1: %\documentclass[12pt,preprint]{aastex}
2:
3: %% manuscript produces a one-column, double-spaced document:
4:
5: %\documentclass[manuscript]{aastex}
6:
7: %% preprint2 produces a double-column, single-spaced document:
8:
9: \documentclass[preprint2]{emulateapj}
10: %\usepackage{natbib}
11:
12: \newcommand{\vdag}{(v)^\dagger}
13: \newcommand{\myemail}{etj@astrax.fis.ucm.es}
14: \newcommand{\reduceme}{\mbox{R\raisebox{-0.35ex}{E}D%
15: \hspace{-0.05em}\raisebox{0.85ex}{uc}\hspace{-0.90em}%
16: \raisebox{-.35ex}{{m}}\hspace{0.05em}E}}
17:
18: %% You can insert a short comment on the title page using the command below.
19: %\slugcomment{Not to appear in Nonlearned J., 45.}
20:
21: \shorttitle{Carbon and Nitrogen in Early-type Galaxies}
22: \shortauthors{Toloba et~al.}
23:
24: \begin{document}
25:
26: %% LaTeX will automatically break titles if they run longer than
27: %% one line. However, you may use \\ to force a line break if
28: %% you desire.
29:
30: \title{Carbon and Nitrogen Abundances in Early-type Galaxies}
31:
32: %% Use \author, \affil, and the \and command to format
33: %% author and affiliation information.
34: %% Note that \email has replaced the old \authoremail command
35: %% from AASTeX v4.0. You can use \email to mark an email address
36: %% anywhere in the paper, not just in the front matter.
37: %% As in the title, use \\ to force line breaks.
38:
39: \author{E. Toloba\altaffilmark{1}}
40: %\email{etj@astrax.fis.ucm.es}
41: \author{P. S\'{a}nchez-Bl\'{a}zquez\altaffilmark{2}}
42: %\email{psanchez-blazquez@uclan.ac.uk}
43: \author{J. Gorgas\altaffilmark{1}}
44: %\email{fjg@astrax.fis.ucm.es}
45: \author{B.~K. Gibson\altaffilmark{2}}
46: %\email{bkgibson@uclan.ac.uk}
47: \altaffiltext{1}{Universidad Complutense de Madrid, 28040, Madrid, Spain}
48: \altaffiltext{2}{Centre for Astrophysics, University of Central Lancashire, Preston, PR1~2HE, UK}
49:
50: \begin{abstract} For the first time, we undertake a systematic
51: examination of the nitrogen abundances for a sample of 35 early-type
52: galaxies spanning a range of masses and local environment. The
53: nitrogen-sensitive molecular feature at 3360\AA\ has been employed in
54: conjunction with a suite of atomic- and molecular-sensitive indices to
55: provide unique and definitive constraints on the chemical content of
56: these systems. By employing NH3360, we are now able to break the
57: carbon, nitrogen, and oxygen degeneracies inherent to the use of the
58: CN-index. We demonstrate that the NH3360 feature shows little
59: dependency upon the velocity dispersion (our proxy for mass) of the
60: galaxies, contrary to what is seen for carbon- and magnesium-sensitive
61: indices. At face value, these results are at odds with conclusions
62: drawn previously using indices sensitive to both carbon and nitrogen,
63: such as cyanogen (CN). With the aid of stellar population models, we
64: find that the N/Fe ratios in these galaxies are consistent with being
65: mildly-enhanced with respect to the solar ratio. We also explore the
66: dependence of these findings upon environment, by analyzing the
67: co-added spectra of galaxies in the field and the Coma cluster. We
68: confirm the previously found differences in carbon abundances between
69: galaxies in low- and high-density environments, while showing that
70: these differences do not seem to exist for nitrogen. We discuss the
71: implications of these findings for the derivation of the star
72: formation histories in early-type galaxies, and for the origin of
73: carbon and nitrogen, themselves.
74: \end{abstract}
75:
76: \keywords{galaxies: abundances --- galaxies: elliptical and lenticular, cD ---
77: galaxies: evolution --- galaxies: formation --- galaxies: stellar content}
78:
79: \section{Introduction}
80:
81: Uncovering the star formation histories of early-type galaxies has
82: proved to be a challenging endeavor for several decades now. Perhaps
83: the most common approach to date has been via the comparison of
84: Lick/IDS spectral indices, each with varying degrees of sensitivity to
85: age and metallicity \citep[][]{Wor94}. Alas, the lack of indices
86: which are \it purely \rm sensitive to age and/or metallicity has
87: limited the uniqueness of the claimed results. Despite a rich
88: literature, there remains fundamental disagreement on the magnitude
89: and even veracity of important trends, including those linking age,
90: metallicity, and relative elemental abundances, with mass and
91: environment. A further limitation to this technique is its
92: sensitivity to the youngest stellar component in a given galaxy, as
93: opposed to the more representative underlying population.
94:
95:
96: An alternative approach, and one employed successfully for our own
97: Galaxy, is to use our knowledge of galactic chemical evolution to
98: inform our derivation of star formation histories. In massive
99: early-type galaxies \citet{Oconnel76} first noticed that Mg abundance
100: was enhanced relative to Fe; Other works confirmed this by comparing
101: absorption line indices with isochrone-based stellar population
102: models. Because Mg is predominantly produced in Type~II supernovae
103: while Fe production is dominated by Type~Ia supernovae, this
104: enhancement is often interpreted as the result of very short
105: timescales for the formation of stars in these galaxies. Apart from
106: Mg, other elements (eg. C, N, Na) have been claimed to be enhanced in
107: large ellipticals (Worthey 1998\nocite{Wor98}; Burstein
108: 2003\nocite{Burs03}; Kelson et al.\ 2006\nocite{Kel06},
109: S\'anchez-Bl\'azquez et al. 2003\nocite{SB03}, 2006a\nocite{SB06a}
110: (SB03 and SB06a hereafter); Schiavon 2007\nocite{Schiav07}; Graves et
111: al. 2008\nocite{Grav08}) and to be strongly correlated with their
112: velocity dispersions \citep[although see][]{Kel06,Clem06} for an
113: alternate perspective regarding Mg and C).
114:
115: Claims have been made for a strong N/Fe correlation with mass
116: \citep[SB06a,][]{Kel06,Schiav07,Grav08}. Nitrogen is produced during
117: hydrogen burning via the CNO and CN cycles, and it is created as both,
118: primary and secondary element. In primary nucleosynthesis N is
119: produced at the same time as C and O and it is independent of
120: metallicity while in secondary production nitrogen is synthesized from
121: the carbon and oxygen already present in the star, and its abundance
122: is therefore proportional to the heavy elements abundance. The
123: correlation of N/Fe with mass have been interpreted as a correlation
124: between N/Fe and metallicity (through the mass-metallicity relation)
125: and has lead the authors above to conclude that most of the N produced
126: in ellipticals is of secondary origin. To derive nitrogen abundances,
127: the CN-band at 4175\AA~ has typically been used; however,
128: disentangling the effects of C, N, and O using the CN-band is
129: difficult \citep[][]{Burs03}. Direct observation of the NH feature at
130: 3360\AA~ offers a more direct, and robust estimator as the NH-feature
131: is insensitive to C and O \citep[][]{Sned73, Norr02} and it is directly measuring N abundances \citep[][]{B&N82,T&L84}. In addition,
132: the near-UV light is produced primarily by dwarf stars
133: \citep[][]{DC94}, while the optical cyanogen indices result from a
134: more complex mix of both dwarf and giant star light. This avoid the
135: variation in the stellar atmospheric abundances due to mixing during
136: the first dredge-up.
137:
138: There are fewer than 15 early-type galaxies with published values of
139: NH3360 in the literature \citep[][]{Pon98,Bou88,DC94}, due in large
140: part to the relative insensitivity of detectors in the near-UV.
141: We present here observations of the NH feature
142: in a sample of 35 early-type galaxies
143: in the field, Virgo, and Coma clusters.
144: We show that previous claims of correlations
145: between nitrogen and velocity dispersion are incorrect
146: and that the NH3360 index-$\sigma$ relation is almost flat.
147:
148:
149:
150: \section{Observations and Data Reduction}
151:
152: Long slit spectroscopic data for 35 ellipticals were obtained at the
153: 4.2m William Herschel Telescope at Roque de los Muchachos Observatory,
154: using the ISIS spectrograph. The instrumental configuration provided
155: and spectral coverage from 3140 to 4040\AA\ in the blue arm with an
156: spectral resolution of 2.3\AA\ (FWHM) and a typical sigmal to noise per \AA{}
157: of 40. Our sample is a subset of those
158: presented in SB06a, allowing us to supplement our near-UV data with
159: the associated optical indices (3500$-$5250\AA). We measure the NH3360
160: index in our new data, and the \nocite{Ser05}Serven, Worthey \& Briley
161: (2005, SWB05 hereafter) optical indices in our SB06a optical spectra.
162:
163: In SB03, we claimed that galaxies in the Coma cluster had much lower
164: CN and C4668 indices than those in lower-density environments. We were
165: unable then to conclude whether the differences were due exclusively
166: to differences in carbon or if nitrogen was also overabundant in
167: galaxies in lower-density environments. With the aim of resolving
168: this issue, we selected our sample to include field, Virgo and Coma
169: ellipticals, spanning a range in velocity dispersions
170: (130$<$$\sigma$$<$330~km~s$^{-1}$). Following the nomenclature of
171: SB03, we denote our Coma cluster sample as High-Density Environment
172: Galaxies (HDEG), while galaxies in the field and Virgo are denoted
173: Low-Density Environment Galaxies (LDEG).
174:
175: We have followed the standard reduction procedure for long-slit
176: spectra, using \reduceme~\citep[][]{Car99}. Galaxy spectra were
177: extracted within a central equivalent aperture of 4$^{\prime\prime}$
178: at the distance of NGC~6703 (corresponding to a physical aperture of
179: 0.76~kpc). We measured the NH3360 index using the definition of
180: \citet{DC94}\footnote{A case could be made to amend the Davidge \&
181: Clark (1994) definition somewhat, in light of Carbon et~al. (1992;
182: Fig~13). Specifically, the NH3360 continua sit within the broad
183: wings of the NH feature itself, reducing its optimal sensitivity to
184: nitrogen. Redefining the red continuum to be 3480$-$3520\AA\ would
185: mitigate this effect to some degree, but for consistency with the
186: canonical definition, we have not pursued this approach.} and the
187: CNO3862, CNO4175, and CO4685 indices, as defined by SWB05. All
188: indices were measured at a dispersion of $\sigma$=200~km~s$^{-1}$ and
189: corrected for the effects of velocity dispersion broadening using a
190: combination of synthetic spectra from Bruzual \& Charlot (2003, BC03
191: hereafter) and Vazdekis et~al. (2008, V08 hereafter).
192:
193:
194: \section{Results}
195:
196: Figure~\ref{figurel} shows the co-added spectra in different
197: wavelength regions of LDEG (blue) and HDEG (red), with velocity
198: dispersions in the range 150$<$$\sigma$$<$250~km~s$^{-1}$. Before
199: being added, the spectra were shifted to the same radial velocity and
200: broadened to the maximum $\sigma$ of all spectra (250~km~s$^{-1}$).
201: The spectral regions correspond to the bands of four different
202: indices: NH3360, CNO3862, CNO4175, and CO4685.
203:
204: Overplotted in these diagrams are the models by BC03 for the NH3360
205: and the models of V08 for the remainder of the indices. We could not
206: use V08 for the comparison of the NH index because the spectra start
207: at 3500~\AA. However, we prefer to use it for the other regions of
208: the spectra as the stellar library used in the BC03 models suffers
209: from wavelength calibration problems. Nevertheless, we have checked
210: and ensured that the results obtained are independent of the models
211: employed. All synthetic spectra were degraded to the same broadened
212: resolution as the galactic spectra. We have plotted four models in
213: each figure, with ages $\sim$6 and $15$~Gyr, and metallicities solar
214: and supersolar ([M/H]=+0.4 for BC03 and [M/H]=+0.2 for V08).
215:
216: Comparing the empirical spectra with the models in Fig.~\ref{figurel}
217: and taking into account the element dependence of the indices (see
218: inset of the figure) we can see: NH3360 is slightly overabundant with
219: respect to the scaled-solar models; CNO3862 is compatible with the
220: scaled-solar models at lower metallicities (for older ages) or higher
221: metallicities (for younger ages); CNO4175 and CO4682 are overabundant
222: in the field galaxies, but not for the cluster galaxies. In consort,
223: these data require nitrogen-based indices be overabundant with respect
224: to solar, carbon-based indices be overabundant with respect to solar
225: only for LDEG, and oxygen be similarly overabundant.
226:
227:
228: \subsection{The index-velocity dispersion relations}
229:
230: In Fig.~\ref{figurel} we present the behavior of the four indices
231: analyzed, in addition to Mgb, as a function of the velocity dispersion
232: of the galaxies. We distinguish between three different groups of
233: plots, depending on the chemical element dominating the intensity of
234: the index: N-based indices (CO3862 and NH3360), C-based indices
235: (CNO4175 and CO4685), and Mgb. Indices dependent on C show a strong
236: dependency upon velocity dispersion, in a similar way to Mgb. However,
237: the trends between the N-sensitive indices and velocity dispersion are
238: much flatter; indeed, they are even flatter than that found for
239: $<$Fe$>$ (see SB03).
240:
241:
242: Differences in the slope of the index-$\sigma$ diagram could appear
243: due to a different sensitivity of the indices to age and/or
244: metallicity and do not necessarily reflect differences in the chemical
245: abundance ratios. Figure~\ref{sensitivity} shows the variation of the
246: indices with age for different metallicities. It can be seen that each
247: of the SWB05 indices show a strong dependence on metallicity,
248: particularly CO4685. The Lick/IDS index Mgb also shows a strong
249: metallicity dependence. On the other hand, for metallicities relevant
250: for elliptical galaxies ($Z>0.004$), NH3360 is essentially constant
251: with age and overall metallicity. This behavior allows NH3360 to be
252: used successfully to derive nitrogen abundances in integrated spectra,
253: with \citet{Bou88} confirming its strong sensitivity to nitrogen
254: abundance .
255:
256: Can we explain the relations between the other indices and $\sigma$ as
257: the consequence only of a relation between the overall metallicity
258: and/or age with $\sigma$? To answer this question we have
259: parametrized the models of BC03 and V08 as a function of age and
260: metallicity, in the age and metallicity restricted to the range
261: covered by our sample, $\sim$4-$\sim$17 Gyrs and -0.70$<$[M/H]$<$0.40.
262: If we allow for the same age variation with $\sigma$ (from 5-12 Gyr in
263: the $\sigma$ range 125-350 kms$^{-1}$) for all the indices, we need
264: $\sim$0.13 dex more metallicity variation to explain the CNO4175- and
265: CO4685-$\sigma$ relations -- the most C-sensitive indices -- while and
266: extra $\sim$ 0.3 dex variation in metallicity is needed to explain the
267: relation of the Mgb-$\sigma$ slope. At a face value, this would be
268: indicating that [C/M], [Mg/M] and [Mg/C] increase with $\sigma$. The
269: conclusions, however, remain very speculative until the oxygen
270: abundance and its variation with $\sigma$ is calculated in these
271: galaxies. In addition, the indices strongly dependent on C show a
272: systematic offset between LDEG and HDEG, with the indices for HDEG
273: systematically weaker than those for LDEG. \footnote{This behavior is
274: also visible in the near-IR C-sensitive indices
275: (M\'armol-Queralt\'o, in preparation).}
276:
277:
278: \section{Discussion}
279:
280: We present here the NH3360 indices for a sample of 35 early-type
281: galaxies. Unlike the ambiguities which exist when inferring nitrogen
282: abundances from cyanogen indices in the optical (due to contamination
283: from carbon and oxygen in the bands and associated continua), the
284: near-UV molecular feature of NH isolates the nitrogen abundance. The
285: main conclusions of this work are: (1) a flat relation exists between
286: the NH3360 index and the velocity dispersion of the galaxies, contrary
287: to what has been previously claimed by other authors from their
288: analysis of the more ambiguous CN-index; (2) N is independent of local
289: environment, whereas C appears stronger for LDEG relative to HDEG. In
290: this sense, carbon and nitrogen production appear to be essentially
291: decoupled from one another.
292:
293: Our results can be useful for constraining the origin of C and N.
294: Nitrogen is produced in both intermediate-mass and massive stars, and
295: it can have a primary and secondary origin
296: \citep[][]{Carig05,Chiapp06}. The abundance ratio of a secondary
297: element is predicted to increase with metallicity. If elliptical
298: galaxies show a mass-metallicity relation, as it has been claimed by
299: many authors \citep[eg.,][]{KA98}, then we should expect the secondary
300: N to increase with the mass of the galaxies, as suggested in previous
301: works based upon cyanogen-inferred nitrogen abundance determinations
302: \citep[eg.,][]{Kel06, Schiav07}. However, we have shown here that when
303: a more ``pure'' nitrogen-sensitive indicator is used, there is no
304: relation between the nitrogen abundance and the mass of the
305: galaxies. This lack of correlation may be indicating primary
306: production of nitrogen.
307:
308:
309: Nucleosynthesis studies predict that primary N can be produced both in
310: the third dredge-up along the Asymptotic Giant Branch (AGB) phase, if
311: nuclear burning at the base of the convective envelope is efficient
312: \citep[hot-bottom burning]{RV81}, and recent yields, including
313: rotationally-induced mixing, also predict significant production of N
314: in massive stars \citep{MM02}. \citet{Chiapp03, Chiapp06} has shown
315: that the inclusion of these yields in a Galactic chemical evolution
316: model of the Milky Way predicts an N/O gradient consistent with
317: observations. Having said that, this primary nitrogen production
318: pathway is apparently only significant at low metallicities and is not
319: predicted to be particularly relevant for systems such as elliptical
320: galaxies that can reach supersolar metallicities very rapidly. In a
321: system more similar to elliptical galaxies -- that of the bulge of our
322: galaxy -- \citet{Ballero07} \citep[see also][]{Matt86} need to include
323: primary N from massive stars of {\it all} metallicities (and all
324: masses), in order too explain the data from planetary nebulae.
325:
326: In elliptical galaxies, the offsets between LDEG and HDEG favor
327: massive stars as the main contributors to primary nitrogen, because
328: for the elements produced by intermediate mass stars, no offsets are
329: detected. However, if that is the case, then we would expect a similar
330: correlation between N and $\sigma$, as seen for Mg and $\sigma$. As
331: we have shown though, the N--$\sigma$ correlation appears flat from
332: our new data and analysis. A possible solution is that the
333: contribution from {\it both} massive and intermediate mass stars is
334: shaping the N--$\sigma$ relation in ellipticals. The delay in the N
335: release from intermediate mass (4$<$M $<$8 M$_{\odot}$) stars is
336: $\sim$100~Myr, so even in a relatively short star formation burst,
337: some contribution from these stars is expected. It is believed that
338: the star formation in ellipticals is both more efficient and of a
339: shorter timescale with increasing galactic mass. If this is the case,
340: N would be produced increasingly more prominently by massive stars in
341: the most massive galaxies, while the less-massive galaxies with their
342: more protracted star formation history would have an additional
343: contribution from intermediate mass stars. As mentioned, the delay in
344: the release of N from intermediate mass stars is $\sim$100~Myrs (the
345: lifetime of 5$-$6~M$_\odot$ solar metallicity stars); intermediate
346: mass stars do not release significant amounts of Mg at high
347: metallicities \footnote{note that significant contribution from AGB
348: stars to the heavy isotopes $^{25}$Mg and $^{26}$Mg is predicted at
349: metallicities lower than [Fe/H]$<-0.5$ \citep{Fenn03}}, which would
350: explain the different behavior of the Mg$-$$\sigma$ relation compared
351: with the N-$\sigma$ relation.
352:
353: Carbon is another controversial element. While some authors argue
354: that it is mainly produced in intermediate mass stars
355: \citep[][]{Chiapp03} (with masses between 1$-$4~M$_{\odot}$) others
356: suggest that massive stars are the main contributors
357: \citep[eg.,][]{Carig05}. The offset between galaxies in different
358: environments that is not seen in other elements produced mainly in
359: massive stars, such as Mg, favors the suggestion that C is produced
360: mainly in lower mass stars. However, if low- and intermediate-mass
361: stars {\it are} the main producers of C, it is difficult to explain
362: the strong correlation between C and $\sigma$ and the overabundance of
363: C/Fe with respect to the Sun. One possibility is that C is produced
364: in both, massive and lower mass stars in comparable proportions.
365: The significant slope in the
366: relation would be the consequence of C production in massive stars
367: while the offset would come from a different production in lower mass
368: stars. This latter C release would increase the C budget in LDEG
369: while in the HDEG, star formation ceases before this can happen.
370:
371: Of course these are very simplified scenarios. The relative
372: importance of the different N-sites and the origin of C in elliptical
373: galaxies will be explored in a future paper using our {\tt GEtool}
374: chemical evolution code (eg., Hughes et~al. 2008).
375:
376: In summary, we report the first determination of ``pure'' nitrogen
377: abundances for a sample of early-type galaxies and compare these with
378: stellar population models. Contrary to what has been claimed
379: previously, the nitrogen abundances do not vary with the velocity
380: dispersion of the galaxies. The environmental differences found for C
381: and N abundances impose an important constraint on chemical evolution
382: and galaxy formation models. It will be critical to independently
383: constrain the oxygen abundance and its dependence upon galaxy mass, in
384: part because of the obvious synergies between CNO in low- and
385: intermediate-mass stars, and in part because oxygen is simply the most
386: dominant component ($\sim$50\%) of global metallicity.
387:
388:
389: %\section{acknowledgments}
390: \acknowledgments
391: We thank Francesca Matteucci and Reynier Peletier for several very
392: useful discussions. The expert guidance of the referee, Judy Cohen, is gratefully acknowledged. This work has been partially supported by the
393: Spanish research project AYA2007-67752-C03-03, the Marie Curie Intra-European
394: Fellowship scheme within the 6th European Community Framework
395: Programme (PSB), UK's Science \& Technology Facilities Council (BKG).
396: E.T. acknowledges the support of the University of Central
397: Lancashire's Livesey Award scheme (PSB).
398:
399: \bibliographystyle{aa}
400: \bibliography{references}{}
401:
402: \clearpage
403:
404: \begin{figure}
405: \resizebox{0.45\textwidth}{!}{\includegraphics[angle=-90]{f1.ps}}
406: \caption{Left column: Line-strength indices versus velocity dispersion. Red
407: solid circles correspond to HDEG, while blue open circles are LDEG.
408: Right column: Blue, green, and red boxes represent the bandpass of the
409: index definition: In blue and red are the two-bands used to define the
410: pseudocontinua. The flux in both,
411: the data and the models have been normalized to these bands.
412: The central band where the feature is located is indicated in green.
413: Black lines represent stellar population models. Only models with solar
414: (dashed lines)
415: and supersolar (dotted lines) metallicity
416: are plotted; 6 and 15 Gyr models are plotted for each
417: metallicity. The stacked spectra for LDEG are plotted in blue,
418: while the HDEG are represented in red. Dominant species and other important
419: species (within brackets) contributing to the indices defined by SWB05 are shown
420: in the inset of the panels. A - symbol indicates that an increase in the abundance of that element makes the index weaker.
421: \label{figurel}}
422: \end{figure}
423:
424: \begin{figure}
425: \resizebox{0.5\textwidth}{!}{\includegraphics[angle=-90]{f2.ps}}
426: \caption{Variations of the analyzed indices with age at different
427: metallicities as indicated in the inset. BC03 models are used for the NH3360
428: while V08 models are employed for the remainder of the
429: plots.\label{sensitivity}}
430: \end{figure}
431:
432: \end{document}
433: