on mass loss, basically due to the RR Lyr properties
658: that cannot be satisfied with a not--peaked mass distribution.
659:
660: How special is the case of M3? The point has been discussed by Catelan
661: (2004) and Castellani et al. (2005), who incline to consider M3 as
662: perhaps a ``pathological'' case. They observe that M5 and M62, both rich
663: in RR Lyr variables, show a less pronounced peak in the period
664: distribution (see also Castellani et al. 2003, Fig. 1). While this is perhaps
665: true, at least for M5, one finds in the quoted Fig. 1 (Castellani et
666: al. 2003),
667: many other Oosterhoff I clusters with about 70 RR variables (as in M62,
668: according to Clement et al. 2001) with strongly peaked period
669: distributions (NGC 6715, 6934, 7006). The problem with these clusters
670: appears similar to that found in M3. Let us quote also the case of
671: Palomar 3 (Catelan et al. 2001), that requires a very small mass
672: dispersion (consistent with zero, according to the authors) to describe
673: the HB distribution.
674:
675: We think therefore that the problem is with us to stay. If the small
676: dispersion in mass loss will be confirmed, we will face a rather
677: impressive change of perspective in the interpretation of HB
678: distributions. The role of the first phases of the dynamical and
679: chemical evolution of GCs will become more and more crucial. In turn,
680: these phases are dominated by the behaviour of the initial mass
681: function, of the chemical processes in asymptotic giant branch stars and
682: on their mass loss efficiency. So, on the one side we think that the
683: hypothesis of multiple stellar generations begins to find a certain
684: amount of observational support, and on the other we are well aware of
685: the long way to reach an understanding of how such a phenomenon took place.
686:
687:
688: \acknowledgements
689: We thank Drs. C. Corsi, T.M. Corwin and F. Ferraro for kindly sending
690: us their data on variable and not variable stars in M3 horizontal
691: branch, and Dr. M. Castellani for useful discussion.
692:
693: %__________________________________________________________________
694: \begin{thebibliography}{}
695: \bibitem[Bedin et al.(2004)]{bedin2004} Bedin, L.~R., Piotto, G.,
696: Anderson, J., Cassisi, S., King, I.~R., Momany, Y., \& Carraro, G.\ 2004,
697: \apjl, 605, L125
698:
699: \bibitem[Benk{\H o} et al.(2006)]{benko2006} Benk{\H o}, J.~M.,
700: Bakos, G.~{\'A}., \& Nuspl, J.\ 2006, \mnras, 372, 1657
701:
702: \bibitem[Bessell, Castelli, \& Plez(1998)]{bessell-castelli-plez1998}
703: Bessell, M.~S., Castelli, F., \& Plez, B.\ 1998, \aap, 333, 231
704:
705: \bibitem[Bono et al.(1995)]{bono1995} Bono, G., Caputo, F., \&
706: Stellingwerf, R.~F.\ 1995, \apjs, 99, 263
707:
708: %\bibitem[Castelli, Gratton \& Kurucz (1997)]{castelli1997}
709: %Castelli F., Gratton R.G., Kurucz R.L., 1997, A\&A 318, 841
710:
711: \bibitem[Brown et al.(2001)]{sweigart2001} Brown, T.~M., Sweigart,
712: A.~V., Lanz, T., Landsman, W.~B., \& Hubeny, I.\ 2001, \apj, 562, 368
713:
714: \bibitem[Buonanno et al.(1994)]{buonanno1994} Buonanno, R., Corsi,
715: C.~E., Buzzoni, A., Cacciari, C., Ferraro, F.~R., \& Fusi Pecci, F.\ 1994,
716: \aap, 290, 69
717:
718: \bibitem[Caloi \& D'Antona(2007)]{cd2007} Caloi, V., \&
719: D'Antona, F.\ 2007, \aap, 463, 949
720:
721: \bibitem[Castellani \& Tornambe(1981)]{ct1981} Castellani, V.,
722: \& Tornambe, A.\ 1981, \aap, 96, 207
723:
724: \bibitem[Castellani et al.(2003)]{cfc2003} Castellani, M.,
725: Caputo, F., \& Castellani, V.\ 2003, \aap, 410, 871
726:
727: \bibitem[Castellani et al.(2005)]{ccc2005} Castellani, M.,
728: Castellani, V., \& Cassisi, S.\ 2005, \aap, 437, 1017
729:
730: \bibitem[Catelan et al.(2001)]{catelan2001} Catelan, M., Ferraro,
731: F.~R., \& Rood, R.~T.\ 2001, \apj, 560, 970
732:
733: \bibitem[Catelan(2004)]{catelan2004} Catelan, M.\ 2004, \apj, 600, 409
734:
735: \bibitem[Clement et al.(2001)]{cle2001} Clement, C.~M., et al.\
736: 2001, \aj, 122, 2587
737:
738: \bibitem[Corwin \& Carney(2001)]{cc2001} Corwin, T.~M., \&
739: Carney, B.~W.\ 2001, \aj, 122, 3183
740:
741: \bibitem[Corwin et al.(1999)]{ccn1999} Corwin, T.~M., Carney,
742: B.~W., \& Nifong, B.~G.\ 1999, \aj, 118, 2875
743:
744: \bibitem[D'Antona et al.(2002)]{dantona2002} D'Antona,
745: F., Caloi, V., Montalb{\' a}n, J., Ventura, P., \& Gratton, R.\ 2002, \aap,
746: 395, 69
747:
748: %\bibitem[D'Antona (2003)]{dantona2003} D'Antona, F.\ 2003, Societa
749: %Astronomica Italiana Memorie Supplement, 3, 64
750:
751: \bibitem[D'Antona \& Caloi(2004)]{dantonacaloi2004} D'Antona, F., \&
752: Caloi, V.\ 2004, \apj, 611, 871
753:
754: \bibitem[D'Antona et al.(2005)]{dantona2005} D'Antona, F., Bellazzini,
755: M., Caloi, V., Fusi Pecci, F., Galleti, S., \& Rood, R.~T.~2005, ApJ,
756: 631, 868
757:
758: \bibitem[D'Antona et al.(2006)]{2006astro.ph.12654D} D'Antona, F.,
759: Ventura, P., \& Caloi, V.\ 2006, invited talk at the Conference: From
760: Stars to Galaxies: Building the Pieces to Build up the Universe, Venice,
761: Astronomical Society of the Pacific Conference Series,
762: Eds. A. Vallenari, R. Tantalo,.L. Portinari and A. Moretti, in press,
763: ArXiv Astrophysics e-prints, arXiv:astro-ph/0612654
764:
765: \bibitem[Di Criscienzo et al.(2004)]{dmc2004} Di Criscienzo,
766: M., Marconi, M., \& Caputo, F.\ 2004, \apj, 612, 1092
767:
768: \bibitem[Ferraro et al.(1998)]{ferraro1998} Ferraro, F.~R.,
769: Paltrinieri, B., Pecci, F.~F., Rood, R.~T., \& Dorman, B.\ 1998, \apj, 500,
770: 311
771:
772: \bibitem[Ferraro et al.(1997)]{ferraro1997} Ferraro, F.~R.,
773: Carretta, E., Corsi, C.~E., Fusi Pecci, F., Cacciari, C., Buonanno, R.,
774: Paltrinieri, B., \& Hamilton, D.\ 1997, \aap, 320, 757
775:
776: \bibitem[Fusi Pecci et al.(1993)]{fp1993} Fusi Pecci, F.,
777: Ferraro, F.~R., Bellazzini, M., Djorgovski, S., Piotto, G., \& Buonanno,
778: R.\ 1993, \aj, 105, 1145
779:
780: \bibitem[Gratton et al.(2001)]{gratton2001} Gratton,
781: R.~G.~et al.\ 2001, \aap, 369, 87
782:
783: \bibitem[Johnson \& Bolte(1998)]{jb1998} Johnson, J.~A., \&
784: Bolte, M.\ 1998, \aj, 115, 693
785:
786: \bibitem[Lee, Demarque, \& Zinn(1994)]{lee1994} Lee, Y.,
787: Demarque, P., \& Zinn, R.\ 1994, \apj, 423, 248
788:
789: \bibitem[Lee et al.(2005)]{lee2005} Lee, Y.-W., et al.\ 2005,
790: \apjl, 621, L57
791:
792: \bibitem[Moehler \& Sweigart(2006)]{moehler2006} Moehler, S., \&
793: Sweigart, A.~V.\ 2006, \aap, 455, 943
794:
795: \bibitem[Norris et al.(1981)]{norris1981} Norris, J., Cottrell,
796: P.~L., Freeman, K.~C., \& Da Costa, G.~S.\ 1981, \apj, 244, 205
797:
798: \bibitem[Norris(2004)]{norris2004} Norris, J.~E.\ 2004, \apjl,
799: 612, L25
800:
801: %\bibitem[Piotto et al.(2002)]{piotto2002} Piotto, G., et al.\
802: %2002, \aap, 391, 945
803:
804: \bibitem[Piotto et al.(2005)]{piotto2005} Piotto, G., et al.\
805: 2005, \apj, 621, 777
806:
807: \bibitem[Piotto et al.(2007)]{piotto2007} Piotto, G., et al.\
808: 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0703767
809:
810: \bibitem[Preston(1961)]{preston1961} Preston, G.\ 1961, \apj, 133, 29
811:
812: \bibitem[Reimers(1977)]{reimers1977} Reimers, D.\ 1977, \aap, 61,
813: 217
814:
815: \bibitem[Rich et al.(1997)]{rich1997} Rich, R.~M., et al.\ 1997,
816: \apjl, 484, L25
817:
818: \bibitem[Rood(1973)]{rood1973} Rood, R.~T.\ 1973, \apj, 184, 815
819:
820: \bibitem[Rood \& Crocker(1989)]{rc1989} Rood, R.~T., \&
821: Crocker, D.~A.\ 1989, IAU Colloq.~111: The Use of pulsating stars in
822: fundamental problems of astronomy, 103
823:
824: \bibitem[Sandage(1990)]{sandage1990} Sandage, A.\ 1990, \apj, 350, 603
825:
826: \bibitem[Sneden et al.(2004)]{2004AJ....127.2162S} Sneden, C., Kraft,
827: R.~P., Guhathakurta, P., Peterson, R.~C., \& Fulbright, J.~P.\ 2004, \aj,
828: 127, 2162
829:
830: \bibitem[Silbermann \& Smith(1995)]{ss1995} Silbermann, N.~A.,
831: \& Smith, H.~A.\ 1995, \aj, 110, 704
832:
833: \bibitem[Sosin et al.(1997)]{sosin1997} Sosin, C., et al.\ 1997,
834: \apjl, 480, L35
835:
836: %\bibitem[Sweigart \& Gross(1976)]{sweigartgross1976} Sweigart, A.~V., \&
837: %Gross, P.~G.\ 1976, \apjs, 32, 367
838:
839: %\bibitem[Sweigart \& Catelan(1998)]{sweigart1998} Sweigart, A.~V.,
840: %\& Catelan, M.\ 1998, \apjl, 501, L63
841:
842: \bibitem[van Albada \& Baker(1973)]{va1973} van Albada, T.~S.,
843: \& Baker, N.\ 1973, \apj, 185, 477
844:
845: %\bibitem[Ventura et al. (1998)]{ventura1998}
846: %Ventura, P., Zeppieri, A., Mazzitelli, I., \& D'Antona, F.\ 1998, \aap, 334, 953
847:
848: \bibitem[Ventura et al. (2001)]{ventura2001} Ventura,
849: P., D'Antona, F., Mazzitelli, I., \& Gratton, R.\ 2001, \apjl, 550, L65
850:
851: \bibitem[Ventura, D'Antona, \& Mazzitelli (2002)]
852: {ventura2002} Ventura, P., D'Antona, F., \& Mazzitelli, I.\ 2002, \aap, 393,
853: 215
854:
855: \bibitem[Ventura \& D'Antona(2005a)]{venturadantona2005a}
856: Ventura, P., \& D'Antona, F.~2005a, A\&A, 431, 279
857:
858: \bibitem[Ventura \& D'Antona(2005b)]{venturadantona2005b}
859: Ventura, P., \& D'Antona, F.~2005b, ApJ, 635, L149
860:
861: \bibitem[Wehlau et al.(1999)]{wehlau1999} Wehlau, A., Slawson,
862: R.~W., \& Nemec, J.~M.\ 1999, \aj, 117, 286
863:
864: %\bibitem[Walker(1994)]{walker1994} Walker, A.~R.\ 1994, \aj, 108,
865: %555
866:
867: \end{thebibliography}
868:
869: \end{document}
870:
871:
872: