1:
2: \documentclass[useAMS,usenatbib]{mn2e}
3: \usepackage{epsfig}
4: \usepackage{amsmath}
5: \usepackage{natbib}
6:
7: \title[An optimized H$_{\beta}$ index for disentangling stellar population ages.]{An optimized H$_{\beta}$ index for disentangling stellar population ages}
8: \author[]{J.L. Cervantes$^{1}$\thanks{E-mail:
9: joseluis@iac.es} and A. Vazdekis$^{1}$\thanks{E-mail: vazdekis@iac.es }\\
10: $^{1}$Instituto de Astrof\'{\i}sica de Canarias, La Laguna, 38200 Tenerife, Spain\\
11: }
12: \begin{document}
13:
14: \date{}
15:
16: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{}
17:
18: \maketitle
19:
20: \label{firstpage}
21:
22: \begin{abstract}
23:
24: We have defined a new H$_{\beta}$ absorption index definition, H$_{\beta_{o}}$,
25: which has been optimized as an age indicator for old and intermediate age
26: stellar populations. Rather than using stellar spectra, we employed for this
27: purpose a library of stellar population SEDs of different ages and metallicities
28: at moderately high spectral resolution. H$_{\beta_{o}}$ provides us with
29: improved abilities for lifting the age-metallicity degeneracy affecting the
30: standard H$_{\beta}$ Lick index definition. The new index, which has also been
31: optimized against photon noise and velocity dispersion, is fully characterized
32: with wavelength shift, spectrum shape, dust extinction and [$\alpha$/Fe]
33: abundance ratio effects. H$_{\beta_{o}}$ requires spectra of similar qualites as
34: those commonly used for measuring the standard H$_{\beta}$ Lick index
35: definition. Aiming at illustrating the use and capabilities of H$_{\beta_{o}}$
36: as an age indicator we apply it to Milky Way globular clusters and to a well
37: selected sample of early-type galaxies covering a wide range in mass. The
38: results shown here are particularly useful for applying this index and
39: understand the involved uncertainties.
40:
41: \end{abstract}
42:
43: \begin{keywords}
44: galaxies: abundances -- galaxies: elliptical and lenticular,cD --
45: galaxies: stellar content -- globular clusters: general
46: \end{keywords}
47:
48: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
49:
50: \section{Introduction}
51:
52: To understand how galaxies form and evolve, we need to study their stellar
53: populations as they are like fossils where the different formation and
54: evolutionary processes are registered. Since stars are not resolved for distant
55: stellar populations, one relies upon intergrated colors and spectra to obtain
56: their physical parameters, such as ages or metallicities. However the integrated
57: light of galaxies suffers the well-known age-metallicity degeneracy, making a
58: galaxy to look redder because it is older or more metal rich (e.g,
59: \cite{1994ApJS...95..107W,1986A&A...164..260A}).
60:
61: Unlike colors, spectroscopic absorption line-strength indices are more promising
62: at breaking the age-metallicity degeneracy, but even the most popular age
63: indicator, i.e., the Lick H$_{\beta}$ index (hereafter, H$_{\beta_{LICK}}$),
64: does show a significant dependance on metallicity, particularly for old stellar
65: populations \citep{1994ApJS...95..107W}. Despite the fact that other age
66: indicators based on H$_{\gamma}$ feature have shown a larger sensitivity to age
67: than H$_{\beta_{LICK}}$ index, their signal-to-noise requirements are extremely
68: high or their dependence on spectral resolution and velocity dispersion make
69: them very difficult to apply for a large variety of data and targets
70: \citep{1995ApJ...446L..31J, 1999ApJ...525..144V}.
71:
72: Balmer lines are commonly used as age indicators, although they are not totally
73: immune to metallicity effects for integrated stellar populations. In order to
74: cope with the fact that H$_{\beta}$ can be filled in with nebular emission
75: \citep{1993PhDT.........7G}, indicators based on higher order Balmer lines
76: (H$_{\delta}$, H$_{\gamma}$) have been proposed (e.g.,
77: \cite{1997ApJS..111..377W}, \cite{1999ApJ...525..144V}). The major drawback of
78: these index definitions based on the H$_{\delta}$ and H$_{\gamma}$ features is
79: that they have been shown to be significantly more sensitive to the total
80: metallicity and [$\alpha$/Fe] ratio than H$_{\beta_{LICK}}$
81: \citep{2004MNRAS.351L..19T,2005A&A...438..685K}. Nowadays the developement of
82: techniques where stellar and ionized-gas contributions to the galactic spectra
83: are simultaneously described allow us to decouple nebular emission from the
84: absorption features (e.g., \cite{2006MNRAS.366.1151S}). Such procedures make it
85: possible to explote the potential of H$_{\beta}$ feature as an age indicator,
86: even when gas emission is present.
87:
88: Plotting H$_{\beta}$ index versus a metallicity indicator, measured on a set of
89: SSP spectra of different ages and metallicities, provides a diagnostic diagram
90: that allows us to partially lift the age-metallicity degeneracy. However it has
91: been shown the disadvantage of using H$_{\beta_{LICK}}$ versus various
92: metallicity indicators for measuring mean luminosity weighted ages of early-type
93: galaxies due to the fact that the resulting model grids are not fully
94: orthogonal. Indeed this method leads to younger age estimates when
95: H$_{\beta_{LICK}}$ is plotted versus Mg$_{b}$ than when is plotted versus an Fe
96: index if a galaxy is [Mg/Fe] overabundant (e.g., \cite{2006ApJ...637..200Y}).
97: However a virtually orthogonal model grid is obtained when H$_{\beta_{LICK}}$ is
98: replaced by H$\gamma_{\sigma}$ \citep{1999ApJ...525..144V}. Unfortunately the
99: very high S/N required for measuring this index ( S/N (\AA) $>$ 150) limits its
100: applicability to nearby and bright objects for which such spectra can be
101: obtained. A more popular approach for obtaining consistent age estimates
102: requires the use of models that specifically take into account the non
103: scaled-solar elements ratios and the simultaneous measurement of several metal
104: lines to constrain the [$\alpha$/Fe] ratio
105: \citep{2004MNRAS.351L..19T,2005A&A...438..685K}. However, the obtained results
106: might depend on the details of modelling and on the particular element partition
107: employed. Furhermore the method is restricted to the use of the Lick/IDS system
108: of indices.
109:
110: In this paper, we explore, in a systematic manner, different index definitions
111: for the H$_{\beta}$ feature to find a new H$_{\beta}$ indicator that is
112: virtually insensitive to metallicity, which does not require particularly high
113: S/N. In section 2 we describe the optimization procedure developed to find the
114: new H$_{\beta}$ indicator, which makes use of a SSP library of SEDs at
115: moderately high resolution. In Section 3 we introduce the new H$_{\beta}$ index
116: definition and in section 4 we tackle its main characteristics. In section 5 we
117: test the reliability of this index on real data. Finally our conclusions are
118: summarized in section 6.
119:
120: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
121:
122: \begin{figure*}
123:
124: \includegraphics[angle=0,height=8cm,width=8cm]{jlcervantes_fig1.eps}
125: \includegraphics[angle=0,height=8cm,width=8cm]{jlcervantes_fig2.eps}
126: \caption{The bandpasses of H$_{\beta_{LICK}}$ (\emph{left panel}) and
127: H$_{\beta_{o}}$ (\emph{right panel}) index definitions are shown on the top of
128: the panels, which include several SSP models. \textsf{Top panels}: Solar
129: metallicity SSP model SEDs of age = 3, 7 and 12\,Gyr (from black to light grey,
130: respectively). \textsf{Bottom panels}: SSP models of age 10\,Gyr and
131: metallicities [Fe/H] = +0.2, +0.0, -0.4 and -0.7 (from light grey to black,
132: respectively). The improvement of the H$_{\beta_{o}}$ definition is achieved by
133: avoiding the metal lines on the red-pseudocontinnum, which are included in the
134: H$_{\beta_{LICK}}$ definition.}
135: \label{figure1}
136: \end{figure*}
137:
138: \section[]{Index definition approach}
139:
140: In the Lick system
141: \citep{1984ApJ...287..586B,1993ApJS...86..153G,1994ApJS...94..687W,1997ApJS..111..377W},
142: an index is defined in terms of a central bandpass enclosing the feature
143: bracketed by two pseudocontinuum bandpasses at either side of the feature (blue
144: and red). Once the average fluxes in the pseudocontinua are obtained, a line is
145: drawn between their midpoints to represent the continuum of the central feature
146: bandpass allowing to define an index as a pseudoequivalent width.
147:
148: The standard Lick/IDS system of indices was defined on the basis of a stellar
149: spectral library that were not flux-calibrated and had a resolution $\sim$ 8.4
150: {\AA} FWHM (three times lower than achieved in modern galaxy surveys, such as
151: SDSS). Furthermore, the Lick/IDS spectral resolution is varying with $\lambda$
152: \citep{1997ApJS..111..377W}. The natural resolution of a galaxy spectrum is
153: given by the convolution of the employed instrumental resolution with the
154: velocity broadening due to galaxy dynamics. Therefore to apply any model
155: predictions based on the Lick/IDS system we need to smooth higher resolution
156: spectra of galaxies with velocity dispersion values lower than that
157: corresponding to the Lick/IDS system to match its resolution (i.e. $\sigma \geq
158: {\rm 200\,km\,s^{-1}}$). Furthermore to apply these model predictions to higher
159: velocity dispersion galaxies we also need to correct index measurements back to
160: the resolution of the Lick/IDS system (an uncertain method that usually
161: requires observing a set of reference Lick stars. Alternatively, we prefer to
162: use here SSP SEDs at higher resolution that allow us to measure absorption line
163: indices directly on the model spectra once smoothed to match galaxy velocity
164: dispersion. This allows a direct comparison to the index values measured on the
165: galaxy spectrum. This approach is advantageous for all stellar population
166: systems, no matter its velocity dispersion nor its instrumental resolution.
167:
168: We explote here the advantadge of the SSP SEDs to derive new indices or to
169: redefine previous ones, rather than employing stellar spectra or polynomial
170: fitting functions that relate the strengths of the absorption lines to stellar
171: atmospheric parameters. This new generation of stellar population synthesis
172: models allows us to perfom such analysis directly on the SSP spectra, allowing
173: us to understand how sensitive to the physical parameters of the stellar
174: populations a trial index definition is, thus avoiding the intermediate step of
175: modeling this index for individual stars. Here we accomplish this approach by
176: using the library of single-age, single-metallicity stellar population model
177: spectra of \citet{1999ApJ...513..224V} (hereafter V99) and its recent extension
178: \citet{vazdekis} (V07) on the basis of the stellar library MILES
179: (\cite{2006MNRAS.371..703S}). The fact that MILES is characterized by an
180: unprecedented stellar atmospheric parameter coverage
181: \citep{2007MNRAS.374..664C} allows predicting scaled-solar SEDs for old and
182: intermediate ages with metallicities $-1.7 \leq [M/H] \leq +0.2$ for the full
183: optical spectral range (3540-7410 {\AA}).
184:
185: The use of SSP SEDs to define new indicators does not only allow us to optimize
186: them to be sensitive to the main population parameters such as age or
187: metallicity, but also to minimize their sensitivity, for example, to the
188: instrumental resolution or S/N requirements. This is possible as we can
189: simulate spectra of stellar populations for different S/N and/or resolution (or
190: velocity dispersion) values. Furthermore with these SSP spectra it is
191: straightforward to simulate and analyze the effects of radial velocity (or
192: rotation curve), spectrum shape and dust extinction on each trial index
193: definition.
194:
195: For finding a new H$_{\beta}$ indicator we have adopted a Lick style index
196: definition, which consists of three bandpasses (feature and two
197: pseudocontinua). We then change the position of these bandpasses or modify
198: their widths until optimizing its sensitivity to a given parameter. Note that
199: there are alternative approaches for defining indices such as those of the Rose
200: system \citep{1985AJ.....90.1927R}, which are defined as the ratio between the
201: intensity of the central peak of two spectral features. Other index definitions
202: are for example the \emph{generic indices} \citep{2001MNRAS.326..959C} or those
203: considered pseudo-colors, e.g. D4000 \citep{1983ApJ...273..105B}.
204:
205: \subsection{Optimizing criteria}
206:
207: The goal of this work is to define a new index based on H$_{\beta}$ feature
208: suitable for obtaining accurate age estimates. We have carried out a
209: comprehensive analysis of this feature in order to obtain an index that is
210: virtually insensitive to the total metallicity, with great stability against
211: the smearing of this feature due to galaxy velocity dispersion (or instrumental
212: resolution) and with as lower S/N requirement as possible. Our method to derive
213: an optimized H$_{\beta}$ index definition developes a multicriteria analysis to
214: evaluate the above requirements for each trial index definition.
215:
216: \subsubsection{Sensitivity to age and metallicity}
217:
218: Our primary criterion is obtaining an index definition for the H$_{\beta}$
219: feature that maximizes our ability to measure ages. To investigate whether
220: metallicity effects can be decreased in the H$_{\beta}$ feature, we have
221: generalized the sensitivity parameter defined in \citet{1994ApJS...94..687W}
222: (see in there footnote 4):
223:
224: \begin{equation}
225: \frac{\beta}{\alpha} = \frac{\sqrt{\sum_{i} \Big[\frac{(\frac{1}{I_{i}})((\frac{\partial I_{i}}{\partial [M/H]})_{age}) }
226: {(\frac{1}{I_{i}})((\frac{\partial I_{i}}{\partial log(age)})_{[M/H]}) } \Big]^{2}}}{N}
227: \end{equation}
228:
229: \noindent where $i$ runs for SSP models with ages older than 5 \,Gyr and -0.7
230: $\leq [M/H] \leq$ +0.2, $i = 1,N$, and $I_{i}$ is the index value measured on
231: the $i$ SSP model for a given index definition and $N$ is the number of
232: employed SSPs. Metallicity indicators should provide values above 1, whereas
233: the value for an age indicator should tend to 0. \looseness-1
234:
235: Among the sistematically-generated configuration of
236: bandpasses, our method consists in choosing the ones that provide mininum
237: values for the $\frac{\beta}{\alpha}$ parameter. Note that the choice of
238: bandpasses that minimazes $\frac{\beta}{\alpha}$ should not depend on the
239: stellar population synthesis models in use.
240:
241: The parameter space where $\frac{\beta}{\alpha}$ is minimized covers the ages
242: and metallicities for most early-type galaxies \citep{1998yCat..21160001T} and,
243: for the H$_{\beta}$ case, represents the parameter subspace where the age
244: sensitivity is lower and where the metallicity has the larger effects. In
245: other words, we are optimizing our indicator to be sensitive to the age for the
246: worst possible cases.
247:
248: \subsubsection{Spectral resolution and velocity dispersion}
249:
250: One of the metioned advantages of using the SSP SEDs at higher resolution is
251: that it is straightforward to study the effects of the velocity dispersion (or
252: resolution) on a given line index definition. The models can be smoothed to
253: different levels, allowing us to evaluate its stability to $\sigma$ variations.
254: Note that $\sigma$ effects might be significant for some index definitions that
255: already are optimal according to the $\frac{\beta}{\alpha}$ parameter, as a
256: sufficiently small $\sigma$ variation might decrease the obtained age
257: sensitivities. To parametrize this effect, we calculate the partial derivative
258: of the index versus $\sigma$ for a solar metallicity SSP model of 10\,Gyr at
259: $\sigma = {\rm 150\,km\,s^{-1}}$:
260:
261: \begin{equation}
262: \Sigma = \frac{1}{I} \cdot\left(\frac{\partial I}{\partial \sigma}\right)_{\rm{[M/H]}=0.0, T=10\,Gyr \arrowvert_{\sigma = 150 kms^{-1}}}
263: \end{equation}
264:
265: In a multicriteria analysis, the parameters that have been taken into account
266: are not assigned the same weight, as our main purpose is to optimize our age
267: resolving power. If we were to obtain an index definition that provides the
268: largest $\sigma$-stability, we would not have retained those definitions
269: providing the better age disentangling sensitivity. Therefore we have discarded
270: index definitions with $ \Sigma > {\rm 4 \cdot 10^{-4}}$ within the subset of
271: solutions for which the $\frac{\beta}{\alpha}$ parameter was optimized. This is
272: equivalent to a maximum variation of 10\% when comparing the index value at the
273: nominal resolution of the models and at ${\rm 250\,km\,s^{-1}}$.
274:
275: \begin{table*}
276: \centering
277: \begin{minipage}{140mm}
278: \caption{H$_{\beta}$ INDEX DEFINITIONS }\label{table1}
279: \begin{tabular}{@{}lrrrrr@{}}
280: \hline
281: {} & {Blue pseudocontinuum} &
282: {Feature} &
283: {Red pseudocontinuum}&{}&{} \\
284: \cline{2-6} \\
285: {Index} & {\AA} & {\AA} &
286: {\AA}&{c$_{1}$}& {c$_{2}$} \\
287: \hline
288: H$_{\beta_{LICK}}$& 4827.875 4847.875 & 4847.875 4876.625& 4876.625
289: 4891.625&7.301& 0.2539\\
290: H$_{\beta_{o}}$& 4821.175 4838.404 & 4839.275 4877.097& 4897.445
291: 4915.845&9.025 & 0.2386\\
292:
293: \hline
294: \end{tabular}
295: \end{minipage}
296: \end{table*}
297:
298: \subsubsection{Signal-to-Noise requirements}
299:
300: We have studied the S/N requirements following the analytical approach of
301: \cite{1999PhDT........12C} (eqs. [9] and [43]-[44]) \footnote{Coefficients c1
302: and c2 are calculated from their equations (43) and (44) using definitions in
303: Table \ref{table1} and added there.}. Obviuously no index definition is
304: appropriate if the required S/N is extremely high. However, achieving minimum
305: index errors does not imply a mininum age uncertainty if the age indicator is
306: not totally insensitive to metallicity effects. In fact, the age-metallicity
307: degeneracy and the uncertainity of age derived from index errors are tightly
308: correlated \citep{1998yCat..21160001T}. We consider that
309: the indicator depends solely on age, so that the uncertanity on measuring
310: this parameter comes mostly from photon noise. Aiming at obtaining index
311: definitions that minimize the S/N requirement, we define $\sigma[I_{a}]$ as the
312: index error required to achieve a precission of 2.5\,Gyr on deriving the age for a
313: a 10\,Gyr and solar metallicity model. Once again we relax our S/N optimizing
314: criterium by discarding only those definitions whose S/N(\AA) requirements are
315: above 70.
316:
317: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
318:
319: \section{The new H$_{\beta}$ index }
320:
321: According to our method for defining indices we show in this section an
322: optimized age indicator based on the H$_{\beta}$ feature, hereafter
323: H$_{\beta_{o}}$. Table \ref{table1} lists the limiting wavelength of the
324: bandpasses for the H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ indices. The
325: bandpasses of two indices are shown in Figure \ref{figure1}. Although farther
326: along we show plots and results obtained on the basis of V07, virtually
327: identical results are achieved with V99 models.
328:
329:
330: H$_{\beta_{LICK}}$ shows little sensitivity to the metallicity, as no strong
331: metallic lines are included in the wavelength range of the index definition
332: \citep{2005A&A...438..685K}. However this dependence is not negligible as
333: inferred from the non-orthogonal model grids resulting when this index is
334: plotted versus the metallicity indicator [MgFe], \citep{2006ApJ...637..200Y}.
335: Furthermore this prevents us to derive a unique age, when different
336: metallicity indicators are used if the analyzed galaxies do not have
337: scaled-solar abundance patterns, particularly for old galaxies (see, for
338: example, their Figure \ref{figure7}). Metallic lines within H$_{\beta}$ are
339: mainly dominated by Mg via MgH absorption \citep{1995AJ....110.3035T}, Ti
340: \citep{2004MNRAS.353..917T} and Cr at $\sim$ 4885 {\AA}
341: \citep{2004MNRAS.351L..19T,2005A&A...438..685K}.
342:
343: H$_{\beta_{o}}$ index definition avoids the spectral range covering those
344: metallic lines as the red-pseudocontinuum is shifted toward the red, whereas
345: the blue-pseudocontinuum is narrower to decrease the metallicity dependency.
346: However, other metallicity variations affect the H$_{\beta_{LICK}}$ central
347: bandpass via a Ti line at 4871\,{\AA} and the H$_{\beta}$ line at $\sim$ 4862
348: {\AA}. Note that the depth of both lines show an opposed behaviour against
349: metallicity: the higher the metallicity, the larger is the strength of the Ti
350: line and the smaller the H$_{\beta}$ line. These opposed responses at
351: increasing metallicity are not balanced by each other within H$_{\beta_{LICK}}$
352: index definition. This index compesates this effect in part by incorporating a
353: metallicity dependence on the red-pseudocontinuum. In fact, it overcompensates
354: the global effect leading to a higher metallicity sensitivity. H$_{\beta_{o}}$
355: extends to the blue the blue-wing of the central bandpass to introduce the
356: required metallicity dependence, thus avoiding to introduce Cr in the
357: red-pseudocontinnuum, which causes the variation of H$_{\beta_{LICK}}$ with the
358: metallicity (Figure \ref{figure1}).
359:
360: Figure \ref{figure2} shows H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ index values
361: as a function of the age of the SSP for different metallicities, {at the nominal resolution
362: of the models (left) and at 225 \,Kms$^{-1}$ (right)\footnote{This broadening
363: is similar to the Lick/IDS resolution at $\sim$ 5000 {\AA}} The plot shows
364: how H$_{\beta_{o}}$ is significatly less sensitive to the metallicity than
365: H$_{\beta_{LICK}}$ as the lines representing models of different metallicities lay
366: almost on the same curve, particularly for the higher metallicities (i.e. [M/H]$\geq$ -0.7)}.
367: Note that this also applies for SSPs with ages below 5\,Gyr.
368:
369: \begin{figure*}
370: \includegraphics[angle=0,scale=.50]{jlcervantes_fig4.eps}
371: \includegraphics[angle=0,scale=.50]{referee_fig3.eps}
372: \includegraphics[angle=0,scale=.50]{jlcervantes_fig3.eps}
373: \includegraphics[angle=0,scale=.50]{referee_fig3a.eps}
374: \caption{H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ indices measured on the V07 SSP
375: spectral library at the nominal resolution of these models (left, FWHM=2.3\AA)
376: and at a resolution similar to that of the Lick/IDS system (right, FWHM $\sim$
377: 8.4\AA)}
378: \label{figure2}
379: \end{figure*}
380:
381: \subsection{Age-metallicity degeneracy}
382:
383: We obtain for H$_{\beta_{o}}$ $\frac{\beta}{\alpha} < 0.01$. For comparison,
384: the standard H$_{\beta_{LICK}}$ index provides $\frac{\beta}{\alpha} \simeq
385: 0.45$. (See Table \ref{table2}), confirming that the new index definition
386: increases significantly the age sensitivity.
387:
388: \begin{figure}
389: \includegraphics[angle=0]{jlcervantes_fig8.eps}
390: \caption{Evolution of H$_{\beta_{LICK}}$ (dashed line) and H$_{\beta_{o}}$
391: (solid line) versus spectral resolution ($\sigma$). The line represents the
392: mean values for SSPs older than 1\,Gyr. Although H$_{\beta_{o}}$ index is more
393: dependent on $\sigma$, the variation is lower than 5\% at $\sigma <$ 250 ${\rm
394: km\,s^{-1}}$.}
395: \label{figure3}
396: \end{figure}
397:
398: \subsection{Spectral resolution and velocity dispersion}
399:
400: H$_{\beta_{LICK}}$ presents a lower $\Sigma$ value than H$_{\beta_{o}} $(Table
401: \ref{table2} $\approx$ 0.1 against 0.19 for H$_{\beta_{o}}$).
402:
403: Figure \ref{figure3} shows that both H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$
404: depend very little on resolution. In fact this dependence is lower than 5 \%
405: when comparing the index value at ${\rm 300\,km\,s^{-1}}$ and at the nominal
406: resolution of the models. Despite the fact of the slightly larger dependence of
407: H$_{\beta_{o}}$ index on $\sigma$, this dependence does not affect its age
408: sensitivity since $\frac{\beta}{\alpha}$ is always lower than 0.15, and much
409: lower than the value obtained for H$_{\beta_{LICK}}$ (See Figure
410: \ref{figure4}).
411:
412:
413: \begin{figure}
414: \includegraphics[angle=0]{jlcervantes_fig8a.eps}
415:
416: \caption{Generalized Worthey's parameter, $\beta/\alpha$, versus model spectral
417: resolution ($\sigma$). This plot shows that H$_{\beta_{o}}$ is virtually insensitive to
418: metallicity as a function of $\sigma$ out to $\sigma \sim 300$. }
419:
420: \label{figure4}
421: \end{figure}
422:
423:
424:
425: \subsection{Signal-to-Noise}
426:
427: Table \ref{table2} lists $\sigma[I_{a}]$ in percentage. H$_{\beta_{o}}$ requires
428: slightly higher S/N than H$_{\beta_{LICK}}$ to distinguish the age once the
429: metallicity is known.
430:
431: However, would we have considered the real error on the age, its derivation
432: would have been different due to the fact that any metallicity information had
433: been taken into account in $\sigma[I_{a}]$. The real S/N requirements to derive
434: the age with precision is tightly related to the metallicity dependence of the
435: H$_{\beta}$ index definition. To estimate the age accuracy determination for an
436: index definition when we neglect the metallicity, we can focus for example on a
437: 10\,Gyr model, which is representative of the sub-space of ages older than
438: 5\,Gyr and metallicities in the range -0.7 $\leq$ [M/H] $\leq$ +0.2, where most
439: early-type galaxies are located \citep{1998yCat..21160001T}. We then simulate
440: the S/N effects on the spectrum and test the age accuracy using a plot such as
441: that of Figure \ref{figure2}. The obtained results are shown in Figure
442: \ref{figure5}, where the age uncertainity of H$_{\beta_{LICK}}$ decreases
443: asymptotically to $\sim$5.1\,Gyr with increasing S/N, while the minimun age
444: uncertainty associated to H$_{\beta_{o}}$ is $\sim$ 1\,Gyr at a S/N (per\,{\AA})
445: $\sim$250 \footnote{This plot can be used to prepare observations for galaxies
446: older than 5 \,Gyr and with metallicities higher than [M/H]=-0.7}. This means
447: that the maximun age accuracy achieved with H$_{\beta_{LICK}}$ is always lower
448: than that of H$_{\beta_{o}}$, no matter the spectrum quality, since this new
449: index is much less sensitive to metallicity.
450:
451:
452:
453: \begin{table*}
454: \centering
455: \begin{minipage}{140mm}
456: \caption{INDEX CHARACTERISTICS AND UNCERTAINTIES}
457: \label{table2}
458:
459: \begin{tabular}{@{}lrr@{}}
460: \hline
461: {Parameter} & {H$_{\beta_{LICK}}$} &
462: {H$_{\beta_{o}}$} \\
463: \hline
464: $\frac{\beta}{\alpha}$ & $\sim$ 0.45& $<$ 0.1\\
465: \\
466: $\Sigma$ ($10^{-4}$ ${\rm km\,s^{-1}}$)&1.05&1.89\\
467: Index variation by ${\rm \pm 10\,km\,s^{-1}}$ in $\sigma$ (\%) & 0.105 & 0.189 \\
468: Maximum stability $\sigma$ range& ${\rm > 300\,km\,s^{-1}}$ & ${\rm \pm 250\,km\,s^{-1}}$\\
469: \\
470: Age uncertainty caused by a $\Delta\lambda$ = 1.5 {\AA} & 5.5-18 \,Gyr&6.7-14.2
471: \,Gyr\\
472: Maximum $\Delta \lambda$ shifts ($\lambda$, z, rotation curve) compatible with a $+/-$ 2.5 \,Gyr error ({\AA}) & 0.85&1.1\\
473: \\
474: (H$_{\beta}$(flux calibrated response curve) - H$_{\beta}$(continuum
475: removed))/H$_{\beta}$(f.c.r.c.) & $\lq$ 0.01 & $\lq$ 0.01 \\
476: \\
477: S/N (per\,{\AA}) required to distinguish 2.5 \,Gyr at 10 \,Gyr& $\sim$ 50 & $\sim$ 65\\
478: Minimum age uncertainity for S/N (per\,{\AA}) $>$ 250 & 5.1 \,Gyr & 1 \,Gyr \\
479: \hline
480:
481: \end{tabular}
482: \end{minipage}
483: \end{table*}
484:
485: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
486:
487: \section{Characterization of H$_{\beta_{o}}$}
488:
489: In this section we fully characterize H$_{\beta_{o}}$ index and discuss its
490: major uncertainties with several aspects that might influence its ability as an
491: age indicator.
492:
493: \subsection{Sensitivity to abundance ratio variations}
494:
495: An index definition includes the contribution of various chemical species,
496: despite the fact that a given element, which uses to name the index, might be
497: its major contributor. As giant elliptical galaxies show [Mg/Fe] overabundance
498: compared to the scaled-solar element partition we need to assess the influence
499: of such abundance ratios on our index definition and on its ability to
500: disentangle mean ages.
501:
502: As this task is difficult to accomplish with models based on empirical stellar
503: spectra, the use of theoretical atmospheres to compute stellar spectra for a
504: large variety of element mixtures is an advantage \citep{1995AJ....110.3035T}.
505: In order to quantify the dependence of H$_{\beta_{o}}$ on [$\alpha$/Fe] we
506: should use SSP models with varying abundance ratios. We use the approach
507: described in \cite{2007IAUS..241..167C} as it is based on the same model that we
508: have employed here, where varying [$\alpha$/Fe] has been implemented via
509: differential correction making use of \cite{2005A&A...443..735C} stellar
510: library. Note that alternative SSP model SEDs computed with non scaled-solar
511: ratios have been recently published by \cite{2007MNRAS.382..498C}.
512: The
513: \cite{2007IAUS..241..167C} model spectra cover two choices of partitions,
514: [$\alpha$/Fe] = 0.0 and 0.4, where O, Ne, Mg, Si, S, Ca and Ti are
515: flat-enhanced, whilst all the other elements follow Fe. The relation between the
516: total metallicity [M/H] and the iron content [Fe/H] is modified as\footnote{When
517: A = 0, we obtain the solar abundance of \cite{1998SSRv...85..161G}, adopted for
518: the synthetic spectra calculations of \cite{2005A&A...443..735C}}
519:
520: \begin{equation}
521: \rm{[M/H]} = [Fe/H] + A[\alpha/Fe]
522: \end{equation}
523:
524: \noindent At a given [M/H], the sentitivity of an index to [$\alpha$/Fe] can
525: be evaluated as
526:
527: \begin{equation}
528: {\Lambda}_{[\alpha/\rm{Fe}]} =
529: \Big[
530: \frac{|I_{[\alpha/{\rm Fe}]=0.4,[\rm{Fe/H}]=-0.3}-I_{[\alpha/{\rm Fe}]=0}|}{I_{[\alpha/{\rm Fe}]=0}}
531: \Big]_{[{\rm M/H}]=0;10\,Gyr}
532: \end{equation}
533:
534: \noindent where $I_{[\alpha/{\rm Fe}]=0}$ is the index value for a 10\,Gyr solar
535: metallicity scaled-solar model, and $I_{[\alpha/{\rm Fe}]=0.4,[\rm{Fe/H}]=-0.3}$
536: is the index value for an $\alpha$-enhanced ([$\alpha$/Fe]=0.4) model of the
537: same age and total metallicity but [Fe/H]=-0.3. ${\Lambda}_{[\alpha/{\rm Fe}]}$
538: should tend to zero if an index does not depend on [$\alpha$/Fe].
539: ${\Lambda}_{[\alpha/{\rm Fe}]}$ can be interpreted as a mean deviation caused by
540: the $\alpha$-enhancement with respect to the scaled-solar composition, which
541: translates to an age (or metallicity) uncertainty. Table \ref{table3} lists the
542: ${\Lambda}_{[\alpha/{\rm Fe}]}$ values, and the associated age uncertainty,
543: corresponding to the various Balmer line index definitions ($\Lambda_{CO}$). We
544: find a greater sensitivity of H$_{\beta_{o}}$ to [$\alpha$/Fe] in comparison to
545: that of H$_{\beta_{LICK}}$. However this effect is smaller than that found for
546: the \cite{1997ApJS..111..377W} higher order Balmer index definitions.
547:
548: We are aware that the SSP SEDs with varying abundance ratios might depend on the
549: atmospheres and spectral synthesis codes employed to compute the theoretical
550: stellar spectra, which feed these models. In fact the computation of stellar
551: spectra with varying [$\alpha$/Fe] ratios requires adopting lists of all
552: relevant atomic and molecular transitions, along with accurate oscillator
553: strengths and damping constants. Furthermore fitting detailed line profile would
554: require the inclusion of NLTE, sphericity, chromosphere effects, among other
555: aspects. We therefore have extended the \cite{2007IAUS..241..167C} analysis by
556: including the $\alpha$-enhanced theoretical stellar library of
557: \cite{2005A&A...442.1127M} with the only purpose of assessing the uncertainties
558: affecting our result. As for \cite{2007IAUS..241..167C}, the model spectra cover
559: two mixtures, [$\alpha$/Fe] = 0.0 and 0.4, with O, Ne, Mg, Si, S, Ca and Ti
560: enhanced as in the \cite{2005A&A...443..735C} stellar library.
561:
562: Table \ref{table3} lists the results obtained for the ${\Lambda}_{[\alpha/{\rm
563: Fe}]}$ parameter based on this alternative library. Surprisingly the two
564: libraries provide similar values for nearly all the Balmer line indices but
565: H$_{\beta}$. Unlike with the library of \cite{2005A&A...443..735C}, we obtain
566: that H$_{\beta_{o}}$ is less sensitive to $[\alpha/{\rm Fe}]$ variations than
567: H$_{\beta_{LICK}}$, (See Table \ref{table3},$\Lambda_{MU}$). Interestingly, the
568: H$_{\gamma_{\sigma}}$ index shows very little sensitivity to $[\alpha/Fe]$ for
569: the two libraries, which is not the case for the \cite{1997ApJS..111..377W}
570: indices. This result is in good agreement with \cite{2003MNRAS.339..897T}
571: conclusion for the latter indices. Such dependence of the H$_{\beta_{LICK}}$
572: index on [$\alpha$/Fe], when the library of \cite{2005A&A...442.1127M} is
573: employed, has been quoted before by \cite{2004MNRAS.353..917T}. This result is
574: however in disagreement with the abundance ratio insensitivity found by Tripicco
575: \& Bell (1995) and Korn et al. (2005) for H$_{\beta_{LICK}}$, as well as with
576: our own calculations on the basis of the library of \cite{2005A&A...443..735C}.
577: Therefore our conclusion on the greater sensitivity of H$_{\beta_{o}}$ on the
578: abundance ratio in comparison to H$_{\beta_{LICK}}$ must be taken with caveat,
579: as it might depend in part on the modelling of the stellar atmospheres. A
580: discussion of the feasibility of those stellar libraries is out of the scope of
581: this paper and we refer the reader to \cite{2007MNRAS.381.1329M} for an extended
582: analysis of these theoretical libraries and their use by SSP models. This
583: possible drawback of the new index definition is minimized by the fact that the
584: relative variation of H$_{\beta_{o}}$ to [$\alpha$/Fe] is very similar to that
585: of H$_{\beta_{LICK}}$ to the total metallicity. H$_{\beta_{o}}$ is however
586: completely safe for scaled-solar element partitions, with the advantage that it
587: provides almost orthogonal model grids. Finally, although not shown here, we
588: note that this orthogonality is preserved when [$\alpha$/Fe] enhanced models of
589: different ages and metallicities are employed to build-up the grids.
590:
591: \begin{figure}
592: \includegraphics[angle=0,scale=0.6]{jlcervantes_fig6.eps}
593: \caption{Age accuracy determination for a 10\,Gyr solar metallicity SSP model as
594: a function of S/N (per {\AA}) obtained on the basis of the H$_{\beta_{LICK}}$
595: and H$_{\beta_{o}}$ indices. This plot shows that the maximum age accuracy
596: reached with H$_{\beta_{o}}$ is $\sim$ 1\,Gyr at SN/{\AA} $\sim$250, while an
597: age error of $\sim$3\, Gyr is obtained at SN/{\AA} $\sim$50. No matter the
598: spectrum quality, the maximum age accuracy for H$_{\beta_{LICK}}$ is $\sim$
599: 5\,Gyr, if we neglect the metallicity information.}
600: \label{figure5}
601: \end{figure}
602:
603: \subsection{Wavelength and radial velocity uncertainties}
604:
605: As absorption line indices provide us with information from relatively narrow
606: bandpasses, errors in wavelength calibration, radial velocity or rotation
607: curve, lead to errors in our age/metallicity estimates. It is worth to note
608: that an accurate wavelength calibration is tipically around 5\% - 10\% of the
609: dispersion resulting from the adopted instrumental setup employed in the
610: observations. To account for this we obtain the largest wavelength errors or
611: shifts allowed to achieve a minimum precision of 2.5\,Gyr on age for a solar
612: metallicity SSP model of 10\,Gyr. In principle we should have discarded index
613: definitions that are unstable for wavelengths shifts of 0.5\,{\AA} (i.e.
614: tipically corresponding to dispersion of 5\,{\AA}), according to our
615: criterion. However the set of preselected definitions that were virtually
616: independent on metallicity did not show such sensitivity to wavelength shifts.
617: For this reason we only characterize the new H$_{\beta_{o}}$ index for
618: wavelength shifts and compare the obtained result with that for
619: H$_{\beta_{LICK}}$. Table \ref{table2} lists the largest wavelength errors
620: allowed to obtain a minimum precision of 2.5\,Gyr for a solar metallicity,
621: 10\,Gyr old SSP model for these two indices. This table also lists the age
622: uncertainty corresponding to a wavelength shift of 1.5\,{\AA} (i.e.
623: ${\rm \sim 90\,km\,s^{-1}}$). Figure \ref{figure11} shows how the $\frac{\beta}{\alpha}$ parameter
624: changes with increasing wavelength shift. We see that wavelength shifts do not
625: affect significantly the age resolving power of these two indices. Note that
626: the age-sensitivity of H$_{\beta_{o}}$ is always larger than that of
627: H$_{\beta_{LICK}}$. Interestingly, H$_{\beta_{LICK}}$ increases its ability for
628: disentangling ages if its bandpasses are shifted 1.7\,\AA\ blueward. The main
629: characteristics and capabilities of this alternative H$_{\beta_{LICK}}$ index
630: definition are sumarized in the Appendix.
631:
632: \begin{figure}
633: \includegraphics[angle=0,scale=0.5]{jlcervantes_fig11.eps}
634: \caption{Generalized Worthey's parameter, $\beta/\alpha$, versus wavelength
635: shifts. This plot shows that the age sensitivity of H$_{\beta_{o}}$ depends very
636: little on this parameter.}
637: \label{figure11}
638: \end{figure}
639:
640: \subsection{The Spectrum Shape}
641:
642: We have tested the effect of the spectral response curve on the H$_{\beta}$
643: indices. We expect this effect to be more significant for H$_{\beta_{o}}$ in
644: comparison to H$_{\beta_{LICK}}$. This is because the definition of
645: H$_{\beta_{o}}$spans a $\sim$ 30 {\AA} wider spectral range. We follow the the
646: test proposed by \cite{1986A&A...164..260A} and show in Table \ref{table2} the largest
647: differences, obtained for the oldest SSPs, when comparing the index mesurements
648: performed on the flux calibrated and continuum removed SSP spectra (using a
649: spline3 of order 4) of similar age and metallicity. We conclude that the effect
650: of the continuum shape on the index measurements is negligible for these two
651: indices.
652:
653: \subsection{Dust extinction}
654:
655: Recently, absorption-line studies of integrated stellar populations are being
656: extended to later type galaxies which may contain significant amounts of dust
657: (e.g., \cite{2007A&A...474.1081G,2007IAUS..241..420D}).
658: In the case of H$_{\beta_{LICK}}$, when dust extinction affects the SSP age
659: determination, the errors on the physical parameters are of the same order as
660: the ones measured in the index and thus would not likely be detected above the
661: noise \citep{2005ApJ...623..795M}. We should not expect significant differences
662: for H$_{\beta_{o}}$, though we have performed the same analysis followed by
663: \cite{2005ApJ...623..795M} who takes into account the two-component model of
664: \cite{2000ApJ...539..718C} for the influence of the interstellar medium on the
665: starlight. The two adjustable parameters of this model are $\tau_{V}$, the
666: total effective V-band optical depth affecting stars younger than $10^{7}$ \,yr
667: and $\mu$, the fraction of the total dust absorption contributed by diffuse
668: interstellar medium dust\footnote{See \cite{2005ApJ...623..795M} for a full
669: description of the method}.
670:
671: The variations of H$_{\beta_{o}}$ as a function of $\tau_{V}$, $\mu$ and age
672: for a solar metallicity SSP model is shown in Figure \ref{figure6}, where
673: $\Delta index$ versus $\tau_{V}$ is plotted. In this figure $\Delta index$ is the
674: difference between the index measured with and without dust, i.e.,$\Delta
675: index = index(\tau_{V}) - index(\tau_{V} = 0)$. Results are shown for model
676: ages of 1 (light grey lines), 5 (grey lines) and 13\,Gyr (black line). The two
677: values for $\mu$ of 0.5 and 1.0 correspond to the solid and dashed lines,
678: respectively. The black horizontal dotted line represents the measurement error
679: required to distinguish 2.5\,Gyr for a 10\,Gyr solar metallicity SSP model,
680: i.e., the accuracy we have imposed during our process of finding the new
681: H$_{\beta_{o}}$ index. For a 1\,Gyr scaled-solar
682: model (light grey lines), H$_{\beta_{LICK}}$ $\sim$ 4.3 \,{\AA} and
683: H$_{\beta_{o}}$ $\sim$ 5.5\,{\AA}, it corresponds to 1\% for H$_{\beta_{o}}$
684: and 0.2\% for H$_{\beta_{LICK}}$, in the two cases for a extreme dust extintion.
685: For a 13\,Gyr model (black lines), H$_{\beta_{LICK}}$ $\sim$ 1.8\,{\AA}
686: and H$_{\beta_{o}}$ $\sim$ 3 \,{\AA}, it means 0.3\% for H$_{\beta_{o}}$,
687: since $\Delta index$ $\sim$ 0.01 \,{\AA} and 0.6\% for H$_{\beta_{LICK}}$.
688:
689: As expected, the effect of extinction is larger in H$_{\beta_{o}}$ than in
690: H$_{\beta_{LICK}}$, since the involved wavelength coverage is larger. However,
691: the obtained effects are lower than the minimum index errors associated to the
692: noise (Figure \ref{figure6}), when extinction has been incorporated on top of
693: the SSP SEDs. It is worth noting than although dust extintion was not
694: considered as a parameter for the index definition proccess, it is not rejecting
695: a posteriori the obtained index.
696:
697: \begin{figure*}
698: \center
699: \includegraphics[angle=0,scale=0.6]{jlcervantes_fig5.eps}
700: \includegraphics[angle=0,scale=0.6]{jlcervantes_fig5a.eps}
701: \caption{H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ index residuals, $\Delta index
702: = index(\tau_{V}) - index(\tau_{V} = 0)$, as a function of the total effective
703: V-band optical depth, $\tau_{V}$, for solar metallicity SSPs. The results for
704: models of 1 (light grey lines), 5 (grey lines), and 13\,Gyr (black lines) are
705: shown. Different values of $\mu$ are represented as 0.5 (solid line), and 1.0
706: (dashed line). The black horizontal dotted lines represent the errors required
707: for distinguishing 2.5\,Gyr for a solar metallicity SSP model of 10\,Gyr.}
708: \label{figure6}
709: \end{figure*}
710:
711: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
712:
713: \section{Discussion}
714:
715: Estimating the mean luminosity-weighted age of an early-type galaxy represents
716: a major step for constraining its Star Formation History. In this section we
717: probe the H$_{\beta_{o}}$ as an age-dating indicator in real data. We apply
718: this index to Milky Way globular clusters and to prototype early-type galaxies
719: for which extremely high quality spectra are available.
720:
721: \subsection{Galactic Stellar Clusters}
722:
723: Globular clusters are ideal laboratories to test the SSP models as they can be
724: considered as stellar populations formed in a single and homogenous process.
725: Milky Way globular clusters allow us to check the consistency of the
726: H$_{\beta_{o}}$ age estimates as it is possible to obtain independent
727: age/metallicity values from detailed Color-Magnitud Diagram (CMD) analyses.
728:
729: Figure \ref{figure10} shows the H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ indices
730: measured on the Milky Way cluster sample of \cite{2005ApJS..160..163S} versus
731: the CMD-derived metallicities. The [Fe/H] values are taken from
732: \cite{1996AJ....112.1487H} (the 2003 version of the McMaster catalog). We
733: obtain for the H$_{\beta_{LICK}}$ sequence a Spearman rank coefficient value of
734: -0.89, whereas for H$_{\beta_{o}}$ we obtain -0.70, indicating a milder
735: anti-correlation for the latter. The slope of a linear fitting is -0.71
736: \,dex/{\AA} for H$_{\beta_{LICK}}$ and -0.36 \,dex/{\AA} for H$_{\beta_{o}}$. Note
737: that the fit for H$_{\beta_{o}}$ is virtually flat for [Fe/H]$>$-1.0.
738: Although we have shown in the previous section that H$_{\beta_{o}}$
739: is particularly optimized for higher metallicities ([M/H]$\geq$-0.7),
740: these results confirm the lower metallicity dependence of H$_{\beta_{o}}$ with
741: respect to H$_{\beta_{LICK}}$ on the basis of real data, without the use of
742: models, for all metallicities.
743:
744: \begin{figure*}
745: \center
746: \includegraphics[angle=0,scale=0.9]{run_cumulos_dual_paper_Hbetalick.ps}
747: \includegraphics[angle=0,scale=0.9]{run_cumulos_dual_paper_Hbetao.ps}
748: \caption{H$_{\beta_{LICK}}$ (left) and H$_{\beta_o}$ (right) vs. CMD-derived
749: metallicities for the globular cluster sample of Schiavon et al. (2005).
750: The [Fe/H] values were taken from the compilation of Harris 1996.}
751: \label{figure10}
752: \end{figure*}
753:
754: Figure \ref{figure7} shows H$_{\beta_{o}}$ and three age-dating indices versus
755: the mean metallicity indicator [MgFe] for the same cluster sample. We see that
756: the stellar clusters fall at the bottom of all these plots, indicating very old
757: ages. In fact for many cases we obtain ages that are larger than the oldest
758: models (i.e. older than the age of the Universe). These plots show the well
759: known model zero-point problem affecting the spectroscopic age determinations
760: from the Balmer line indices (e.g. \cite{1999AJ....118.1268G,
761: 2001ApJ...549..274V, 2002ApJ...580..850S}). Recently,
762: \cite{2007MNRAS.379.1618M} using multi-index $\chi^{2}$ minimization technique
763: analized two Milky Way globular cluster samples (the Schiavon's data used here
764: and Puzia et al. 2002) to test the age estimates obtained with three different
765: sets of updated models (Thomas et al. 2004, Lee \& Worthey 2005 and our
766: models). Although the ages inferred on the basis of their multi-index method
767: are in agreement with the CMD-derived ages, their Figure 2 shows basically the
768: same result, i.e. the observed H$_{\beta}$ values fall below the model grids for
769: the three models. Nonetheless the large age values is a result that is common
770: to all the Balmer line-strength index definitions shown in Figure \ref{figure7}.
771: However there are other works, working in the Lick/IDS system, in which this
772: offset is not seen, e.g. Thomas et al. (2003) where these models are used to fit
773: the Puzia et al.'s data (both, the models and the data, also employed in
774: \cite{2007MNRAS.379.1618M}). Note also that according to the $\Lambda_{CO}$
775: value listed in Table \ref{table3} for H$_{\beta_{o}}$ this offset is clearly
776: minimized when employing models based on the \cite{2005A&A...443..735C} library.
777: As this problem does not affect the relative age/metallicity sensitivities of
778: the H$_{\beta}$ index definitions used here we refer the interested reader to
779: the above papers for further details on this issue.
780:
781: In the H$_{\beta_{o}}$ plot we are able to distinguish two populations of
782: clusters for [MgFe] $>$ 1.5 {\AA}, which can also be seen in Figure
783: \ref{figure10}. With this evidence in our hands we can identify these two
784: cluster populations in the H$_{\beta_{LICK}}$ plot as well. However the two
785: cluster populations cannot be distinguished in the remaining two panels of
786: Figure \ref{figure7}, which include lower age-sensitivity indices. Therefore
787: this intriguing feature becomes more evident as the metallicity dependance of
788: the Balmer index definition has been significantly minimized. An extensive
789: study devoted to understand the origin of this feature is presented in
790: \cite{cenarrosb}. In this study we compare the CMDs of these
791: clusters and discuss among other aspects, the effects of the Horizontal Branch
792: morphology and the Blue Stragglers on the integrated H$_{\beta}$ indices. In
793: that paper we convincingly show that the latter is directly linked to the
794: observed two age-population feature and that this feature is not driven by the
795: zero-point problem.
796:
797: \begin{table*}
798: \centering
799: \begin{minipage}{140mm}
800: \caption{INDEX SENSITIVITY TO [$\alpha$/Fe] VARIATIONS}
801: \label{table3}
802: \begin{tabular}{@{}lrrrrr@{}}
803: \hline
804:
805: Index & $\Lambda_{CO}$ &
806: Age uncertainity & $\Lambda_{MU}$ &
807: Age uncertainity\\
808: & &
809: at 10 \,Gyr (in \,Gyr) & & at 10 \,Gyr (in \,Gyr) \\
810:
811: \hline
812: H$_{\beta_{LICK}}$& 0.011 & $<$ 1 &0.079 & 8-13\\
813: H$_{\beta_{o}}$& 0.119 & 5-18&$<$0.01& $<$ 1 \\
814: H$_{\sigma_{130}}$& 0.045 & 8-12 &0.020& 9-12\\
815: H$_{\delta_{A}}$& 0.75 & 2.5-18 &1.1 & 1-18\\
816: H$_{\gamma_{A}}$& 0.23 & 4-18 &0.17& 3.5-18\\
817: H$_{\delta_{F}}$& 1.7 &3.5-18 & 0.4& 7-15\\
818: H$_{\gamma_{F}}$& 0.6 &3.5-18 & 0.5& 3.5-18 \\
819:
820: \hline
821: \end{tabular}
822: \end{minipage}
823: \end{table*}
824:
825: \begin{figure}
826: \includegraphics[angle=0,scale=0.85]{diagrama_ver_cum.eps}
827: \caption{From top to bottom H$_{\beta_{o}}$, H$_{\beta_{LICK}}$,
828: H$_{\gamma_{A}}$ and H$_{\delta_{F}}$ versus [MgFe] index for the
829: Schiavon et al. (2005) sample of GCs (top to bottom). Model grids smoothed
830: to the resolution of the instrumental setup employed in the data are
831: overplotted. Metallicity increases from left to right (solid lines): [Fe/H] =
832: -1.7, -1.3, -0.7, -0.4, 0.0 and +0.2.
833: Age increases from top to bottom (dotted lines): Age = 3, 5, 8, 13 and 18\,Gyr.}
834: \label{figure7}
835: \end{figure}
836:
837: \subsection{Elliptical galaxies}
838:
839: In this section we use the H$_{\beta_{o}}$ index to estimate the ages of a well
840: selected sample of ellitpical galaxies. For selecting the sample we mainly
841: followed two criteria: a wide coverage in galaxy mass and availability of
842: spectra of very high S/N. The galaxy sample is composed of M\,32
843: \citep{1985AJ.....90.1927R} and six elliptical galaxies of Virgo, selected
844: along the Color-Magnitude Relation of this cluster \citep{2001ApJ...551L.127V}.
845: The long-slit spectra for all the galaxies have S/N (per \AA) above 150. In
846: Figures 8 and \ref{figure9}, we show H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ as
847: a function of various metallicity indicators: [MgFe], Fe$_{3}$, Mg$_{\rm{b}}$,
848: Ca4227 and CN$_{2}$. Galaxies with similar velocity dispersions were grouped
849: separately: $\sigma_{group}$ $\approx$ 135, 180, 225 km\,s$^{-1}$. To allow
850: direct comparations between the galaxies having similar velocity dispersions,
851: some small $\sigma$ corrections were applied: the galaxies with
852: $\sigma_{total}[=(\sigma^{2}_{galaxy}+\sigma^{2}_{instr.})^{1/2}] <
853: \sigma_{group}$ were convolved with the appropiate Gaussian to reach the
854: corresponding $\sigma_{group}$.
855:
856: It is commonly used the H$_{\beta_{LICK}}$ versus [MgFe] to determine both the
857: age and the total metallicity, as the latter has been shown to be rather
858: insensitive to possible non scaled solar element ratios which are found in
859: massive ellipticals \citep{2004MNRAS.351L..19T}. However as H$_{\beta_{LICK}}$
860: has some sensitivity to metallicity the age estimates depend on the metallicity
861: indicator in use, i.e. younger for a Magnesium dominated index and older for an
862: Iron dominated index. This problem is usually alleviated through an iterative
863: process with the aid of models that specifically take into account the non
864: scaled-solar ratios (e.g. \cite{2004MNRAS.353..917T, 1998yCat..21160001T,
865: 2003MNRAS.339..897T}). This is no longer a problem if we use H$_{\beta_{o}}$
866: index as it can be seen in Figure \ref{figure9}. This figure shows that the
867: derived ages are consistent irrespectivly of the metallicity indicator in use.
868: Table \ref{table2} lists the H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ age
869: estimates for M\,32 and the Virgo galaxies as derived from all the diagramas in
870: \ref{figure8} and \ref{figure9}.
871:
872: Our H$_{\beta_{o}}$ age estimates are consistent within the error bars with
873: those derived by \cite{1985AJ.....90.1927R} for M\,32 and
874: \cite{2001ApJ...551L.127V} and \cite{2006ApJ...637..200Y} for the Virgo sample
875: using H$_{\gamma_{\sigma}}$ high S/N requirement set of indices. This sample
876: of high quality spectra confirms H$_{\beta_{o}}$ as an advantageous age-dating
877: indicator. A more detailed analysis of the ages and metallicities of these
878: galaxies has already been performed by these authors.
879:
880: \begin{figure*}
881: \includegraphics[angle=0]{jlcervantes_fig7a.eps}
882: \caption{[Mg/Fe], Fe$_{3}$, Mg$_{b}$, Ca4227 and CN2 vs. H$_{\beta_{LICK}}$.
883: Top, middle and bottom panels are for galaxies with $\sigma_{group}$ = 135, 180,
884: and 300 ${\rm km\,s^{-1}}$, respectively. Model grids with various ages (dotted
885: lines) and metallicities (solid lines) are overplotted; the age increases from
886: top to bottom (5.6, 8.0, 11.2 and 17.8 \,Gyr), whereas the metallicity increases
887: from left to right ([Fe/H]=-0.7, -0.4, 0.0 and +0.2).}
888: \label{figure8}
889: \end{figure*}
890:
891: \begin{figure*}
892: \includegraphics[angle=0]{jlcervantes_fig7.eps}
893: \caption{Same as Figure \label{fig8} but for H$_{\beta_{o}}$.}
894: \label{figure9}
895: \end{figure*}
896:
897: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
898:
899: \section{Conclusions}
900:
901: We have defined a new spectroscopic age indicator, H$_{\beta_{o}}$, which has
902: been optimized for disentangling stellar cluster and galaxy ages. This index
903: has a larger ability for lifting the age-metallicity degeneracy than the
904: standard index H$_{\beta_{LICK}}$. To achieve this, we have employed the
905: evolutionary stellar population synthesis model of \cite{1999ApJ...513..224V}
906: and its recent extension \cite{vazdekis} based on MILES stellar spectral
907: library (S\'anchez-Bl\'azquez et al. 2006; Cenarro et al. 2007). As these
908: models provide full spectra at moderately high resolution for stellar
909: populations of different ages and metallicities, it is straighforward to
910: investigate the behaviour of prospective index definitions as a function of
911: relevant parameters by measuring the indices directly on the SSP spectra. This
912: avoids us going through an intermediate step that requires the parametrization
913: of each index definition as a function of the stellar atmospheric parameters to
914: compute the integrated index, as it has been done, for example, for the
915: Lick/IDS system of indices. The latter approach is not functional for
916: optimizing trial index definitions as a function of age or metallicity, or
917: other effects such as velocity dispersion.
918:
919: We have shown that the stronger age disentangling power of H$_{\beta_{o}}$ is
920: achieved by avoiding the metallic lines of the red-pseudocontinuum of
921: H$_{\beta_{LICK}}$. The main characteristics and uncertainties affecting
922: H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$ indices have been studied in detail. We
923: find that H$_{\beta_{o}}$ has a slightly higher velocity dispersion sensitivity
924: than H$_{\beta_{LICK}}$, but this effect is negligible in comparison to that
925: from photon noise. The S/N required to measure H$_{\beta_{o}}$ is not
926: significantly higher than that for H$_{\beta_{LICK}}$, and much lower than that
927: needed for applying H$\gamma_{\sigma}$ set of age indicators of
928: \cite{1999ApJ...525..144V}. We have fully characterized the behaviour of
929: H$_{\beta_{o}}$ with wavelengths shifts, spectrum shape and dust extinction. We
930: find that none of these effects are particularly relevant. We also have studied
931: the effects of the [$\alpha$/Fe] enhancement on H$_{\beta_{o}}$ making use of
932: the theoretical stellar spectral libraries of \cite{2005A&A...443..735C} and
933: \cite{2005A&A...442.1127M} following the approach of \cite{2007IAUS..241..167C}.
934: We find a greater sensitivity of H$_{\beta_{o}}$ to [$\alpha$/Fe] in
935: comparison to that obtained for H$_{\beta_{LICK}}$ when the library of
936: \cite{2005A&A...443..735C} is employed. Furthermore the relative variation of
937: H$_{\beta_{o}}$ to [$\alpha$/Fe] is very similar to that of H$_{\beta_{LICK}}$
938: to the total metallicity, but this result must be taken with caution as the
939: opposite trend is obtained when employing an alternative stellar library
940: (\cite{2005A&A...442.1127M}). The sensitivities of H$_{\beta_{o}}$ and
941: H$_{\beta_{LICK}}$ to [$\alpha$/Fe] variations are smaller than those found
942: for the \cite{1997ApJS..111..377W} higher order Balmer index definitions.
943: Interestingly the H$_{\gamma_{\sigma}}$ indices of \cite{1999ApJ...525..144V}
944: and \cite{2001ApJ...549..274V} show negligible sensitivity to [$\alpha$/Fe].
945:
946: We have analyzed the Milky Way globular cluster spectra of
947: \cite{2005ApJS..160..163S} to test H$_{\beta_{o}}$. The plots of their
948: CMD-derived metallicities against H$_{\beta_{LICK}}$ and H$_{\beta_{o}}$
949: indices, show how the metallicity sensitivity has been decreased significantly
950: for latter. The comparison of the observed values and the model grids resulting
951: from plotting H$_{\beta_{o}}$ versus the [MgFe] metallicity indicator provide
952: very old ages, in good agreement with the results obtained from various Balmer
953: age indicators, when employing scaled solar SSP models. However the obtained
954: ages are older than the CMD-derived ages, confirming a model zero-point problem
955: that might be affecting the analyses of the integrated light based on the Balmer
956: indices (e.g. \cite{1999AJ....118.1268G, 2001ApJ...549..274V,
957: 2002ApJ...580..850S, 2007MNRAS.379.1618M}).
958:
959: We also probe the reliability of H$_{\beta_{o}}$ to obtain mean luminosity
960: weighted ages of early-type galaxies. For this purpose we used a sample of
961: ellipticals covering a wide range in mass and $\alpha$-enhancement, for which
962: spectra of extremely high quality are available
963: \citep{2005AJ....129..712R,2001ApJ...551L.127V}. Unlike with H$_{\beta_{LICK}}$,
964: the ages inferred from plotting H$_{\beta_{o}}$ versus various metallicity
965: indicators and scaled-solar model grids are consistent irrespective of the
966: metallicity indicator in use. This also applies to the more massive galaxies
967: with larger [Mg/Fe] values. We also find that the H$_{\beta_{o}}$ ages are in
968: good agreement with the values obtained from the very high S/N requirement
969: H$\gamma_{\sigma}$ set of indices of \cite{1999ApJ...525..144V} as listed in
970: \cite{2006ApJ...637..200Y} on the basis of the same SSP models.
971:
972: The results and plots shown here might be very usefull, and could be taken as a
973: guide, for preparing and optimizing observations for those willing to use
974: H$_{\beta_{o}}$ and H$_{\beta_{LICK}}$ indices in their analyses.
975:
976:
977: \begin{table*}
978: \centering
979: \begin{minipage}{140mm}
980: \caption{AGE ESTIMATES FOR VIRGO ELLIPTICAL GALAXIES} \label{table5}
981: \begin{tabular}{@{}rrrrrrrrrrrr@{}}
982: \hline
983: {} & {Age$^{a}$ } &&&&&{Age$^{b}$}&&&&& {Age$^{c}$}\\
984: {Galaxy} & {[MgFe]} & {Fe$_{3}$} &{Mg$_{b}$} &{Ca4227} &{CN2} &
985: {[MgFe]} & {Fe$_{3}$} &{Mg$_{b}$} &{Ca4227} &{CN2} & \\
986: \hline
987: M32&3.8$^{-0.4}_{+1.2}$&3.4&4.5&3.8&3.6&3.1$^{-0.2}_{+0.3}$&3.0&3.1&3.1&3.0& \\
988: &&&&&&&&&&&\\
989: NGC4387&12.9$^{-1.4}_{+1.8}$&13.4&11.1&15.5&13.4& 14.5$^{-1.9}_{+1.9}$&14.5&14.8&14.3&14.5& 12.9$^{-2.2}_{+5.5}$\\
990: &&&&&&&&&&&\\
991: NGC4464&15.3$^{-1.8}_{+2.4}$&$>$17.8&10.5&$>$17.8&$>$17.8&$>$17.8&$>$17.8&$>$17.8&14.3& $>$ 17.8 &18.5$^{-1.6}_{+11.5}$\\
992: &&&&&&&&&&&\\
993: NGC4478&5.8$^{-1.0}_{+1.2}$&7.2&3.7&7.8&5.5& 8.2$^{-1.3}_{+1.3}$&8.7&7.4&8.7&8.2& 8.6$^{-1.3}_{+2.4}$\\
994: &&&&&&&&&&&\\
995: NGC4473&10.0$^{-0.8}_{+0.4}$&12.3&6.0&17.2&4.6& 11.1$^{-0.5}_{+0.8}$&13.2&10.0&13.2&10.1& 9.9$^{-1.4}_{+1.9}$\\
996: &&&&&&&&&&&\\
997: NGC4365&11.1$^{-0.4}_{+1.1}$&17.0&8.3&$>$17.8&10.4& 13.8$^{-0.9}_{+1.2}$&$>$17.8&10.6&$>$17.8&11.8&20.0$^{-7.2}_{+10.0}$\\
998: &&&&&&&&&&&\\
999: NGC4621&10.5$^{-0.2}_{+0.3}$&16.2&7.0&$>$17.8&8.4& 10.6$^{-0.2}_{+0.3}$&14.3&9.5&14.1&9.9& 10.6$^{-0.3}_{+0.4}$\\
1000: &&&&&&&&&&&\\
1001: \hline
1002:
1003: \end{tabular}
1004: \end{minipage}
1005: \end{table*}
1006:
1007: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1008:
1009: \section*{Acknowledgments} The authors would like to thank P. Coelho and B.
1010: Barbuy for their help for implementing their stellar library, N. Cardiel for his
1011: help for simulating error computations and M. Beasley, J. Cenarro and J.
1012: Falc\'on-Barroso for very useful suggestions and discussions. The authors thank
1013: the referee for relevant suggestions that improved the original version of the
1014: paper. JLC is a FPU PhD student and AV is a Ram\'on y Cajal Fellow of the
1015: Spanish Ministry of Education and Science. This work has been supported by the
1016: Spanish Ministry of Education and Science grants \emph{AYA2004-03059},
1017: \,\emph{AYA2005-04149} and \emph{AYA2007-67752-C03-01}.
1018:
1019: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1020:
1021: \begin{thebibliography}{99}
1022:
1023: \bibitem[\protect\citeauthoryear{Arimoto \&
1024: Yoshii}{1986}]{1986A&A...164..260A} Arimoto N., Yoshii Y., 1986, A\&A, 164,
1025: 260
1026:
1027: %\bibitem[\protect\citeauthoryear{Bower, Lucey, \&
1028: %Ellis}{1992}]{1992MNRAS.254..601B} Bower R.~G., Lucey J.~R., Ellis R.~S.,
1029: %1992, MNRAS, 254, 601
1030:
1031: %\bibitem[\protect\citeauthoryear{Bower, Lucey, \&
1032: %Ellis}{1992}]{1992MNRAS.254..589B} Bower R.~G., Lucey J.~R., Ellis R.~S.,
1033: %1992, MNRAS, 254, 589
1034:
1035: \bibitem[\protect\citeauthoryear{Burstein et
1036: al.}{1984}]{1984ApJ...287..586B} Burstein D., Faber S.~M., Gaskell C.~M.,
1037: Krumm N., 1984, ApJ, 287, 586
1038:
1039: \bibitem[\protect\citeauthoryear{Bruzual A.}{1983}]{1983ApJ...273..105B}
1040: Bruzual A.~G., 1983, ApJ, 273, 105
1041:
1042: \bibitem[\protect\citeauthoryear{Cardiel}{1999}]{1999PhDT........12C}
1043: Cardiel N., 1999, PhDT,
1044:
1045: \bibitem[\protect\citeauthoryear{Cardiel et
1046: al.}{2003}]{2003A&A...409..511C} Cardiel N., Gorgas J.,
1047: S{\'a}nchez-Bl{\'a}zquez P., Cenarro A.~J., Pedraz S., Bruzual G., Klement
1048: J., 2003, A\&A, 409, 511
1049:
1050: \bibitem[\protect\citeauthoryear{Cenarro et
1051: al.}{2001}]{2001MNRAS.326..959C} Cenarro A.~J., Cardiel N., Gorgas J.,
1052: Peletier R.~F., Vazdekis A., Prada F., 2001, MNRAS, 326, 959
1053:
1054: \bibitem[\protect\citeauthoryear{Cenarro et
1055: al.}{2007}]{2007MNRAS.374..664C} Cenarro A.~J., et al., 2007, MNRAS, 374, 664
1056:
1057: \bibitem[\protect\citeauthoryear{Cenarro et
1058: al.}{2007b}]{cenarrosb} Cenarro A.~J., Cervantes J.~L., Beasley
1059: M., Mar{\'\i}n A., Vazdekis A., 2008, ApJL, accepted
1060:
1061: \bibitem[\protect\citeauthoryear{Cervantes et
1062: al.}{2007a}]{2007IAUS..241..167C} Cervantes J.~L., Coelho P., Barbuy B.,
1063: Vazdekis A., 2007, IAUS, 241, 167
1064:
1065: \bibitem[\protect\citeauthoryear{Coelho et al.}{2007}]{2007MNRAS.382..498C}
1066: Coelho P., Bruzual G., Charlot S., Weiss A., Barbuy B., Ferguson J., 2007,
1067: MNRAS, 382, 498
1068:
1069: \bibitem[\protect\citeauthoryear{Coelho et al.}{2005}]{2005A&A...443..735C}
1070: Coelho P., Barbuy B., Mel{\'e}ndez J., Schiavon R.~P., Castilho B.~V.,
1071: 2005, A\&A, 443, 735
1072:
1073: \bibitem[\protect\citeauthoryear{Charlot \&
1074: Fall}{2000}]{2000ApJ...539..718C} Charlot S., Fall S.~M., 2000, ApJ, 539,
1075: 718
1076:
1077: \bibitem[\protect\citeauthoryear{de Lorenzo-C{\'a}ceres, Vazdekis, \&
1078: Aguerri}{2007}]{2007IAUS..241..420D} de Lorenzo-C{\'a}ceres A., Vazdekis
1079: A., Aguerri J.~A.~L., 2007, IAUS, 241, 420
1080:
1081: \bibitem[\protect\citeauthoryear{Gibson et al.}{1999}]{1999AJ....118.1268G}
1082: Gibson B.~K., Madgwick D.~S., Jones L.~A., Da Costa G.~S., Norris J.~E.,
1083: 1999, AJ, 118, 1268
1084:
1085: \bibitem[\protect\citeauthoryear{Gonzalez}{1993}]{1993PhDT.........7G}
1086: Gonzalez G., 1993, PhDT,
1087:
1088: \bibitem[\protect\citeauthoryear{Gorgas, Jablonka, \&
1089: Goudfrooij}{2007}]{2007A&A...474.1081G} Gorgas J., Jablonka P., Goudfrooij
1090: P., 2007, A\&A, 474, 1081
1091:
1092: \bibitem[\protect\citeauthoryear{Gorgas et al.}{1993}]{1993ApJS...86..153G}
1093: Gorgas J., Faber S.~M., Burstein D., Gonzalez J.~J., Courteau S., Prosser
1094: C., 1993, ApJS, 86, 153
1095:
1096: \bibitem[\protect\citeauthoryear{Grevesse \&
1097: Sauval}{1998}]{1998SSRv...85..161G} Grevesse N., Sauval A.~J., 1998, SSRv,
1098: 85, 161
1099:
1100: \bibitem[\protect\citeauthoryear{Harris}{1996}]{1996AJ....112.1487H} Harris
1101: W.~E., 1996, AJ, 112, 1487
1102: (February 2003 version on line: http://coihue.rutgers.edu/~andresj/gccat.html)
1103:
1104: \bibitem[\protect\citeauthoryear{Jones \&
1105: Worthey}{1995}]{1995ApJ...446L..31J} Jones L.~A., Worthey G., 1995, ApJ,
1106: 446, L31
1107:
1108: \bibitem[\protect\citeauthoryear{Korn, Maraston, \&
1109: Thomas}{2005}]{2005A&A...438..685K} Korn A.~J., Maraston C., Thomas D.,
1110: 2005, A\&A, 438, 685
1111:
1112: %\bibitem[\protect\citeauthoryear{Lee, Yoon, \&
1113: %Lee}{2000}]{2000AJ....120..998L} Lee H.-c., Yoon S.-J., Lee Y.-W., 2000,
1114: %AJ, 120, 998
1115: \bibitem[\protect\citeauthoryear{Lee \&
1116: Worthey}{2005}]{2005ApJS..160..176L} Lee H.-C., Worthey G., 2005, ApJS,
1117: 160, 176
1118:
1119: \bibitem[\protect\citeauthoryear{MacArthur}{2005}]{2005ApJ...623..795M}
1120: MacArthur L.~A., 2005, ApJ, 623, 795
1121:
1122: %\bibitem[\protect\citeauthoryear{Maraston \&
1123: %Thomas}{2000}]{2000ApJ...541..126M} Maraston C., Thomas D., 2000, ApJ, 541,
1124: %126
1125:
1126: \bibitem[\protect\citeauthoryear{Martins \&
1127: Coelho}{2007}]{2007MNRAS.381.1329M} Martins L., Coelho P., 2007, MNRAS, 381, 1329
1128:
1129: \bibitem[\protect\citeauthoryear{Mendel, Proctor, \&
1130: Forbes}{2007}]{2007MNRAS.379.1618M} Mendel J.~T., Proctor R.~N., Forbes
1131: D.~A., 2007, MNRAS, 379, 1618
1132:
1133: \bibitem[\protect\citeauthoryear{Munari et al.}{2005}]{2005A&A...442.1127M}
1134: Munari U., Sordo R., Castelli F., Zwitter T., 2005, A\&A, 442, 1127
1135:
1136: %\bibitem[\protect\citeauthoryear{Oconnell}{1976}]{1976ApJ...206..370O}
1137: %Oconnell R.~W., 1976, ApJ, 206, 370
1138:
1139: %\bibitem[\protect\citeauthoryear{Oconnell}{1980}]{1980ApJ...236..430O}
1140: %Oconnell R.~W., 1980, ApJ, 236, 430
1141:
1142: \bibitem[\protect\citeauthoryear{Puzia et al.}{2002}]{2002A&A...395...45P}
1143: Puzia T.~H., Saglia R.~P., Kissler-Patig M., Maraston C., Greggio L.,
1144: Renzini A., Ortolani S., 2002, A\&A, 395, 45
1145:
1146: %\bibitem[\protect\citeauthoryear{Recio-Blanco et
1147: %al.}{2006}]{2006A&A...452..875R} Recio-Blanco A., Aparicio A., Piotto G.,
1148: %de Angeli F., Djorgovski S.~G., 2006, A\&A, 452, 875
1149:
1150: %\bibitem[\protect\citeauthoryear{P{\'e}rez, S{\'a}nchez-Bl{\'a}zquez, \&
1151: %Zurita}{2007}]{2007A&A...465L...9P} P{\'e}rez I., S{\'a}nchez-Bl{\'a}zquez
1152: %P., Zurita A., 2007, A\&A, 465, L9
1153:
1154: \bibitem[\protect\citeauthoryear{Rose}{1985}]{1985AJ.....90.1927R} Rose
1155: J.~A., 1985, AJ, 90, 1927
1156:
1157: \bibitem[\protect\citeauthoryear{Rose et al.}{2005}]{2005AJ....129..712R}
1158: Rose J.~A., Arimoto N., Caldwell N., Schiavon R.~P., Vazdekis A., Yamada
1159: Y., 2005, AJ, 129, 712
1160:
1161: %\bibitem[\protect\citeauthoryear{Salaris \&
1162: %Weiss}{2002}]{2002A&A...38 M., Weiss A., 2002, A\&A, 388,
1163: %492
1164:
1165: \bibitem[\protect\citeauthoryear{S{\'a}nchez-Bl{\'a}zquez et
1166: al.}{2006}]{2006MNRAS.371..703S} S{\'a}nchez-Bl{\'a}zquez P., et al., 2006,
1167: MNRAS, 371, 703
1168:
1169: \bibitem[\protect\citeauthoryear{Sarzi et al.}{2006}]{2006MNRAS.366.1151S}
1170: Sarzi M., et al., 2006, MNRAS, 366, 1151
1171:
1172: \bibitem[\protect\citeauthoryear{Schiavon et
1173: al.}{2002}]{2002ApJ...580..850S} Schiavon R.~P., Faber S.~M., Castilho
1174: B.~V., Rose J.~A., 2002, ApJ, 580, 850
1175:
1176: \bibitem[\protect\citeauthoryear{Schiavon et
1177: al.}{2005}]{2005ApJS..160..163S} Schiavon R.~P., Rose J.~A., Courteau S.,
1178: MacArthur L.~A., 2005, ApJS, 160, 163
1179:
1180: \bibitem[\protect\citeauthoryear{Tantalo \&
1181: Chiosi}{2004}]{2004MNRAS.353..917T} Tantalo R., Chiosi C., 2004, MNRAS,
1182: 353, 917
1183:
1184: \bibitem[\protect\citeauthoryear{Thomas, Maraston, \&
1185: Bender}{2003}]{2003MNRAS.339..897T} Thomas D., Maraston C., Bender R.,
1186: 2003, MNRAS, 339, 897
1187:
1188: \bibitem[\protect\citeauthoryear{Thomas, Maraston, \&
1189: Korn}{2004}]{2004MNRAS.351L..19T} Thomas D., Maraston C., Korn A., 2004,
1190: MNRAS, 351, L19
1191:
1192: %\bibitem[\protect\citeauthoryear{Trager et al.}{1998}]{1998yCat..21160001T}
1193: %Trager S.~C., Worthey G., Faber S.~M., Burstein D., Gonzalez J.~J., 1998,
1194: %yCat, 211, 60001
1195:
1196: \bibitem[\protect\citeauthoryear{Tripicco \&
1197: Bell}{1995}]{1995AJ....110.3035T} Tripicco M.~J., Bell R.~A., 1995, AJ,
1198: 110, 3035
1199:
1200: \bibitem[\protect\citeauthoryear{Vazdekis}{1999}]{1999ApJ...513..224V}
1201: Vazdekis A., 1999, ApJ, 513, 224
1202:
1203: \bibitem[\protect\citeauthoryear{Vazdekis \&
1204: Arimoto}{1999}]{1999ApJ...525..144V} Vazdekis A., Arimoto N., 1999, ApJ,
1205: 525, 144
1206:
1207: \bibitem[\protect\citeauthoryear{Vazdekis et
1208: al.}{2001a}]{2001ApJ...549..274V} Vazdekis A., Salaris M., Arimoto N., Rose
1209: J.~A., 2001, ApJ, 549, 274
1210:
1211: \bibitem[\protect\citeauthoryear{Vazdekis et
1212: al.}{2001b}]{2001ApJ...551L.127V} Vazdekis A., Kuntschner H., Davies R.~L.,
1213: Arimoto N., Nakamura O., Peletier R., 2001, ApJ, 551, L127
1214:
1215: \bibitem[\protect\citeauthoryear{Vazdekis et
1216: al.}{2008 (in preparation)}]{vazdekis} Vazdekis A., et al., 2008, (in preparation)
1217:
1218: \bibitem[\protect\citeauthoryear{Worthey}{1994}]{1994ApJS...95..107W}
1219: Worthey G., 1994, ApJS, 95, 107
1220:
1221: \bibitem[\protect\citeauthoryear{Worthey et
1222: al.}{1994}]{1994ApJS...94..687W} Worthey G., Faber S.~M., Gonzalez J.~J.,
1223: Burstein D., 1994, ApJS, 94, 687
1224:
1225: \bibitem[\protect\citeauthoryear{Worthey \&
1226: Ottaviani}{1997}]{1997ApJS..111..377W} Worthey G., Ottaviani D.~L., 1997,
1227: ApJS, 111, 377
1228:
1229: \bibitem[\protect\citeauthoryear{Yamada et al.}{2006}]{2006ApJ...637..200Y}
1230: Yamada Y., Arimoto N., Vazdekis A., Peletier R.~F., 2006, ApJ, 637, 200
1231:
1232: %\bibitem[\protect\citeauthoryear{York et al.}{2000}]{2000AJ....120.1579Y}
1233: %York D.~G., et al., 2000, AJ, 120, 1579
1234:
1235: \end{thebibliography}
1236: \appendix
1237: \section[]{INCREASING THE AGE SENSITIVITY OF H$_{\beta_{LICK}}$ }
1238:
1239: As a consecuence of the characterization of the standard H$_{\beta_{LICK}}$
1240: index definition as a function of wavelength shifts (see section 4.2),
1241: we have shown how the H$_{\beta_{LICK}}$ index
1242: increases its ability to disentagle ages if all the bandpasses
1243: are shifted 1.7\,{\AA} blueward. In this apendix we provide the main
1244: properties of this alternative index definition and show the most important
1245: differences with respect to the standard, i.e. not shifted, H$_{\beta_{LICK}}$
1246: index definition.
1247:
1248: Table \ref{aptable1} lists the limiting wavelegth of the optimized version of
1249: this index definition (hereafter H$_{\beta_{LICK,o}}$).
1250: The main characteristics are compiled in Table \ref{aptable2}.
1251: H$_{\beta_{LICK,o}}$ shows a slightly larger age disentangling power than
1252: H$_{\beta_{LICK}}$ (see Table 2). On the other hand, H$_{\beta_{LICK,o}}$ is
1253: less stable against velocity dispersion than H$_{\beta_{LICK}}$ is. Note,
1254: however, that this dependance, tends to flatten for $\sigma {\rm >
1255: 150\,km\,s^{-1}}$ (see Figure
1256: \ref{apfigure1}). Figure \ref{apfigure2} shows that the slightly larger age
1257: resolving power is maintained as function of $\sigma$.
1258:
1259: Figure \ref{apfigure3} shows an age/metallicity diagnostic diagram based on
1260: H$_{\beta_{LICK,o}}$. The plot shows similar age values as those obtained with
1261: which provides similar ages as those obtained with H$_{\beta_{LICK}}$.
1262:
1263: \begin{table*}
1264: \centering
1265: \begin{minipage}{140mm}
1266: \caption{H$_{\beta_{LICK}}$ INDEX CONFIGURATION OF BANDPASSES }\label{aptable1}
1267: \begin{tabular}{@{}lrrrrr@{}}
1268: \hline
1269: {} & {Blue pseudocontinuum} &
1270: {Feature} &
1271: {Red pseudocontinuum}&{}&{} \\
1272: \cline{2-6} \\
1273: {Index} & {\AA} & {\AA} &
1274: {\AA}&{c$_{1}$}& {c$_{2}$} \\
1275: \hline
1276: H$_{\beta_{LICK,o}}$& 4826.175 4846.175 & 4846.175 4874.925& 4874.925
1277: 4889.925&8.590& 0.2439\\
1278:
1279: \hline
1280: \end{tabular}
1281: \end{minipage}
1282: \end{table*}
1283:
1284: \begin{table*}
1285: \centering
1286: \begin{minipage}{140mm}
1287: \caption{INDEX CHARACTERISTICS AND UNCERTAINTIES} \label{aptable2}
1288:
1289: \begin{tabular}{@{}lr@{}}
1290: \hline
1291: {Parameter} &
1292: {H$_{\beta_{LICK,o}}$} \\
1293: \hline
1294: $\frac{\beta}{\alpha}$ & 0.3\\
1295: \\
1296: $\Sigma$ ($10^{-4}$ ${\rm km\,s^{-1}}$)&2.9\\
1297: Index variation by ${\rm \pm 10\,km\,s^{-1}}$ in $\sigma$ (\%) & 0.29 \\
1298: Maximum stability $\sigma$ range& ${\rm \pm 100\,km\,s^{-1}}$\\
1299: \\
1300: Age uncertainty caused by a $\Delta\lambda$ = 1.5 {\AA} &5.7-18
1301: \,Gyr\\
1302: Maximum $\Delta \lambda$ shifts ($\lambda$, z, rotation curve) compatible with a $+/-$ 2.5 \,Gyr error ({\AA}) & 0.85\\
1303: \\
1304: (H$_{\beta}$(flux calibrated response curve) - H$_{\beta}$(continuum
1305: removed))/H$_{\beta}$(f.c.r.c.) & $\lq$ 0.01 \\
1306: \\
1307: S/N (per\,{\AA}) required to distinguish 2.5 \,Gyr at 10 \,Gyr & $\sim$ 50\\
1308: Minimum age uncertainity for S/N (per\,{\AA}) $>$ 250 & 4.9 \,Gyr \\
1309: \\
1310: Index Sensitivity to [$\alpha$/Fe] variations ([$\Lambda_{CO}$, $\Lambda_{MU}$]) & [0.12,0.23]\\
1311:
1312: \hline
1313:
1314: \end{tabular}
1315: \end{minipage}
1316: \end{table*}
1317:
1318: \begin{figure}
1319: \includegraphics[angle=0]{jlcervantes_ap_fig1.eps}
1320: \caption{Evolution of H$_{\beta_{LICK,o}}$ versus spectral resolution
1321: ($\sigma$). The line represents the mean values for SSPs older than 1\,Gyr.}
1322: \label{apfigure1}
1323: \end{figure}
1324:
1325: \begin{figure}
1326: \includegraphics[angle=0]{jlcervantes_ap_fig2.eps}
1327: \caption{Generalized Worthey's parameter, $\beta/\alpha$, versus model spectral
1328: resolution ($\sigma$). }
1329: \label{apfigure2}
1330: \end{figure}
1331:
1332: \begin{figure}
1333: \includegraphics[angle=0]{jlcervantes_ap_fig3.eps}
1334: \caption{[Mg/Fe] vs. H$_{\beta_{LICK,o}}$. Top, middle and bottom panels are for
1335: galaxies with $\sigma_{group}$ = 135, 180, and 300 ${\rm km\,s^{-1}}$,
1336: respectively. Model grids with various ages (dotted lines) and metallicities
1337: (solid lines) are overplotted; the age increases from top to bottom (5.6, 8.0,
1338: 11.2 and 17.8 \,Gyr), whereas the metallicity increases from left to right
1339: ([Fe/H]=-0.7, -0.4, 0.0 and +0.2).}
1340: \label{apfigure3}
1341: \end{figure}
1342:
1343: \bsp
1344:
1345: \label{lastpage}
1346:
1347: \end{document}
1348: