0705.2060/ms.tex
1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3: %\documentclass[manuscript]{aastex}
4: %\documentclass[preprint2]{aastex}
5: % \documentclass[preprint2,longabstract]{aastex}
6: %\usepackage{booktabs}
7: \slugcomment{Accepted for publication in ApJ Letters }
8: 
9: \shorttitle{TRACING THE GALACTIC THICK DISK TO SOLAR METALLICITIES}
10: \shortauthors{BENSBY ET AL.}
11: 
12: %=====================================================================
13: %   MACROS
14: \newcommand\teff{T_{\rm eff}}
15: \newcommand\kms{\rm\,km\,s^{-1}}
16: \newcommand\zmin{z_{\rm min}}
17: \newcommand\zmax{z_{\rm max}}
18: \newcommand\ulsr{U_{\rm LSR}}
19: \newcommand\vlsr{V_{\rm LSR}}
20: \newcommand\wlsr{W_{\rm LSR}}
21: \newcommand\uasym{U_{\rm asym}}
22: \newcommand\vasym{V_{\rm asym}}
23: \newcommand\wasym{W_{\rm asym}}
24: \newcommand\usigma{\sigma_{\rm U}}
25: \newcommand\vsigma{\sigma_{\rm V}}
26: \newcommand\wsigma{\sigma_{\rm W}}
27: \newcommand\uw{(U_{\rm LSR}^2+W_{\rm LSR}^2)^{1/2}}
28: \newcommand\uvw{(U_{\rm LSR}^2 + V_{\rm LSR}^2 + W_{\rm LSR}^2)^{1/2}}
29: 
30: %=====================================================================
31: \begin{document}
32: 
33: %=====================================================================
34: \title{
35: Tracing the Galactic thick disk to solar metallicities\altaffilmark{1}
36: }
37: \author{
38: T. Bensby,\altaffilmark{2}
39: A.R. Zenn,\altaffilmark{2}
40: M.S. Oey,\altaffilmark{2}
41: and
42: S. Feltzing\,\altaffilmark{3}
43: }
44: 
45: \altaffiltext{1}{Based on data collected with the 6.5\,m Magellan
46: telescopes at the Las Campanas Observatory, and with the Very
47: Large Telescope at the European Southern Observatory
48: (ESO proposal 72.B-0179).}
49: \altaffiltext{2}{Department of Astronomy, University of Michigan, 
50: Ann Arbor, MI, USA; {\tt tbensby@umich.edu}}
51: \altaffiltext{3}{Lund Observatory, Lund, Sweden;
52: {\tt sofia@astro.lu.se}}
53: 
54: %=====================================================================
55: \begin{abstract}
56: We show that the Galactic thick disk reaches at least solar 
57: metallicities, and that it experienced strong chemical enrichment
58: during a period of $\sim3$\,Gyr, ending around 8-9\,Gyr ago. 
59: This finding puts further constraints on 
60: the relation and interface between the thin and thick disks, and their 
61: formation processes.
62: Our results are based on a detailed elemental abundance analysis of 261 
63: kinematically selected F and G dwarf stars in the solar neighborhood:
64: 194 likely members of the thick disk and 67 likely members of the thin disk, 
65: in the range $\rm -1.3\lesssim[Fe/H]\lesssim+0.4$. 
66: \end{abstract}
67: 
68: \keywords{
69: Galaxy: disk ---
70: Galaxy: formation ---
71: Galaxy: evolution ---
72: solar neighbourhood ---
73: stars: abundances ---
74: stars: kinematics
75: }
76: 
77: %==============================================================================
78: \section{Introduction}
79: 
80: The Milky Way has since the early 1980s been known to have two disk 
81: components, a thin disk and a thick disk \citep{gilmore1983}.
82: Since then, several studies, using high-resolution spectra to derive
83: elemental abundances in disk dwarf stars, have been aimed at establishing
84: the properties of the thick disk, and to better understand
85: its origin and role in our Galaxy
86: \citep[e.g.,][]{gratton2000,prochaska2000,fuhrmann2004,mishenina2004,
87: reddy2006,bensby2003,bensby2005}.
88: The thick disk is now known to be a major Galactic stellar 
89: population, and that its stars have hotter kinematics, higher ages, 
90: and are chemically distinct from the stars of the thin disk.
91: All this points to separate origins and different chemical histories
92: for the thin and thick  disks.
93: 
94: Recently, a lot of structure has been observed amongst the stars in the Galaxy.
95: In the disk in the solar neighbourhood this is seen as various stellar streams and 
96: moving groups \citep[e.g.][]{famaey2005short,helmi2006short}; 
97: and at larger distances, features such as e.g. 
98: "The Field of streams" \citep[e.g.][]{belokurov2006short}
99: have been detected. So, did the thick disk form as a single 
100: entity in the initial collapse of the protogalactic cloud 
101: \citep[e.g.,][]{eggen1962}, and/or is it a result of an ancient merger event, 
102: or is it made up of a stars coming from
103: streams and merger debris, i.e. a hierarchical origin 
104: \citep[e.g.][]{abadi2003,brook2004,robertson2004}?
105: A persistent question is why the Milky Way has two
106: disk populations.
107: 
108: Abundance trends and the metallicity distribution function 
109: of the thick disk are vital records to its formation and evolution.
110: However, the high metallicity limit of the thick disk remains poorly defined. 
111: For instance, \cite{fuhrmann2004,mishenina2004,reddy2006} 
112: suggest that the thick disk extends only up to $\rm [Fe/H]\,\approx\,-0.3$,
113: because their candidate thick disk stars at higher [Fe/H] either fall
114: within their thin disk abundance trends and/or have
115: highly eccentric orbits that are near the Galactic plane.
116: Hence their origins should be sought elsewhere, perhaps in stellar streams like 
117: the Hercules stream \citep[see e.g.][]{famaey2005short}. But, even if possible 
118: Hercules stream stars are weeded out, stars with thick-disk-like kinematics 
119: at high [Fe/H] still remain 
120: \citep[][see also Fig.~\ref{fig:contour}]{soubiran2005,bensby2007letter}. 
121: Furthermore, in \cite{bensby2003,bensby2004,bensby2005,bensby2006}
122: we find that the thick disk stars differ significantly from the thin disk 
123: stars, both in terms of abundance ratios as well as stellar ages, 
124: even at [Fe/H] close to solar. However, those results are based on a small 
125: stellar sample and need confirmation.
126: 
127: As described, the current data for the metal-rich thick disk are
128: confusing and ambiguous. It is therefore necessary to isolate the thick disk
129: abundance relations from those of other populations.
130: Therefore, we have carried out an extensive spectroscopic survey of 
131: metal-rich stars that are kinematically associated with the Galactic thick 
132: disk. In this Letter, we discuss Ti and Ba abundance trends, and combine 
133: our new results with our thin and thick disk results from
134: \cite{bensby2003,bensby2005}. Other $\alpha$-, $r$-, $s$-, and iron peak 
135: elements will be discussed in an upcoming paper (Bensby et al., in prep) 
136: together with the details of the kinematic selection criteria and the 
137: abundance analysis.
138: 
139: 
140: 
141: %==============================================================================
142: \section{Selection of targets, observations and abundance analysis}
143: \label{sec:observations}
144: 
145: \begin{figure}
146: \resizebox{\hsize}{!}{
147: \includegraphics[bb=18 215 592 650,clip]{f1a.eps}}
148: \resizebox{\hsize}{!}{
149: \includegraphics[bb=18 170 592 465,clip]{f1b.eps}}
150: \caption{
151:         {\sl (a)-(c):} Velocity-metallicity plots
152:         for the stellar sample. Thin disk and
153:         candidate thick disk stars are marked by
154:         open and filled circles, respectively.
155: 	{\sl (d):} $TD/D$ probability ratios versus metallicity
156: 	for the thick disk candidates. Bottom lines indicate
157: 	$TD/D=4$ (dotted line) and $TD/D=9$ (dashed line).
158: 	Lines on top give the corresponding fraction of stars
159: 	that have $TD/D$ larger than these ratios (scale on the
160: 	right hand side).
161:         [Fe/H] values are from our spectroscopic work.
162: \label{fig:contour}}
163: \end{figure}
164: 
165: The kinematic method from \cite{bensby2003,bensby2005} was used to select 
166: possible thick disk F and G dwarf stars from the \cite{nordstrom2004short} 
167: catalogue. Briefly, the method assumes Gaussian velocity distributions
168: for all stellar populations, and that the solar neighbourhood can be
169: represented as a mixture of only the thin disk, the thick disk, the Hercules 
170: stream, and the halo. Candidate thick disk stars are selected as 
171: those that have probabilities of belonging to the thick disk that are at least 
172: twice the probabilities of belonging to any of the other populations 
173: (and likewise for the other populations). The space velocities for the 159 
174: new thick disk and 10 new thin disk stars are shown in 
175: Fig.~\ref{fig:contour}a-c together with 35 thick 
176: disk and 57 thin disk stars from \cite{bensby2003,bensby2005}.
177: Also shown, in Fig.~\ref{fig:contour}d-e, is how the thick disk-to-thin disk
178: probability ratios ($TD/D$) vary with [Fe/H]. 
179: 
180: Echelle spectra were obtained in 2005 and 2006 with the MIKE spectrograph,
181: on the Magellan Clay 6.5\,m telescope, for 145 new thick disk stars
182: ($R\approx65\,000$, $S/N\gtrsim250$),
183: and in 2004 with the UVES spectrograph, on the ESO Very Large Telescope,
184: for 14 new thick disk and 10 new thin disk stars
185: ($R\approx110\,000$, $S/N\gtrsim250$).
186: 
187: The MARCS model stellar atmospheres 
188: \citep{gustafsson1975,edvardsson1993short,asplund1997} were used in the abundance
189: analysis. Excitation balance, and balance with line strength, of 
190: abundances from Fe\,{\sc i} lines, were used to determine effective 
191: temperatures and the microturbulence parameter. 
192: For the surface gravities we exploited accurate distances based 
193: on {\sc\sl Hipparcos} parallaxes \citep{esa1997}. Final abundances were
194: normalised on a line-by-line basis with our solar values as reference and
195: then averaged for each element.
196: 
197: Stellar ages were determined with the help of the Yonsei-Yale (Y$^{2}$)
198: isochrones \citep{kim2002,demarque2004}, with appropriate $\alpha$-enhancements,
199: in the $\teff$-$M_{\rm V}$ plane.
200: Upper and lower limits on the ages were estimated from the error bars due
201: to an uncertainty of $\pm 70$\,K in $\teff$
202: and the uncertainty in $M_{\rm V}$ due to the error in the parallax
203: \citep[see also][]{bensby2003}.
204: 
205: %==============================================================================
206: \section{Results and discussion} \label{sec:results}
207: 
208: 
209: %==============================================================================
210: \subsection{Abundance trends}
211: 
212: \begin{figure*}
213: \resizebox{\hsize}{!}{
214: \includegraphics[bb=18 165 592 655,clip]{f2a.eps}
215: \includegraphics[bb=18 165 592 655,clip]{f2b.eps}}
216: \caption{
217:         {\sl Top four panels:}
218:         [Ti/Fe] and [Ba/Fe] versus [Fe/H]. Thin and thick disk stars are
219:         marked by open and filled circles, respectively.
220:         Solid lines marks the boundary for the thin sample.
221:         {\sl Bottom panels:}
222:         Distribution of all 195 thick disk stars
223:         {\sl (white histograms)}, and the thick disk stars that
224:         fall within the thin disk abundance trends {\sl (black histograms)}.
225:         The dotted line in the bottom panels shows the fraction (scale on the
226:         right-hand side) of thick disk stars in each
227:         bin that fall within the thin disk abundance trends.
228: \label{fig:haltplottar0}}
229: \end{figure*}
230: 
231: Figure~\ref{fig:haltplottar0} shows the resulting [Ti/Fe] and [Ba/Fe] 
232: versus [Fe/H] trends. The stars associated with the 
233: thick disk first show a [Ti/Fe] plateau at $\rm [Fe/H] \lesssim -0.4$, 
234: a signature of fast enrichment from massive stars. At higher metallicities, 
235: the thick disk [Ti/Fe] ratio declines, indicating the delayed enrichment 
236: from SN\,Ia. The thin disk shows an overall 
237: shallow decline  in [Ti/Fe], characteristic of 
238: slow enrichment by {\sl both} massive and low-mass stars. 
239: At $\rm [Fe/H]\approx0$, the trends for the two disks converge.
240: 
241: [Ba/Fe] for the thick disk evolves almost in lockstep with [Fe/H].
242: As solar metallicity is approached, the thin and thick disk [Ba/Fe] 
243: trends diverge. At $\rm [Fe/H] >0$ it again becomes 
244: hard to differentiate the two disks.
245: 
246: Both [Ti/Fe] and [Ba/Fe] versus [Fe/H] 
247: demonstrate that kinematically hot stars associated with the thick disk
248: extend to solar metallicities. However, it is also evident that
249: there are thick disk stars 
250: that do not follow the general thick disk abundance trends. Instead, they
251: chemically behave as thin disk stars. This is at least
252: evident in the [Ti/Fe]-[Fe/H] plot as $\rm [Fe/H]\approx0$ is approached.
253: 
254: To try to determine the nature of these ambiguous stars, we use 
255: the thin disk sample (Figs.~\ref{fig:haltplottar0}b and e) to visually 
256: define boundaries on [Ti/Fe] and [Ba/Fe] for the thin disk 
257: (shown as solid lines in the {\sl upper four panels} of 
258: Fig.~\ref{fig:haltplottar0}). 
259: The number of candidate thick disk stars that fall within the thin disk 
260: abundance trends are shown in the {\sl bottom two panels} of 
261: Fig.~\ref{fig:haltplottar0}. There is a steady increase with metallicity of candidate 
262: thick disk stars that fall within the thin disk [Ti/Fe] trend, suggesting that 
263: the contamination from the high-velocity tail of the thin disk increases with 
264: [Fe/H]. The fraction that fall within the thin 
265: disk [Ba/Fe]-[Fe/H] trend is, on the other hand, generally small, and 
266: with no apparent trend.  This suggests that essentially all candidate thick 
267: disk stars could be genuine thick disk stars.
268: 
269: Due to the closeness of the thin and thick disk [Ti/Fe] trends at higher [Fe/H]
270: one can expect true members of the thick disk to fall within the thin disk 
271: trend, and vice versa. And, since the Ba abundances are based on only 3-4 
272: Ba\,{\sc ii} lines, there are larger measurement uncertainties in [Ba/Fe] 
273: than in [Ti/Fe]. Ba abundances could also be influenced by 
274: NLTE effects, hyperfine and isotopic structure, and blends from other 
275: spectral lines \citep[see, e.g.,][]{mashonkina2006}; effects that we have not 
276: accounted for. On the other hand, we present a strictly
277: differential abundance analysis. If the above effects were severe, 
278: we would not find well-defined and distinct Ba trends for 
279: two kinematically selected samples. Hence, we believe that our Ba abundances
280: are well determined. 
281: 
282: %==============================================================================
283: \subsection{Age trends}
284: 
285: \begin{figure}
286: \resizebox{\hsize}{!}{
287: \includegraphics[bb=18 165 592 670,clip]{f3.eps}}
288: \caption{
289:         Top and middle panel show age distributions for
290:         thick disk stars that fall above and below the [Ti/Fe] boundary,
291:         respectively. Bottom panel shows the thin disk distribution.
292: 	All age distributions are divided into four metallicity bins,
293: 	as indicated in the square brackets at the
294: 	top of each panel.
295: \label{fig:agemess}}
296: \end{figure}
297: 
298: \begin{figure}
299: \resizebox{\hsize}{!}{\includegraphics[bb=18 165 592 718,clip]{f4.eps}}
300: \caption{
301:         [Fe/H] and [Ba/Fe] versus age. Stars for which the upper and
302:         lower age estimates differ by at most 35\,\% are included.
303:         Thin and thick disk stars are marked by open and filled circles,
304:         respectively. The solid line  shows the running
305:         median ages (see text) for the thick disk stars, and the dashed line 
306:         when excluding thick disk stars that have thin disk [Ti/Fe]
307:         values (marked by crosses).
308:         The vertical, dotted line, is the age of the Sun (4.5\,Gyr).
309: \label{fig:agefeh}}
310: \end{figure}
311: 
312: The top panel in Fig.~\ref{fig:agemess} shows the age distributions for the 
313: candidate thick disk stars that follow the thick disk [Ti/Fe] trend
314: (as defined by the thin disk boundary line in Fig.~\ref{fig:haltplottar0}a)
315: while the middle panel shows those that do not. Each sub-sample has been 
316: divided into four metallicity bins, as shown. The bottom panel shows the age 
317: distributions for the thin disk sample. 
318: 
319: The candidate thick disk stars with thin disk [Ti/Fe] ratios appear to be 
320: younger than those above the boundary. 
321: For instance, in the $\rm -0.35<[Fe/H]<0$ bin, only one out of 26 candidate 
322: thick disk stars (4\,\%) that have a thin disk [Ti/Fe] ratio is older than 
323: 8\,Gyr. In the same metallicity bin, 23 out of 33 stars (70\,\%)
324: that remain above the thin disk [Ti/Fe] boundary are older than 8\,Gyr.
325: This duality in both ages and abundances again points to two distinct 
326: Galactic disk populations, both reaching $\rm [Fe/H]=0$.
327: 
328: Figure~\ref{fig:agefeh} shows [Fe/H] and [Ba/Fe] as a
329: function of age, excluding stars with estimated
330: upper and lower age limits (see Sect.~\ref{sec:observations}) that differ by 
331: more than 35\,\%. Running medians of the ages for the thick disk, 
332: calculated in steps of 0.1\,dex in [Fe/H], using a 0.2\,dex wide window in
333: [Fe/H], both with and without thick disk stars that have thin 
334: disk [Ti/Fe] ratios, are shown in Fig.~\ref{fig:agefeh}a.
335: For $\rm [Fe/H]\lesssim-0.8$, median ages are typically $\sim 12$\,Gyr.
336: The median age at higher [Fe/H] depends on whether
337: thick disk candidates that have thin disk [Ti/Fe] ratios
338: are included or not.  As many of the stars that fall below the 
339: [Ti/Fe] boundary have ages comparable to the stars that do not, 
340: it is likely that the actual relation
341: is intermediate to the solid and dashed lines. The age of the 
342: thick disk at solar metallicities is then $\sim9$\,Gyr, i.e. it takes the 
343: thick disk $\sim3$\,Gyr to reach $\rm [Fe/H]\approx0$.
344: 
345: %==============================================================================
346: \subsection{The metal-rich end of the Galactic thick disk}
347: 
348: Our kinematically hot stars appear to come from an old stellar population, with 
349: ages of 8-12\,Gyr, that extend at least to $\rm [Fe/H]\approx0$.  
350: This population is not only old, but also its stars have kinematic properties 
351: typical of the Galactic thick disk, and chemical 
352: properties similar to what is found in the Galactic thick disk. 
353: Furthermore, preliminary results show that the
354: abundance and age trends do not vary with
355: either of the $\ulsr$, $\vlsr$, and $\wlsr$ velocities
356: \citep[Bensby et al.~in prep., but see also][]{bensby2006venice_short}.
357: Therefore, this appears to be manifest evidence that this stellar 
358: population indeed is the Galactic thick disk.
359: That the thick disk really reach all the way up solar metallicities verifies
360: the existence of the ``knee'' present in most
361: thick disk $\rm [\alpha/Fe]$ trends.
362: Hence the thick disk
363: formed stars for at least 3\,Gyr and experienced strong enrichment, from both
364: SN\,II and SN\,Ia, during this period, ending $\sim 8$-9\,Gyr ago.
365: 
366: 
367: %==============================================================================
368: \subsection{The relation between the thin and thick disks}
369: 
370: In our sample, the most metal-poor stars with thin disk kinematics have
371: metallicities of $\rm [Fe/H]\approx-0.7$ and ages 
372: around 5\,Gyr. Hence, these stars are considerably younger than the 
373: most metal-rich thick disk stars at $\rm [Fe/H]\approx0$ whose ages 
374: are 8-9\,Gyr.
375: At super-solar metallicities, the thin disk stars appear to have ages 
376: comparable to those of the most metal-poor ones, i.e. $\sim5$\,Gyr, 
377: suggesting that they formed at the same time! This phenomenon could be 
378: explained by the infall of gas into the Galaxy, which initially was
379: poorly mixed with the remains of the old metal-rich gas. 
380: The first stars of the thin disk could then be 
381: metal-rich ($\gtrsim 0.3$), metal-poor ($\lesssim -0.5$), or, depending
382: on the degree of mixing of the gas, of any metallicity in the range
383: $\rm -0.7\lesssim [Fe/H] \lesssim +0.4$. 
384: This scenario may explain why there is no well-defined
385: age-metallicity relation in the solar neighbourhood
386: \citep[e.g.,][]{edvardsson1993short,feltzing2001,haywood2006}. However, we caution that age
387: uncertainties can be large and that the increase in dispersion of the metallicity
388: with stellar age, for nearby stars, partly could be due to migration of stellar 
389: orbits \citep[e.g.,][]{haywood2006,wielen1996}. 
390: 
391: Figure~\ref{fig:agefeh}b  shows [Ba/Fe] versus stellar age.
392: The two disks appear to follow smoothly in
393: time and there also appears to be a quiescent period of 1-2\,Gyr 
394: when almost no stars were formed, some 6-7\,Gyr ago.
395: However, our thin disk stellar
396: sample has by no means been selected to probe its oldest parts.
397: Hence, a possible hiatus, and the fact that that there are (a few) 
398: stars that have ages in betwen the two disks, should be investigated with
399: a sample targeted for the oldest thin disk.
400: 
401: Figure~\ref{fig:agefeh}b also helps to further understand the origin and
402: evolution of Ba in the Galactic disks. The "bump'' in the
403: thin disk [Ba/Fe]-[Fe/H] trend (Fig.~\ref{fig:haltplottar0}e) is no longer 
404: seen. As the most metal-rich thin disk stars evidently are not the youngest ones 
405: there is now instead a steady increase in [Ba/Fe] toward younger ages. 
406: The first, flat portion of the [Ba/Fe] trend is consistent with being 
407: due to the $r$-process. As the $s$-process becomes
408: significant (due to AGB stars), [Ba/Fe] will rise. 
409: The position of the Sun is consistent with an origin during the
410: early times of the thin disk, when Ba enrichment 
411: was mainly $r$-process dominated but started to give way to being 
412: $s$-process dominated. However, we caution that the solar Ba composition 
413: is thought to be $\sim80$\,\% $s$-process and $\sim20$\,\% $r$-process
414: \citep[e.g.,][]{arlandini1999short}.
415: 
416: 
417: 
418: %==============================================================================
419: \section{Summary}
420: 
421: We have presented clear evidence that 
422: the Galactic thick disk reaches at least solar metallicities, and thus that it 
423: experienced strong chemical enrichment during an early period ending some
424: 8-9\,Gyr ago. 
425: 
426: The plot of [Ba/Fe] versus time, instead of [Fe/H], offers a more straightforward
427: interpretation of  the evolution of Ba at high metallicities.
428: 
429: We find that even the most metal-rich stars of the thick disk are older than
430: the thin disk population, with a possible hiatus in the star formation between
431: these two populations. We are continuing to investigate these relationships
432: with a stellar sample designed to target the oldest stars of the thin disk.
433: 
434: %==============================================================================
435: \acknowledgments
436: 
437: This work was supported by the National Science Foundation, grant AST-0448900.
438: SF is supported by a grant from the Knut and Alice Wallenberg Foundation.
439: We thank the anonymous referee for valuable comments.
440: 
441: \bibliographystyle{apj}
442: \bibliography{referenser}
443: 
444: 
445: \end{document}
446: 
447: