0808.3584/ms.tex
1: \documentclass{emulateapj}
2: 
3: \usepackage{graphicx}
4: \usepackage{apjfonts}
5: \citestyle{aa}
6: \special{papersize=8.5in,11in}
7: \usepackage{longtable}
8: \usepackage{natbib}
9: \newcommand{\super}[1]{\scriptsize \ensuremath{^\textrm{#1}} \normalsize}
10: \newcommand{\sub}[1]{\tiny \ensuremath{_\textrm{#1}} \normalsize}
11: \newcommand{\Msun}[0]{$M_{\sun}$}
12: 
13: 
14: \begin{document}
15: 
16: \title{Lithium Depletion of Nearby Young Stellar Associations}
17: 
18: \author{Erin Mentuch, Alexis Brandeker\altaffilmark{1}, Marten H. van Kerkwijk, Ray Jayawardhana}
19: \affil{Department of Astronomy \& Astrophysics, University of Toronto, 50 St.\,George Street, Toronto, Ontario M5S\,3H4, Canada}
20: \and
21: \author{Peter H. Hauschildt}
22: \affil{Hamburger Sternwarte, Gojenbergsweg\,112, 21029, Hamburg, Germany}
23: 
24: \altaffiltext{1}{Current address: Stockholm Observatory, AlbaNova University centre, SE-106\,91 Stockholm, Sweden}
25: 
26: \begin{abstract}
27: 
28: We estimate cluster ages from lithium depletion in five
29: pre-main-sequence groups found within 100\,pc of the Sun: TW~Hydrae
30: Association, $\eta$~Chamaeleontis Cluster, $\beta$~Pictoris Moving
31: Group, Tucanae-Horologium Association and AB~Doradus Moving Group.  We
32: determine surface gravities, effective temperatures and lithium
33: abundances for over 900 spectra through least squares fitting to
34: model-atmosphere spectra.  For each group, we compare the dependence
35: of lithium abundance on temperature with isochrones from
36: pre-main-sequence evolutionary tracks to obtain model dependent ages.
37: We find that the $\eta$\,Chamaelontis Cluster and the TW~Hydrae
38: Association are the youngest, with ages of $12{\pm6}$\,Myr and $12{\pm8}$\,Myr,
39: respectively, followed by the $\beta$~Pictoris Moving Group at $21{\pm9}$\,Myr, the
40: Tucanae-Horologium Association at $27{\pm11}$\,Myr, and the AB Doradus
41: Moving Group at an age of at least $45$\,Myr (where we can only set a
42: lower limit since the models -- unlike real stars -- do not show much
43: lithium depletion beyond this age).  Here, the ordering is robust, but
44: the precise ages depend on our choice of both atmospheric and
45: evolutionary models.  As a result, while our ages are consistent with
46: estimates based on Hertzsprung-Russell isochrone fitting and dynamical
47: expansion, they are not yet more precise.  Our observations do show
48: that with improved models, much stronger constraints should be
49: feasible: the intrinsic uncertainties, as measured from the scatter
50: between measurements from different spectra of the same star, are very
51: low: around 10\,K in effective temperature, 0.05\,dex in surface
52: gravity, and 0.03\,dex in lithium abundance.
53: 
54: \end{abstract}
55: \keywords{open clusters and associations: individual ($\eta$
56:   Chamaelontis, TW Hydrae, $\beta$ Pictoris, Tucanae-Horologium, AB
57:   Doradus) --- stars: abundances --- line: profiles --- stars:
58:   pre-main-sequence} 
59: 
60: \section{Introduction}\label{s:intro}
61: 
62: Triggered largely by the discovery of young stars in the {\em ROSAT} X-ray Satellite
63: All-Sky Survey, over the last decade several nearby pre-main-sequence
64: (PMS) star groups have been identified (for a review, see 
65: \citealt{zuc04}).  Ranging in age from roughly 6\,Myr to $\sim$100\,Myr, five
66: separate groups can be distinguished: TW Hydrae Association (TWA),
67: $\eta$~Chamaeleontis cluster ($\eta$\,Cha), $\beta$~Pictoris Moving
68: Group (BPMG), Tucanae-Horologium association (TUCHOR) and AB~Doradus
69: Moving Group (ABD).  The common space motions and localized sky positions 
70: suggest that these groups are likely connected to the Sco-Cen star
71: forming region located $\sim$100\,pc away in the southern
72: hemisphere.  \citet{mam99} and \citet{son03} have traced back the space
73: motion of members in BPMG, TWA and $\eta$\,Cha and argue that the
74: groups are related to a star formation burst in the
75: Sco-Cen region as a result of the passing of the Carina arm
76: $\sim$60\,Myr ago. 
77: 
78: These groups, because of their close vicinity, are excellent
79: laboratories for studying star and planet formation.  Well constrained
80: ages are necessary to make conclusions about timescales of, e.g., disk
81: dissipation and planet formation.  Already, observations 
82: from the same sample presented
83: in this paper have revealed that accretion disks can last up to
84: $\sim$10\,Myr, but beyond this is rare \citep{jay06}.
85: 
86: \defcitealias{bar98}{BCAH98} 
87: Previously, ages have been derived from Hertzsprung-Russell (HR)
88: diagram fitting, group dynamics and lithium abundance
89: measurements. \citet{luh04} provide an HR diagram isochrone age for
90: $\eta$\,Cha of $6^{+2}_{-1}$\,Myr derived from the evolutionary
91: models of \citeauthor{bar98} (\citeyear{bar98}; hereafter \citetalias{bar98})  and \citet{pal99}, which agrees well with the
92: dynamical expansion age of 6.7\,Myr determined by \citet{jil05}. The
93: dynamical age of TWA has been harder to determine because of
94: inconsistent space motions among its more than 30 members. An inferred
95: age of $8.3\pm0.8$\,Myr is given to TWA based on the dynamical motion
96: of four members \citep{del06}. However, the likely complex dynamical
97: evolution of TWA has led to several plausible evolutionary
98: scenarios. \citet{mak05} attribute this complex evolution to a chance 
99: encounter with Vega, while \citet{law05} suggest
100: TWA is composed of two separate groups, based on bimodal rotation
101: period distributions with distinctly separate ages of $\sim$\,10\,Myr
102: and $\sim$\,17\,Myr. More recently, \citet{bar06} finds a conservative
103: age of $10^{+10}_{-7}$\,Myr by comparing ages from HR diagram
104: isochrone comparisons (from BCAH98) and lithium abundances.
105: 
106: The slightly older group BPMG has an estimated age of $12^{+8}_{-4}$\,Myr 
107: based on HR diagram isochrone comparisons (from \citetalias{bar98}) and lithium abundances 
108: \citep{zuc01}, with three dimensional motions that are consistent 
109: with a dynamical expansion age of 11.5\,Myr \citep{ort02}. \citet{fei06}
110: independently derived an age of $13^{+4}_{-3}$\,Myr for the recently 
111: confirmed wide binary system of 51~Eri and GJ\,3305, part of BPMG.
112: 
113: Known to be older than BPMG, but younger than the Pleiades, TUCHOR
114: has an age of 20--40\,Myr based on H$\alpha$ measurements, X-ray
115: luminosity, rotation and lithium abundances in comparison to other
116: young clusters like TWA and the Pleiades \citep{zuc00,ste00}.
117:  
118: Perhaps, the most debated age is that of the ABD group. \citet{zuc04b}
119: derived an age of 50\,Myr by comparing the H$\alpha$ emission strength of
120: ABD members to members of the younger TUCHOR association, in addition
121: to fitting its three M-type members to HR diagram isochrones. In
122: contrast, \citet{luh05} compared HR isochrones of ABD members to those 
123: of two well-observed clusters with ages of 50 and 125\,Myr and suggested
124: that the ABD group is coeval with the Pleiades at an age of 100--125\,Myr.
125:  The latter age is strongly supported by \citet{ort07},
126: who compute full 3D galactic orbits of ABD and the Pleiades
127: cluster and show the dynamics of the two groups can be traced back to
128: a common origin of $119\pm20$\,Myr ago.
129: 
130: A relatively new approach to age estimates is to use the evolution of
131: the lithium abundance for low-mass, partially and fully convective PMS
132: stars \citep{bil97,jef05}. The initiation and duration of
133: lithium depletion in PMS stars is dependent on mass and is very
134: sensitive to the central temperature. Lithium is converted into helium
135: in $p,\alpha$ reactions in cores of low-mass stars when the
136: temperature reaches $2.5\times10^{6}$\,K. The lower the stellar mass,
137: the longer the time it takes to reach this critical temperature. For
138: example, a 0.6\,\Msun\ star begins to burn lithium at an age of
139: 3\,Myr, while a lower mass star at 0.1\,\Msun\ begins to burn lithium
140: at an age of 40\,Myr. Stars with $M<0.06$\,\Msun\ never reach this
141: typical temperature, while stars with 0.6\,\Msun\ $<M<$
142: 1.2\,\Msun\ burn lithium for a short period (1--2\,Myr) until a
143: radiative core develops, and more massive pre-main sequence stars do
144: not destroy lithium in their envelope at all. The result of these
145: processes is a dip in the lithium abundance as a function of luminosity
146: (and consequently, a function of effective temperature),
147: only affecting stars with spectral types late than F5. As a group
148: ages, this dip becomes deeper and widens on the cool end as
149: progressively cooler stars reach the critical core temperature.
150: 
151: The cool end of this dip has been used to date coeval groups
152: containing late M-dwarf stars by identifying the lithium depletion
153: boundary (LDB). The LDB marks the
154: luminosity above which all stars will
155: have depleted their lithium. The lithium is very quickly depleted in
156: these low-mass stars, so the LDB marks a sharp jump from initial to
157: near depleted lithium abundances. As the temperature in the cores of
158: PMS stars increases in time, the LDB will shift to cooler temperatures
159: as a cluster ages. LDB ages have been determined for the Pleiades (125
160: $\pm$ 8 Myr), $\alpha$\,Per (90 $\pm$ 10 Myr), IC\,2391 (53 $\pm$ 5
161: Myr), and NGC\,2547 (35 $\pm$ 4 Myr)
162: \citep{sta98,bar99,sta99,bar04,jef05}.
163: 
164: In this paper, we fit more than 900 multi-epoch, high-resolution spectra, introduced
165: in \S\ref{s:obs}, of 121 low-mass PMS stars to synthetic spectra
166: created from PHOENIX model atmospheres (see \S\ref{s:models}). We
167: begin by identifying the model with the best-fitting surface gravity, $\log
168: g$, and effective temperature, $T_\mathrm{eff}$ for each 
169: observed spectrum, as described in
170: \S\ref{s:gravtemp}. In \S\ref{s:Li}, we use these
171: best-fit $T_\mathrm{eff}$ and $\log g$ to find the lithium abundance
172: by fitting model spectra to both the 6\,104\,\AA\ and
173: 6\,708\,\AA\ lithium features in the observations, but now using
174: lithium abundance as a free parameter.  For comparison, we also measure the
175: equivalent width (EW) of the 6\,708\,\AA\ lithium doublet which allows us
176: to chronologically order the groups based solely on empirical 
177: measurements. We proceed by comparing the
178: measured lithium abundance distribution to that predicted by PMS
179: evolutionary models of \citetalias{bar98} and \citet{sie00} and
180: estimate model dependent ages for each of the PMS groups.
181: 
182: \section{Observations}\label{s:obs}
183: 
184: High-resolution spectra were obtained during six separate observation
185: runs, on a total of 19 nights, between December 2004 and April 2006,
186: utilizing the MIKE spectrograph at the Magellan-Clay 6.5\,m telescope
187: on Las Campanas, Chile. MIKE is a double echelle instrument, covering
188: two separate wavelength regions. For this study, the red region from
189: 4\,900\,\AA\ to 9\,300\,\AA\ is used. The raw data were bias subtracted 
190: and flat-fielded, and before extraction, the
191: scattered background in the spectrograph was subtracted by fitting splines
192: to interorder pixels. The spatial direction of the projected slit produced
193: by MIKE is wavelength dependent and not aligned with the CCD. We therefore
194: extracted spectra using a custom procedure develped in ESO-MIDAS, that
195: takes into account the tilt and optimally extracts the spectrum by
196: iteratively estimating the slit illumination function. For wavelength
197: calibration, exposures of a Thorium-Argon lamp were used, as well as
198: observed telluric absorption lines. A more detailed account of the
199: reductions and a log of the observations will appear in a forthcoming paper
200: (A.\ Brandeker et al.\ 2008, in prep.). Multiple spectra for many 
201: of the targets in our sample were taken in
202: order to search for multiplicity and perform variability studies.
203: 
204: With no binning and using the 0\farcs35 slit, the spectra in this
205: study have a resolution of R$\sim$60\,000. The pixel scale was
206: 0\farcs13\,pix$^{-1}$ in the spatial direction, and about
207: 24\,m\,\AA\,pix$^{-1}$ at 6\,500\,\AA\ in the spectral
208: direction. Integration times were chosen so that the signal-to-noise
209: ratio (S/N) $\gtrsim$70 per spectral resolution element at
210: 6\,500\,\AA, except for the brightest stars where this would have
211: implied an exposure shorter than 120\,s. In those cases, we used the
212: longest exposure time shorter than 120\,s that did not saturate the
213: detector, giving (S/N) = 70--500, depending on seeing.
214: 
215: For this study, we only use objects with spectral types later than F5,
216: and earlier than M5.  Hotter stars do not show depleted lithium
217: abundances in their atmospheres, because their cores are already
218: radiative when lithium burning starts. Cooler stars were not included
219: in the survey because they were too faint.  We also exclude obvious
220: spectroscopic binaries. In total we have a sample of 121 stars, with
221: 11 from $\eta$\,Cha, 32 from TWA, 23 objects from BPMG, 35 
222: from TUCHOR, and 22 from ABD. For comparison, we also analysed 
223: 20 radial-velocity standard stars from the field.
224: 
225: \section{Models}\label{s:models}
226: 
227: We compute synthetic spectra using version 14 of the PHOENIX model
228: atmosphere package \citep{hau98}. The model atmospheres are described
229: in \cite{kuc05,kuc06}. In total, we have 240 model spectra
230: covering a spectral region from 3\,200\,\AA\ to 10\,000\,\AA\, with
231: 0.03\,\AA\ spectral resolution, with temperatures ($T_\mathrm{eff}$)
232: ranging from 2\,500\,K to 6\,500\,K in steps of 100\,K, and surface
233: gravities ($\log g$) ranging from 3.0 to 6.0 in steps of 0.5
234: (i.e.\ gravities ranging from $10^3$\,cm\,s$^{-2}$ to
235: $10^6$\,cm\,s$^{-2}$).  For each of these temperature and gravity
236: combinations, we have additional models with different lithium
237: abundances ($N_{\rm Li}$) ranging from $\log N_{\mathrm{Li}}=0.0$ to
238: 4.0 in steps of 0.5 (where the normalization is such that $\log
239: N_{\mathrm{H}}=12$), which cover the 6\,104\,\AA\ and
240: 6\,708\,\AA\ lithium absorption lines. All models were calculated at solar
241: metallicity.
242: 
243: The atmospheric models are calculated under the assumptions 
244:  of local thermodynamic equilibrium (LTE). \citet{car94} have systematically
245:  analyzed the effects of non-LTE on the formation of the \ion{Li}{1} line in 
246:  cool stars. They show that that the non-LTE effects can lead to discrepancies
247:  in lithium abundances measured from the 6\,708\,\AA\ lithium doublet
248:   from about -0.3 to 0.3 dex depending on the depletion
249:  and temperature. We attempt to minimize the exclusion of non-LTE 
250:  effects by also fitting the 6\,104\,\AA\ lithium feature whose non-LTE
251:  corrections are opposite to those of the 6\,708\,\AA\ line \citep{car94}.
252:  
253:  In addition, we do not consider the effects of 
254: chromospheric and/or magnetic activity. To compensate for the former, we exclude 
255: from our fitting method emission lines due to chromospheric activity
256: that are frequently seen in classical T~Tauri and post T~Tauri stars 
257: (described further in \S\ref{s:mods}). We also note here that  
258: very strong lines --- which are completely dominated 
259: by pressure broadening, a relatively poorly understood process ---
260: require special treatment to reproduce the line profile. Although the
261:  models include attempts to do this, it does not always work satisfactory.
262: The resulting mismatches primarily affects lines 
263: from neutral metals that are highly abundant and very optically 
264: thick (such as \ion{K}{1} and \ion{Na}{1}). For less abundant neutral 
265: metals, such as \ion{Li}{1}, the effects should be minimal.
266: 
267: \section{Surface Gravities and Effective Temperatures}\label{s:gravtemp}
268: 
269: We derive surface gravities and
270: effective temperatures, following the methodology of \citet{moh04}, 
271: by fitting theoretical spectra to
272: our observed spectra in small, 80\,\AA\ wide, spectral regions that
273: contain absorption lines or molecular bands that are highly sensitive
274: to surface gravity and/or effective temperature.
275: 
276: \subsection{Modifications to the data}\label{s:mods}
277: 
278: To ensure our measurements would not be influenced by telluric
279: absorption lines, or by stellar emission related to activity, we made
280: some alterations to the observed spectra before fitting. For the
281: telluric lines, we first used multi-epoch observations of one of our
282: stars ($\beta$\,Pic) to identify the stronger atmospheric lines and determine their
283: depths relative to the continuum. Next, for each spectrum, we marked
284: a fixed spectral width of data points around the center of each
285: telluric line as contaminated; these regions are ignored in all fits
286: done below. We used widths of the marked regions that depended on the
287: strength of the telluric line (as measured for $\beta$\,Pic): a region
288: of 0.2\,\AA\ was removed for relatively weak lines, which had a depth
289: less than 80\% of the continuum. Regions of 0.6\,\AA\ were removed for
290: line depths between 80\% and 88\%, while for the strongest, deeply
291: saturated absorption lines, 1.0\,\AA\ wide regions were removed.
292: 
293: As our sample consists of young stellar groups, many of the stars show
294: signatures of activity, with relatively narrow chromospheric emission
295: lines found in many of our targets, and strong and wide emission lines
296: indicative of accretion in a few others \citep{jay06,sch07}.  The
297: chromospheric activity causes emission features at a number of
298: wavelengths, with the strengths and locations depending on spectral
299: types and age \citep{eis90}. To mitigate the effects of activity, we
300: exclude 2.4\,\AA\ regions around lines that frequently are seen in
301: emission from classical T~Tauri and post-T~Tauri stars \citep{sta05}:
302: \ion{He}{1} at 5\,875.6\,\AA, \ion{Na}{1} at 5\,890.0 and
303: 5\,895.9\,\AA, \ion{O}{1} at 8\,446.5\,\AA, and \ion{Ca}{2} at
304: 8\,498.0, 8\,542.1 and 8\,662.1\,\AA.
305: 
306: Some stars also show signs of accretion, which leads to strong, wide
307: emission features, potentially leading to less accurate fits.
308: Since only three out of eleven $\eta$\,Cha and two out of 32 TWA
309: members show signs of accretion \citep{jay06}, this will affect
310: relatively few stars in the sample and in practice we do not find any effect of 
311: accretion on the uncertainties involved in constraining $T_{\mathrm{eff}}$, 
312: $\log g$ and $\log N_{\mathrm{Li}}$, except for the EW of one star 
313: (see \S\ref{s:ew}). There are no known accretors in any of the older 
314: PMS groups investigated in this study.
315: 
316: \subsection{Selection of spectral regions}\label{s:spec_regions}
317: 
318: We locate spectral regions that are
319: strongly sensitive to surface gravity and temperature. Both
320: \citet{tor93} and \citet{kir91} have compiled lists of prominent
321: absorption features useful for spectral classification of low-mass
322: stars. For all, we checked whether our model spectra (see
323: \S\ref{s:models}) indeed showed strong temperature and/or gravity
324: sensitivity, and selected only those that did.
325: 
326: \begin{deluxetable}{cccccc}
327: \tablecaption{\small{Spectral Regions Fitted}}
328: \tablecolumns{12}
329: \tablewidth{0pc}
330: \tabletypesize{\small}
331: \tablehead{
332: 	\colhead{Line ID} 				&
333: 	\colhead{$\lambda$ (\AA)} 		&
334: 	\colhead{$\Delta\lambda$ (\AA)} 	&
335: 	\colhead{T (K)}					&
336: 	\colhead{Sens.}					
337: }
338: \startdata 
339: \ion{Na}{1}\tablenotemark{a}		& 5\,893		& 	5\,850--5\,930 	&	2\,500--4\,500	&	T, $\log g$	\\ 
340: \ion{Fe}{1}\tablenotemark{a} 		& 5\,893, 5\,898 	&				&				&				\\
341: TiO\tablenotemark{a} 			& 5\,847--6\,058 	&				&				&	T 			\\
342: \tableline
343: VO\tablenotemark{b}	 		& 7\,851--7\,973 	&	7\,900--7\,980 	& 	4\,000--6\,500 	&	T 			\\ 
344: CN\tablenotemark{a,b} 			& 7\,916, 7\,941, 7963 & 			&				&				\\ 
345: \tableline
346: \ion{Na}{1}\tablenotemark{b} 		& 8\,183, 8\,195 	&	8\,150--8\,230 	&	2\,500--3\,000 	&	$\log g$ , T 	\\
347: \tableline
348: TiO\tablenotemark{a,b} 			& 8\,432, 8\,442, 8\,452 & 8\,400--8\,480 	&	2\,500--6\,500 	&	T 			\\ 
349: \ion{Fe}{1}\tablenotemark{b} 		& 8\,440, 8\,468 	&				&				&				\\ 
350: \tableline
351: \ion{Ca}{2}\tablenotemark{a,b}	 	 & 8\,498, 8\,542 & 	8485--8565 	&	2\,500--6\,500 	&	T 			\\ 
352: VO\tablenotemark{b} 	 		& 8\,521, 8\,538 	&				&				&				\\ 
353: \enddata	
354: \label{tab:spectralregions}
355: \tablecomments{A list of the spectral regions selected for our fitting 
356: method as described in \S\ref{s:spec_regions} \& \S\ref{s:methods}. 
357: The first and second columns identify and locate spectral features within the 
358: chosen spectral range that show strong sensitivity to temperature and/or 
359: surface gravity. The third column lists the 80\,\AA\ region that was selected in
360: our fitting procedure. The fourth column indicates what range in effective
361: temperature the selected region shows strong sensitivity to varying parameters
362: and the fifth column indicates whether the region is more sensitive to 
363: effective temperature or surface gravity.}
364: \tablenotetext{a}{\citet{tor93}}
365: \tablenotetext{b}{\citet{kir91}}
366: \end{deluxetable}%
367: 
368: %
369: \begin{figure}[t]
370: \begin{center}
371: \includegraphics[width=3.25in]{f1.eps}
372: \caption{Spectral fits (solid red lines) to $\eta$\,Cha~10
373:   (black dots), a K7 star, in the regions listed in Table
374:   \ref{tab:spectralregions}. Quality of fit ($\chi^2$) contours at 68.3,\%, 95.4\,\% and 99.99\,\%
375:   confidence for each spectral window are  shown in the bottom panel with each 
376:   level given by $\chi^{2}_{\mathrm{min}} \times [1 + (2.30, 6.17, 18.4)/N_{\rm
377: dof}]$, where $N_{\rm dof}$ is the degrees of freedom in the fit and
378: $\chi^{2}_{\mathrm{min}}$ is the best-fit value. The summed $\chi^2$ values for all 
379: spectral windows are represented by the tight solid black contours.  \label{fig:specfit1138}}
380: \end{center}
381: \end{figure}
382: 
383: In general, the dependence of these features on our parameters is
384: somewhat degenerate, as different $T_{\mathrm{eff}}$ and $\log g$
385: combinations can fit a given absorption feature equally well. As
386: discussed in \citet{moh04}, features that show contrasting dependence
387: on gravity and temperature are needed to constrain the parameter
388: space.  Thus, we choose combinations of molecular absorption features
389: like titanium oxide (TiO) and vanadium oxide (VO) with lines from
390: neutral alkali elements, like \ion{K}{1}, \ion{Na}{1} and
391: \ion{Mg}{1}. This works well because, for neutral alkali elements, an
392: increasing temperature can be compensated by increasing gravity, while
393: the molecular absorption bands show less correlation between gravity
394: and temperature. For instance, TiO bands in the vicinity of
395: 7\,100\,\AA\ are much more sensitive to temperature than to gravity,
396: while the triple headed TiO bands in the range of 8\,440\,\AA\ are
397: more sensitive to gravity than to temperature. Examples of this
398: dependence can be seen in the $\chi^2$ contours shown in the bottom 
399: panels of Figs.~\ref{fig:specfit1138} and \ref{fig:specfit8095}.  Of course, 
400: these dependencies change with temperature, and hence different sets of
401: regions are best for different spectral types. In principle, one could
402: choose to fit different regions in different temperature ranges, but
403: 
404: this risks (borne out in practise) that at the borders there are false
405: jumps in parameters due to systematic problems with the models. Thus,
406: we fit all regions for all stars, which gives very tight constraints
407: on $T_{\mathrm{eff}}$ and $\log g$ for all spectral types (and more
408: smoothly varying systematic offsets; see below).  After investigating
409: spectral fits of twelve individual regions collected from
410: \citet{tor93} and \citet{kir91}, we settled on the five separate
411: spectral regions listed in Table \ref{tab:spectralregions}.  Also
412: listed is the effective temperature range where the features provide
413: strong constraints, as well as whether these constraints are
414: predominantly on $T_{\mathrm{eff}}$ and/or $\log g$.
415: 
416: We excluded regions which either tended to overestimate $\log g$ or
417: provided $T_{\mathrm{eff}}$/$\log g$ values extremely offset from the
418: majority of spectral regions. For instance, for many of the neutral metal
419: lines, particularly \ion{K}{1} at 7\,665 and 7\,699\,\AA, the best fits are 
420: at surface gravities much higher than are expected for
421: late-type PMS stars. Since these lines are sensitive to gravity and generally fit
422: the spectrum very well, the entire fit was very sensitive to the
423: \ion{K}{1} line. Due to this inconsistency between the line strengths
424: in the atmospheric models and predicted surface gravity, this region
425: and several others were not used. We note that we 
426: use other neutral metal lines such as \ion{Na}{1} that could be flawed 
427: similarly. In practise though, these produced fits that were 
428: consistent with the other spectral regions investigated.
429: 
430: \begin{figure}[t]
431: \begin{center}
432: \includegraphics[width=3.25in]{f2.eps}
433: \caption{As Fig.~1, but for HIP 32235 (black dots), a G6V-type TUCHOR
434:   member.   \label{fig:specfit8095}}
435: \end{center}
436: \end{figure}
437: 
438: \subsection{Fitting Methods}\label{s:methods}
439: 
440: All spectra in our study with spectral types later than F5
441: (corresponding to about 6\,500\,K, the highest temperature for which
442: we have models) were fitted to the synthetic spectra for each spectral
443: window listed in Table \ref{tab:spectralregions}. Prior to the fit,
444: the synthetic spectra were convolved with a gaussian filter to match
445: the observed resolution. The fit used a variant of the
446: broadening-function formalism introduced by \citet{ruc02}. In this
447: formalism, a least-squares fit is made of the observed spectrum to a
448: set of reference spectra that differ only in their velocity offset.
449: This way, rotational broadening is automatically accounted for. The
450: difference with the formalism of \citeauthor{ruc02} is that the sum of
451: the model spectra at various velocities is also multiplied with a
452: polynomial, to account for not only the normalization, but also
453: for small errors in the flux calibration. For our small wavelength
454: regions, we found that a third-degree polynomial sufficed.
455: 
456: We note that our fitting method effectively introduces a relatively
457: large number of parameters that are not of direct physical interest,
458: viz., the line shape. Since these parameters might be covariant with
459: some of the effects of temperature and/or gravity changes, the
460: constraints on $T_{\rm eff}$ and $\log g$ we find are thus not as
461: strong as might be possible if, e.g., instead we broadened the
462: synthetic spectra with an analytic broadening model (with only the
463: projected rotational velocity as parameter). In practice, however,
464: the covariance is small, and our errors are clearly dominated by
465: systematic mismatches between the models and the observations.
466: 
467: With the above least-squares fitting method, we fitted each of the
468: five regions from each observed spectrum to 77 of the synthetic
469: spectra, covering all seven values of $\log g$ (from 3.0 to 6.0 in
470: steps of 0.5), and eleven temperatures in a range of 1\,000\,K around
471: the effective temperature inferred from the object's spectral type (as listed
472: in Tables~\ref{tab:TWA}--\ref{tab:RVstd}; we used the temperature scale
473: for dwarfs of \citealt{luh98}). For new companions with unknown
474: spectral types, initial estimates for the temperatures were found
475: through trial and error. 
476: 
477: To determine combined constraints on $T_\mathrm{eff}$ and $\log g$ for
478: a given spectrum, we add up the $\chi^{2}$ values derived for the five
479: regions. The results are similar to fitting the five regions
480: together, except that our method leaves greater freedom for variations
481: in continuum (which might be expected) or line shape (which would
482: not). The difference with fitting the entire spectrum is that we have
483: effectively ignored the parts of the spectrum that are either
484: insensitive to $T_\mathrm{eff}$ and $\log g$ or where the models are
485: shown to fit poorly to the observations. 
486: 
487: For our final estimates of the best-fit values of $T_\mathrm{eff}$ and
488: $\log g$ for each spectrum, we interpolate in the 7 by 11 grid by
489: determining the minimum of a two-dimensional parabola (of the form $a
490: + bx + cy + dx^{2} + exy + fy^{2}$) fit to the sixteen grid points
491: with the lowest values of $\chi^{2}$.
492: 
493: \subsection{Examples}\label{s:example}
494:  
495: In Fig.~\ref{fig:specfit1138} and \ref{fig:specfit8095}, we show our
496: fits for one spectrum of each of the stars $\eta$\,Cha~10 and TUCHOR member
497: HIP\,33235. The gaps in the data in the panels centered on 5\,890
498: and 8\,525\,\AA\ are due to the removal of emission lines as discussed
499: in \S\ref{s:mods}. Similarly, on close inspection, one sees that
500: multiple telluric lines have been removed from the 8\,190\,\AA\
501: region.
502: 
503: Overall, the broadened models reproduce the observed spectra well. In
504: detail, however, there are clear inconsistencies between the observed
505: and synthetic spectra, apparent in many of the panels. In particular,
506: a number of lines appear to be absent in the models, or are clearly
507: too weak. We also found examples of the reverse in some other
508: wavelength regions.
509: 
510: In the bottom panels of Figs.~\ref{fig:specfit1138} and
511: \ref{fig:specfit8095}, the resulting 68.3\,\%, 95.4\,\% and 99.99\,\% confidence contours in the
512: $T_{\mathrm{eff}}$-$\log g$ parameter space are shown, with levels set
513: according to \citet{pre92} at $\chi^{2}_{\mathrm{min}} \times [1 + (2.30, 6.17, 18.4)/N_{\rm
514: dof}]$, where $N_{\rm dof}$ is the degrees of freedom in the fit and
515: $\chi^{2}_{\mathrm{min}}$ is the best-fit value inferred from the
516: parabolic fit. Contours are shown for each spectral window as well as
517: for the total $\chi^{2}$ values. For the different
518: spectral regions, one notices the different covariances in
519: $T_{\mathrm{eff}}$ and $\log g$, and how, by using a number of these
520: sensitive regions, it is possible to constrain gravity and temperature
521: precisely. Indeed, the contours for the summed $\chi^{2}$
522: distribution (solid black contours) are extremely tight. We will see below that different
523: spectra of the same star lead to similarly small scatter in the
524: inferred temperature and gravity. 
525: 
526: One also notices, in particular in Fig.~\ref{fig:specfit8095}, that the
527: contours for the different regions are statistically inconsistent with
528: each other, with differences of several 100\,K in
529: $T_{\mathrm{eff}}$ and up to 1\,dex in $\log g$.  These differences
530: likely reflect systematic uncertainties in the models, similar to what
531: we find for the resulting best-fit average values below.
532: 
533: \begin{figure}[t]
534: \begin{center}
535: \plotone{f3.eps}
536: \caption{Surface gravity, $\log g$, as a function of effective
537:   temperature, $T_{\mathrm{eff}}$, for all five PMS groups, as well as
538:   field stars. PMS tracks from \citetalias{bar98} are overdrawn, with, 
539:   from top to bottom, ages of 4, 8, 15, 25, 35, 45~and~150\,Myr. 
540:   We represent the external error in the plotted parameters 
541:   for the ensemble in the bottom-left corner. It is evident that there are
542:   systematic problems with the values of $\log g$ derived from the
543:   models, particularly for $3\,500< T_{\mathrm{eff}}<4\,100\,$K.
544:   However, the relative ordering is clear: stars that are older, such
545:   as ABD and TUCHOR members, have higher surface gravities than
546:   younger stars. The ultrafast rotators and accretors in
547:   our sample are marked with circles and squares, respectively. \label{fig:logg_vs_t}}
548: \end{center}
549: \end{figure}
550: 
551: \subsection{Results}
552: \label{s:LiRes}
553: 
554: The best-fit model $T_{\mathrm{eff}}$ and $\log
555: g$ with corresponding errors, for each star, including field stars,
556: can be found in Tables~\ref{tab:TWA}--\ref{tab:RVstd}. We consider two
557: ways of estimating the associated uncertainties. First, we follow
558: \citet{pre92} and use the curvature of the best-fit two-dimensional
559: parabola to the $\chi^2$ values to find regions which are enclosed
560: within a level of 68\% confidence.  Second, we consider the error in
561: the mean between the results from different spectra of the same source
562: taken at different epochs (e.g.\ the standard deviation divided by the
563: square root of the number of observations taken). 
564: 
565: For the temperatures, our statistical uncertainties are typically around 9\,K,
566:  while the scatter derived from multiple observations
567: of the same object is on average 11\,K.  For the surface gravities, our 
568: statistical uncertainties numbers are 0.02\,dex on average, and the 
569: scatter derived from the error in the mean from multiple observations 
570: is about 0.03\,dex. The above suggests the true intrinsic uncertainties 
571: ($\sigma_\mathrm{int}$) in our temperature
572: and gravity measurements are very small, about 10\,K and 0.05\,dex
573: for a single observation.  We will see below, however, that systematic
574: mismatches as a function of, e.g., spectral type, are much larger.
575: 
576: In Fig.~\ref{fig:logg_vs_t}, we show the distribution of $\log g$ as a
577: function of $T_{\mathrm{eff}}$ for each group as well as for the field
578: stars observed. In addition, we draw isochrones from the
579: \citetalias{bar98} PMS evolutionary models.  One immediately sees
580: there is a clear systematic problem in determining $\log g$ for stars
581: with $3\,500<T_{\mathrm{eff}}<4\,100$\,K: $\log g$ increases
582:  with temperature from 3\,000\,K to 4\,000\,K, but at around 4\,000\,K, 
583:  it becomes almost 1 dex smaller, an unrealistic
584:  physical trend. The systematic differences in $\log g$ can result from a few
585: effects. If the resolution of the model spectra is not much better
586: than the observations before smoothing, line
587: depths can be systematically off.  In addition, stellar activity also
588: introduces systematic errors, as has been found for young M-dwarfs,
589: where the chromosphere feeds back into the photosphere \citep{fuh05}.
590: This feedback is not incorporated into the model, and may lead to
591: systematic errors of about 0.5 dex in $\log g$.
592: 
593: To evaluate the dependence of the fitted $T_\mathrm{eff}$ on $\log g$, we re-fit 
594: our spectra three more times using the same method described in 
595: \S\ref{s:mods}, except with $\log g$ fixed to 4.0, 4.5 and 5.0. The change 
596: in $T_\mathrm{eff}$ across the surface gravity space is small in an absolute 
597: sense. Going from log g fixed at 4.0 to 4.5, the average change in temperature
598: within our sample is $\Delta T_{\mathrm{eff}} = 71$\,K. It is 
599: slightly higher going from 4.5 to 5.0, with
600: $\Delta T{\mathrm{eff}} = 124$\,K. Changes from $\log g =$ 4.0 to 5.0, yield absolute 
601: average changes in $T_\mathrm{eff}$ of 162 K.
602: 
603: Thus, the external errors related to the models are much higher than 
604: the internal errors. Tables 2 to 7 list the best-fit surface 
605: gravities and temperatures with quoted errors representing the error 
606: in the mean of multi-epoch observations of the object. More conservatively,
607: we consider our external errors to be 150 K in $T_\mathrm{eff}$ and 
608:  0.5 dex in $\log g$.
609: 
610: In Fig.~\ref{fig:fitt_vs_sptt}, we compare our computed temperatures
611: to those obtained by converting spectral types to effective
612: temperatures using the temperature scale for dwarf-type stars of
613: \citet{luh98}. Generally, the two scales are consistent within the 
614: external errors just discussed, with the largest deviations
615: show by the accretors (marked by squares) and ultrafast 
616: rotators (marked by circles, see also \S\ref{s:rotation}). 
617: In addition, most of the objects with
618: $T_\mathrm{eff} >5000$\,K, while individually consistent within the
619: uncertainties, appear systematically to have fitted temperatures
620: slightly greater than those inferred from spectral type, suggesting a
621: small systematic error.
622: 
623: \begin{figure}[t]
624: \begin{center}
625: \plotone{f4.eps}
626: \caption{Comparison of the temperatures obtained from spectral fitting
627:   (see \S\ref{s:methods}) with those derived from spectral types using
628:   the scale of \citet{luh98}. We represent the external error in the plotted parameters for the ensemble in the top left corner. The ultrafast rotators and accretors in
629:   our sample are marked with circles and squares, respectively. \label{fig:fitt_vs_sptt}}
630: \end{center}
631: \end{figure}
632: 
633: \section{Lithium Abundances}\label{s:Li}
634:  
635: We analyse lithium abundances using two independent methods. In one
636: method, which follows previous work \citep{jef05}, we measure the
637: equivalent width (EW) of the 6\,708\,\AA\ lithium absorption doublet
638: for each spectrum. We use this measurement and the spectral types quoted
639: in the literature to chronologically order the groups. The other
640: method uses the best-fit model calculated in \S\ref{s:gravtemp} to fit
641: model spectra for various lithium abundances to regions surrounding
642: the 6\,104 and 6\,708\,\AA\ lithium absorption features. As with our
643: EW measurements, we chronologically order the groups. Further, we 
644: compare the observations with PMS models of \citetalias{bar98} 
645: and \citet{sie00} to see if the isochrone ages based on lithium 
646: abundances are consistent with other age determinations 
647: of the five nearby PMS groups in this study
648: 
649: \subsection{Lithium equivalent widths}\label{s:ew}
650: 
651: First, we use the EW of the 6\,708\,\AA\ lithium doublet feature in
652: each observed spectrum as an empirical measurement
653: of the lithium abundance. To measure it, we interactively chose
654: the edges of the lithium feature and the boundaries of sufficiently 
655: large regions around it to define the stellar continuum. We chose 
656: to consistently reject the lowest 15\%
657: of the flux points in the continuum region (corresponding to the
658: deepest absorption features), so that we do not underestimate the
659: continuum flux. We do not apply any 
660: correction for the 5 identified accretors in our sample, but find 
661: that veiling due to accretion can reduce the measured EW, 
662: as discussed briefly below.
663: 
664: The resulting mean EWs, averaged over all available spectra, are
665: listed in Tables~\ref{tab:TWA}~to~\ref{tab:ABD}, with uncertainties
666: being the error in the mean between multiple spectra.
667: Fig.~\ref{fig:ew_vs_t} displays the resulting EWs as a function of
668: spectral type. From this purely empirical figure,
669: the chronological order of the groups is evident: from oldest to
670: youngest, they are ABD, TUCHOR, BPMG, TWA and $\eta$\,Cha. 
671: The same ordering was found by \citet{zuc04} from lithium EWs for
672: a smaller sample of stars. Although
673: some TWA members appear to be as young as those in $\eta$\,Cha, it is
674: quite clear that no members are older than BPMG, contrary to the
675: suggestion by \citet{law05}. (Note that we implicitly assume 
676: here the initial lithium abundance was
677: the same for all groups.  We return to this below.)
678: 
679: We identify the ultrafast rotators and accretors in our sample with
680: black circles and squares, respectively. The EWs of the rotators in
681: BPMG (red squares) and TUCHOR (green stars) stand out above the
682: general trend for each group; we will return to this in
683: \S\ref{s:rotation}.  We also note that one accretor in $\eta$\,Cha,
684: ($\eta$\,Cha\,13) has a lithium EW that is lower than most objects at
685: the same temperature ($\sim3100\,$K) and age, but we will find below a
686: lithium abundance near initial from the fits to the spectra of this
687: star (see \S\ref{s:Li-method}).  This suggests the low estimate of the
688: EW is related to the fact that it is accreting, either by its effect
689: on the continuum level or alternatively, by us 
690: underestimating the temperature.  Indeed, the
691: latter may well play a role: from our spectral fit, we infer
692: $T_\mathrm{eff}=3139\pm96$\,K, while that from its M2 spectral type,
693: one would estimate $T_\mathrm{eff}=3514$\,K \citep{luh98}.  With the latter
694: temperature, the object would match the trend much better. Because of the 
695: large accretion lines in the spectra, the models are a poor representation
696: of the data throughout the regions selected for our fits. A higher 
697: temperature model is just a slightly worse fit to the observations.
698: 
699: \begin{figure}[t]
700: \begin{center}
701: \plotone{f5.eps}
702: \caption{Equivalent width (EW) of the 6708\,\AA\ lithium line as a
703:   function of spectral type. Without depletion, the EW would increase
704:   with decreasing temperature, as is seen with the youngest two
705:   groups, TWA and $\eta$\,Cha.  But in older groups lithium is being
706:   depleted, and the order, from youngest to oldest, is evident: BPMG, TUCHOR,
707:   and ABD.  We identify the ultrafast rotators and accretors in our
708:   sample with black circles and squares, respectively. \label{fig:ew_vs_t}}
709: \end{center}
710: \end{figure}
711: 
712: %
713: 
714: \subsection{Lithium Line Analysis}\label{s:Li-method}
715: 
716: %
717: \begin{figure*}[t]
718: \begin{center}
719: \plotone{f6.eps}
720: \caption{Examples of fitting the two lithium features to three stars
721:   (from left to right: TWA 02A (M2e), CD-60\,416 in TUCHOR
722:   (K3/4) and GJ 3305 in BPMG (M0.5)).  For each star, the top panel
723:   shows the fit to the strong 6\,708\,\AA\ lithium doublet and the
724:   middle panel the fit to the much weaker 6\,104\,\AA\ lithium line.
725:   The bottom panel shows the normalized $\chi^{2}$ values for the
726:   abundances at which we had models, and the parabolic interpolations
727:   used to determine the best-fit abundances (which is taken to be at
728:   the minimum of the sum of these two curves; see
729:   \S\ref{s:Li-method}). \label{fig:lifit-montage} }
730: \end{center}
731: \end{figure*}
732: 
733: Using the average best-fit effective temperatures and surface
734: gravities listed in Tables~\ref{tab:TWA}~to~\ref{tab:RVstd}, synthetic
735: spectra, varying in lithium abundance from $\log N_{\mathrm{Li}}$ =
736: 0.0 to 4.0 in steps of 0.5, are, for each spectrum, fitted to small
737: spectral regions around the 6\,104\,\AA\ and 6\,708\,\AA\ lithium
738: absorption lines. The fitting method is identical to the method
739: outlined in \S\ref{s:methods}. However, we constrain the minimization
740: of the least-squares fit in a slightly different manner.
741: 
742: Synthetic spectra of varying lithium abundance are fitted to
743: 20\,\AA\ wide spectral regions -- specifically
744: 6\,095--6\,115\,\AA\ and 6\,695--6\,715\,\AA.  The
745: 6\,104\,\AA\ lithium triplet line is a weaker transition than the
746: 6\,708\,\AA\ doublet line. It is also blended into the strong
747: 6\,103\,\AA\ \ion{Ca}{2} absorption line. As a result, the line is
748: only detectable for high lithium abundance. Overall, we are not very
749: sensitive to this line, detecting only a small change in $\chi^{2}$
750: over the entire lithium abundance range. This is not the case with the
751: stronger lithium doublet at 6\,708\,\AA, which is very sensitive to
752: lithium abundance, and shows sharp transitions from good to bad in its
753: least-squares fits.  For high lithium abundances, however, the line
754: saturates and without the 6\,104\,\AA\ line no good abundance
755: estimates are possible.
756: 
757: In order to be able to treat all data uniformly, irrespective of
758: lithium abundance, we proceeded as follows.  First, we use the
759: 6\,708\,\AA\ region, with its higher sensitivity, to determine the
760: approximate abundance, and select the points with $\log N_\mathrm{Li}$
761: corresponding to the 4 lowest $\chi^{2}$ values from the fit to this
762: region.  For both spectral regions, we then fit a 2nd order polynomial
763: to the $\chi^2$ for these 4 selected $\log N_\mathrm{Li}$.  Next, we
764: normalize both fitted polynomials by dividing by the minimum
765: $\chi^{2}$ value for each spectral region.  The two normalized curves
766: are then added together to give an average curve, and the minimum of
767: this curve is what we take to be the best-fit $\log N_\mathrm{Li}$.
768: Thus, in our procedure for determining the lithium abundances, 
769: we give equal weight to both regions, unlike
770: the procedure outlined in \S\ref{s:methods}, where we desired to keep
771: the weight on the best fitted regions by summing the raw $\chi^{2}$
772: values.
773: 
774: By way of example, we show, in Fig.~\ref{fig:lifit-montage}, the
775: spectral fits to the two lithium features for three stars with
776: distinctly different lithium abundances.  The resulting normalized
777: $\chi^2$ curves are shown in the bottom panel for each star.  Averages
778: of the best-fit $\log N_\mathrm{Li}$ for all stars can be found in
779: Tables~\ref{tab:TWA}~to~\ref{tab:RVstd}, with uncertainties
780: representing the error in the mean between multi-epoch observations.
781: 
782: As discussed in \S\ref{s:LiRes}, uncertainties in the model atmospheres
783: lead to larger errors than the internal errors quoted in 
784: Tables~\ref{tab:TWA}~to~\ref{tab:RVstd}. Thus, we also handle external 
785: errors for the lithium abundances with the same approach. We 
786: investigate changes in the fitted abundances by re-fitting the 
787: lithium lines to models with $T_{\mathrm{eff}}$ perturbed by $\pm 100$
788: from the initial best-fit $T_{\mathrm{eff}}$ and also to models with
789: $\log g$ also perturbed by $\pm 0.5$\,dex. We find that this 
790: parameters leads to changes in  $\log N_\mathrm{Li}$ of 0.15 on average. 
791: We show this error on all relevant figures to indicate our estimate of the
792: external uncertainties.
793: 
794:  %
795: \begin{figure}[ht]
796: \begin{center}
797: \plotone{f7.eps}
798: \caption{Lithium abundance derived by fitting the 6\,104\,\AA\ and
799:   6\,708\,\AA\ spectral features as a function of temperature. We 
800:   identify the ultrafast rotators and accretors in our
801:   sample with black circles and squares, respectively.\label{fig:li-vs-t}}
802: \end{center}
803: \end{figure}
804: %
805: 
806: \subsection{Ages from the lithium abundance versus temperature isochrones}\label{s:ages}
807: 
808: As a PMS group ages, the lithium abundances of group members deplete
809: as a function of luminosity and time (see \S\ref{s:intro}). As with EWs in \S\ref{s:ew}, 
810: we can order the groups in age using the relative depletion of
811: members of different groups and by comparison with models, 
812: we can also determine absolute ages.  Of course, the
813: absolute ages will only be as good as the models. To get an idea of 
814: the associated uncertainty, we try both the models of \citetalias{bar98}
815: and \citet{sie00}.
816: 
817: In Fig.~\ref{fig:li-vs-t}, we show the distribution of measured
818: lithium abundance as a function of temperature for each of the groups
819: in this study, as well as for the field stars (which should be fully
820: depleted).  As with the EWs in \S\ref{s:ew}, it is easy to order the groups
821: chronologically based on the lithium depletion distribution: from
822: oldest to youngest, we again find the order ABD, TUCHOR, BPMG, and
823: then $\eta$\,Cha and TWA both at about the same age. It is also clear
824: that the field stars are older than all of the PMS groups in this
825: study.
826: 
827: In Fig.~\ref{fig:li-vs-t-montage}, the dependence of $\log
828: N_{\mathrm{Li}}$ on stellar $T_{\mathrm{eff}}$ is shown for each group
829: individually, and a range of PMS isochrones from \citetalias{bar98} 
830: (solid lines) and \citet{sie00} (dashed lines).  Both models use a convection
831: mixing length of $\alpha_{\mathrm{MLT}}=1.9$  and are scaled to the initial lithium 
832: abundance. We chose an initial lithium abundance of  
833: $\log N_{\mathrm{Li}} = 3.7$ to match the abundance we measured
834: from our method (see \S\ref{s:Li-method}) for the majority of undepleted
835:  stars in the entire sample. The inferred ages do depend on this
836:  choice, as well as the choice of model.
837:  
838:  From Fig.~\ref{fig:li-vs-t-montage}, we can infer ages by eye for each group.
839:  We tried but decided not to use quantitative analysis, because it is 
840:  apparent that the models do not reproduce the data to enough accuracy,
841:  especially in the temperature range where the lithium abundances are 
842:  most dependent on age. 
843: In general, we find that the models of \citetalias{bar98}
844:  fit the nature of the depletion slightly better for all of the groups, 
845:  but that the ages derived are similar, apart from a small
846: systematic offset (with \citetalias{bar98} giving slightly higher ages).
847: 
848:   We find that a lithium depletion age of $12{\pm6}$\,Myr for $\eta$\,Cha 
849:  using the models of  \citetalias{bar98} ($12{\pm8}$\,Myr using the 
850:  \citet{sie00} models; hereafter given in parentheses), and and 
851:  an age of $12{\pm8}$\,Myr ($12{\pm8}$\,Myr) for TWA,
852: consistent with dynamical expansion ages \citep{jil05,del06} and other
853: age estimates also based on \citetalias{bar98} PMS models 
854: \citep{luh04,zuc04,bar06}.  For BPMG, we find an age of
855: $21{\pm9}$\,Myr ($13{\pm5}$\,Myr), which is in agreement with the estimate of
856: 9--17\,Myr from other methods \citep{zuc04,fei06}.  It is slightly
857: higher than its dynamical expansion age of 11.5\,Myr from
858: \citet{ort02}, but agrees with a different age based on lithium dating
859: of 10--20\,Myr recently found by \citet{mam07}.  For TUCHOR, we find
860: an age of $27{\pm11}$\,Myr ($22{\pm10}$\,Myr), which is consistent with
861: all previous age estimates for this group \citep{zuc00,ste00}. 
862: 
863: For ABD, we find that it is clearly older than TUCHOR and
864: clearly younger than the field stars, however the age estimate from
865: PMS models is poorly constrained.  Although the field stars show
866: more depletion than ABD, this is not predicted by the PMS isochrones;
867: both models show no depletion after $\sim\!45$\,Myr for stars
868: with $4\,000<T_{\mathrm{eff}}<6\,000$\,K. Until the PMS evolutionary
869: models are improved, the best way to find an
870: upper limit to the age of the ABD group would be to use the cool end of the LDB.
871: This would require stars with spectral types later than M3
872: ($T_{\mathrm{eff}}\lesssim$3\,300\,K), but, unfortunately, no such
873: members are known in ABD.  Within our present large uncertainties, our
874: age estimate is consistent with both a younger estimate of 50\,Myr
875: based on H$\alpha$ emission strength \citep{zuc04b}, as well as an
876: older one of 100--140\,Myr from HR isochrones \citep{luh05} and
877: dynamical expansion \citep{ort07}.
878: 
879: We close with a number of notes.  First, while the poor fit of the
880: data to the models leads to rather large uncertainties on the ages,
881: these should be considered overall shifts: the age ordering of the
882: groups is secure.  Second, an additional uncertainty in 
883: the derived ages is our choice of initial
884: lithium abundance, of $\log N_{\mathrm{Li}}=3.7$ based
885: on our observations. Decreasing the initial lithium abundance to
886: $\log N_{\mathrm{Li}}=3.3$ (as used in \citet{jef05}), predicts 
887: younger group ages by about 5 Myrs. On the other hand, using a higher
888: initial lithium abundance of $\log N_{\mathrm{Li}} = 4.0$, yields
889: ages larger by 5-10 Myrs. Third, we have ignored non-LTE effects in the lithium lines
890: (\S\ref{s:models}).  While our scatter is larger than the predicted
891: effects, the systematic changes with temperature and abundance will
892: lead to additional systematic age differences.  
893: It also may be the underlying reason for our need for
894: a relatively high initial abundance: \citet{car94} found that around
895: 6000\,K, the correction for the 6708 line is about $-0.3\,$dex, which
896: would imply initial abundances more in line with expectations (at
897: these temperatures, the 6\,104\,\AA\ line is very weak and contributes little to our
898: fits.  
899: 
900: %
901: \begin{figure*}[t]
902: \begin{center}
903: \includegraphics[width=6.5in]{f8.eps}
904: \caption{Lithium abundances derived from fits to the spectra as a
905:   function of temperature for each group.  Overdrawn are predictions
906:   from the evolutionary models of \citetalias{bar98} (solid line) and 
907:   \citet{sie00} (dashed line) as indicated by the ages in 
908:   the bottom-right corner of each plots. Above the ages, 
909:   external errors are shown. In addition, the ultrafast rotators are
910:   identified by black circle outlines, and accretors are identified
911:   by black square outlines.  \label{fig:li-vs-t-montage}}
912: \end{center}
913: \end{figure*}
914: 
915: \subsection{The Effect of Rotation on Lithium Depletion}\label{s:rotation}
916: 
917: We examine the effect of stellar rotation on lithium depletion using 
918: projected rotational velocities found previously from our observations
919: \citep{jay06,sch07}. In Fig.~\ref{fig:li-vs-t-montage}, we identify 
920: ultrafast rotators as those stars that have $v$\,$\sin i>70$\,km\,s$^{-1}$.
921: One member (TWA 6) from TWA, two members (PZ\,Tel, CD\,64d1208) of BPMG and 
922: three members (CD\,53\,544, HIP\,108422, HIP\,2729) of TUCHOR are identified
923: as ultrafast rotators. In three of the cooler stars ($T_\mathrm{eff}\sim$3800\,K), CD\,64d1208, 
924: CD\,53\,544 and HIP\,2729 , the lithium EWs are noticeably higher, 
925: and the derived abundances larger, than the trend in lithium depletion for 
926: the entire group. The correlation between fast rotation and slower lithium 
927: depletion has also been seen previously in a sample of weak-line T Tauri stars 
928: \citep{mart94} and in the 115\,Myr Pleiades cluster \citep{sod93,gar94}. 
929: It may be related to rapidly rotating stars being relatively cooler as the rapid 
930: rotation inhibits convection \citep{cha07}. This alternative view is supported by the location of the rotators in Fig.~\ref{fig:fitt_vs_sptt}. It is evident that the $T_\mathrm{eff}$ 
931: derived for these rotators is 100--300\,K cooler than the temperature 
932: derived from their spectral types. It may be that the presence of a colder 
933: equatorial region and a hotter polar one affects the model fits differently than
934: the spectral typing.
935: 
936: However, the trend that lithium depletion is slowed down by rotation
937: is not seen in all of our ultrafast rotators. The relatively slower rotators 
938: PZ\,Tel ($v$\,$\sin i=77.5$\,km\,s$^{-1}$) and TWA\,6 ($v$\,$\sin i=79.5$\,km\,s$^{-1}$)
939: have lithium abundances comparable to other members in their groups. 
940: For our faster rotator, HIP\,108422 ($v$\,$\sin i=$139.8\,km\,s$^{-1}$),
941: the spectrum is so strongly broadened that our fitting method
942: does a poor job and we see no change in the quality of fit
943: for varying lithium abundances. For this, reason, we do not quote an abundance
944: for this object, but do note, however, that the equivalent width measured is consistent with 
945: other group members.
946: 
947: \subsection{Notes on Individual Systems}
948: 
949: \textit{BD\,17$\,^{\circ}$6128 --} This binary system from BPMG
950: consists of a K7 primary with $T_{\mathrm{eff}}=$4\,140\,K and 
951: $\log N_{\mathrm{Li}}=2.78$ and a lithium depleted secondary of 
952: $T_\mathrm{eff}=$3\,350\,K and $\log N_{\mathrm{Li}}=0.54$. This
953: system is unique to our sample as it is the only case of the cool end
954: of the LDB in effect within a binary, and provides a precise, if model 
955: dependent, age of the system. Using the models from \citetalias{bar98}, an age of
956: 15--50\,Myr is predicted, consistent with the age we inferred for BPMG
957: as a group. 
958: 
959:  \textit{GSC\,08056-0482 --} The measured $\log N_{\mathrm{Li}}$ for TUCHOR member
960: GSC\,08056-0482 is much higher than expected for a $\sim\!30$\,Myr
961: old, M3 dwarf.  Indeed, another M3 star in the younger BPMG,
962: GSC\,08491-1194, has an abundance over two orders of magnitude smaller
963: (consistent with models at $\sim\!20\,$Myr). With a modest $v\sin i$
964: of $34.2{\rm\,km\,s^{-1}}$ \citep{jay06}, rotation does not explain
965: the high lithium abundance. As pointed out already by previous authors
966: (see Table 3 of \citealt{zuc04}), the lithium abundance suggests that
967: GSC\,08056-0482 is likely not a member of TUCHOR, but a star slightly
968: older than TWA and $\eta$\,Cha, but definitely younger than BPMG.
969: 
970: \section{Summary and outlook}\label{s:conclusions}
971: 
972: We have measured effective temperatures, surface gravities, lithium
973: equivalent widths and lithium abundances for 121 low-mass 
974: PMS stars from five nearby, PMS
975: groups ranging in age from 8--125\,Myr by performing least-squares
976: fits of high resolution spectra to synthetic spectra created from
977: PHOENIX model atmospheres \citep{hau98}. To investigate the
978: reliability of our measurements we compare the derived
979: $T_\mathrm{eff}$ and $\log g$ with isochrones from PMS evolutionary
980: models \citepalias{bar98} as well as temperatures derived from
981: spectral types. 
982: 
983: Isochrones from PMS models for $\log N_{\mathrm{Li}}$ as a function of
984: $T_\mathrm{eff}$ are visually compared to the observed distribution. 
985: We find agreement between ages derived from PMS isochrones of
986: \citetalias{bar98} and \citet{sie00} to ages calculated from other methods such
987: as dynamical expansion ages. We find that $\eta$\,Cha and TWA have
988: ages of $12{\pm6}$\,Myr and $12{\pm8}$\,Myr,
989: respectively. BPMG has an age of $21{\pm9}$\,Myr, and TUCHOR has
990: an age of $27{\pm11}$\,Myr. We can only constrain a tight lower
991: limit for ABD, with an age greater than $45$\,Myr, since, according
992: to the PMS models, there is no more lithium depletion after
993: $\sim\!45$\,Myr for stars with $4\,000<T_{\mathrm{eff}}<6\,000\,$K.
994: However, the halting of lithium depletion at this age and temperature
995: is inconsistent with observations of radial velocity standards which 
996: demonstrate more depletion than the ABD group and the model predictions.
997: Finally, we find that some of the ultrafast rotators in our sample have
998: significantly less lithium depletion than other stars in the same
999: group at the same temperature.
1000: 
1001: The consistent determination of $T_\mathrm{eff}$ and $\log g$ between
1002: multiple epochs ($\sigma_T \simeq 10$\,K, $\sigma_{\log g} \simeq
1003: 0.05\,$dex) means that, in principle, we should be able to constrain
1004: those parameters with this precision.  As revealed by
1005: Fig.~\ref{fig:logg_vs_t}, however, there is an apparent
1006: systematic offset between the $\log g$ inferred from model spectra and
1007: $\log g$ expected from models of stellar evolution
1008: (\S\ref{s:LiRes}). To account for these offsets, we introduce rough
1009: conservative external errors by examining how the measured parameters
1010: depend on each other. We find that the systematic errors in $\log g$ of 
1011: 0.5\,dex lead to systematic errors of 100\,K in our ability to constrain the 
1012: $T_{\mathrm{eff}}$. These external errors, similarly lead to offsets in the 
1013: measured lithium abundances of 0.15\,dex. 
1014: 
1015: The small internal errors that we have measured imply that currently our 
1016: accuracy is limited by the models.  With further improvements in the atmospheric
1017: models, there is a potential of comparing $T_\mathrm{eff}$ and $\log
1018: g$ directly to evolutionary models, thereby finding an age constraint
1019: independent of other estimators, such as color-magnitude diagram or
1020: lithium depletion boundary fitting.  Our data set would be well-suited
1021: for use with such future improved models.  Another use of our data set
1022: would be to derive both overall metallicity and abundances for
1023: individual elements.  While we do not believe this would affect our
1024: derived temperatures, etc., to a significant degree, it may be
1025: interesting to see how uniform the abundances are within (and between)
1026: groups, and whether there is any dependence on binarity, etc.
1027: 
1028: \acknowledgements
1029: We thank the referee for an excellent and thorough review, which
1030: helped to improve our manuscript greatly.  We also thank the
1031: outstanding support staff at Magellan for their assistance  
1032: during multiple observing runs.  This work was supported in part by NSERC 
1033: and the DFG (via Graduiertenkolleg 1351).  Some of the calculations presented 
1034: here were performed at the H\"ochstleistungs Rechenzentrum Nord (HLRN); at the
1035: NASA's Advanced Supercomputing Division's Project Columbia, at the
1036: Hamburger Sternwarte Apple G5 and Delta Opteron clusters financially
1037: supported by the DFG and the State of Hamburg; and at the National
1038: Energy Research Supercomputer Center (NERSC), which is supported by
1039: the Office of Science of the U.S. Department of Energy under Contract
1040: No.\ DE-AC03-76SF00098. We thank all these institutions for a generous
1041: allocation of computer time.
1042: 
1043: \begin{thebibliography}{}
1044: \bibitem[Asplund et al.(2005)]{asp05} Asplund, M., Grevesse, 
1045: N., \& Sauval, A.~J.\ 2005, Cosmic Abundances as Records of Stellar 
1046: Evolution and Nucleosynthesis, 336, 25 
1047: \bibitem[Baraffe et al.(1998)]{bar98} Baraffe, I., Chabrier, 
1048: G., Allard, F., \& Hauschildt, P.~H.\ 1998, \aap, 337, 403 
1049: \bibitem[Barrado y Navascu{\'e}s et al.(1999)]{bar99} Barrado 
1050: y Navascu{\'e}s, D., Stauffer, J.~R., \& Patten, B.~M.\ 1999, \apjl, 522, 
1051: L53
1052: \bibitem[Barrado y Navascu{\'e}s et al.(2004)]{bar04} Barrado 
1053: y Navascu{\'e}s, D., Stauffer, J.~R., \& Jayawardhana, R.\ 2004, \apj, 614, 
1054: 386 
1055: \bibitem[Barrado y Navascu{\'e}s(2006)]{bar06} Barrado y 
1056: Navascu{\'e}s, D.\ 2006, \aap, 459, 511 
1057: \bibitem[Bildsten et al.(1997)]{bil97} Bildsten, L., Brown, 
1058: E.~F., Matzner, C.~D., \& Ushomirsky, G.\ 1997, \apj, 482, 442 
1059: \bibitem[Chabrier et al.(2007)]{cha07} Chabrier, G., Gallardo, 
1060: J., \& Baraffe, I.\ 2007, \aap, 472, L17 
1061: \bibitem[Carlsson et al.(1994)]{car94} Carlsson, M., Rutten, 
1062: R.~J., Bruls, J.~H.~M.~J., \& Shchukina, N.~G.\ 1994, \aap, 288, 860 
1063: \bibitem[de la Reza et al.(2006)]{del06} de la Reza, R., 
1064: Jilinski, E., \& Ortega, V.~G.\ 2006, \aj, 131, 2609 
1065: \bibitem[Eisl\"{o}ffel et al.(1990)]{eis90} Eisl\"{o}ffel, J., 
1066: Solf, J., \& Boehm, K.~H.\ 1990, \aap, 237, 369
1067: \bibitem[Feigelson et al.(2006)]{fei06} Feigelson, E.~D., 
1068: Lawson, W.~A., Stark, M., Townsley, L., \& Garmire, G.~P.\ 2006, \aj, 131, 
1069: 1730 
1070: \bibitem[Fuhrmeister et al.(2005)]{fuh05} Fuhrmeister, B., 
1071: Schmitt, J.~H.~M.~M., \& Hauschildt, P.~H.\ 2005, \aap, 439, 1137 
1072: \bibitem[Garc\'{i}a L\'{o}pez et al.(1994)]{gar94} Garc\'{i}a L\'{o}pez, 
1073: R.~J., Rebolo, R., \& Mart\'{i}n, E.~L.\ 1994, \aap, 282, 518 
1074: \bibitem[Hauschildt \& Baron(1999)]{hau98} Hauschildt, P.~H. \&
1075:  Baron, E.\ 1999, \ Journal of Computational and Applied Mathematics, 109, 41
1076: \bibitem[Jayawardhana et al.(2006)]{jay06} Jayawardhana, R., 
1077: Coffey, J., Scholz, A., Brandeker, A., \& van Kerkwijk, M.~H.\ 2006, \apj, 
1078: 648, 1206
1079: \bibitem[Jeffries \& Oliveira(2005)]{jef05} Jeffries, R.~D., 
1080: \& Oliveira, J.~M.\ 2005, \mnras, 358, 13 
1081: \bibitem[Jilinski et al.(2005)]{jil05} Jilinski, E., Ortega, 
1082: V.~G., \& de la Reza, R.\ 2005, \apj, 619, 945 
1083: \bibitem[Kirkpatrick et al.(1991)]{kir91} Kirkpatrick, J.~D., 
1084: Henry, T.~J., \& McCarthy, D.~W., Jr.\ 1991, \apjs, 77, 417
1085: \bibitem[Ku{\v c}inskas et al.(2005)]{kuc05} Ku{\v c}inskas, 
1086: A., Hauschildt, P.~H., Ludwig, H.-G., Brott, I., Vansevi{\v c}ius, V., 
1087: Lindegren, L., Tanab{\'e}, T., \& Allard, F.\ 2005, \aap, 442, 281 
1088: \bibitem[Ku{\v c}inskas et al.(2006)]{kuc06} Ku{\v c}inskas, 
1089: A., Hauschildt, P.~H., Brott, I., Vansevi{\v c}ius, V., Lindegren, L., 
1090: Tanab{\'e}, T., \& Allard, F.\ 2006, \aap, 452, 1021 
1091: \bibitem[Lawson \& Crause(2005)]{law05} Lawson, W.~A., \& 
1092: Crause, L.~A.\ 2005, \mnras, 357, 139 
1093: \bibitem[Ludwig et al.(2006)]{lud06} Ludwig, H.-G., Allard, 
1094: F., \& Hauschildt, P.~H.\ 2006, \aap, 459, 599 
1095: \bibitem[Luhman \& Rieke(1998)]{luh98} Luhman, K.~L., \& 
1096: Rieke, G.~H.\ 1998, \apj, 497, 354 
1097: \bibitem[Luhman et al.(2005)]{luh05} Luhman, K.~L., Stauffer, 
1098: J.~R., \& Mamajek, E.~E.\ 2005, \apjl, 628, L69 
1099: \bibitem[Luhman \& Steeghs(2004)]{luh04} Luhman, K.~L., \& 
1100: Steeghs, D.\ 2004, \apj, 609, 917 
1101: \bibitem[Makarov et al.(2005)]{mak05} Makarov, V.~V., Gaume, 
1102: R.~A., \& Andrievsky, S.~M.\ 2005, \mnras, 362, 1109 
1103: \bibitem[Mamajek et al.(1999)]{mam99} Mamajek, E.~E., Lawson, 
1104: W.~A., \& Feigelson, E.~D.\ 1999, \apjl, 516, L77
1105: \bibitem[Mamajek et al.(2007)]{mam07} Mamajek, E.~E., Barrado 
1106: y Navascu{\'e}s, D., Randich, S., Jensen, E.~L.~N., Young, P.~A., Miglio, A., 
1107: \& Barnes, S.~A.\ 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0702074 
1108: \bibitem[Mart\'in et al.(1994)]{mart94} Mart\'in, E.L., Rebolo, R., Magazz\`u, A. 
1109: \& Pavlenko, Ya.V. 1994, A\&A, 282, 503 
1110: \bibitem[Mathieu et al.(2007)]{mat07} Mathieu, R.~D., 
1111: Baraffe, I., Simon, M., Stassun, K.~G., \& White, R.\ 2007, Protostars and 
1112: Planets V, 411
1113: \bibitem[Mohanty et al.(2004)]{moh04} Mohanty, S., Basri, G., 
1114: Jayawardhana, R., Allard, F., Hauschildt, P., \& Ardila, D.\ 2004, \apj, 
1115: 609, 854
1116: \bibitem[Ortega et al.(2002)]{ort02} Ortega, V.~G., de la 
1117: Reza, R., Jilinski, E., \& Bazzanella, B.\ 2002, \apjl, 575, L75
1118: \bibitem[Ortega et al.(2007)]{ort07} Ortega, V.~G., Jilinski, 
1119: E., de la Reza, R., \& Bazzanella, B.\ 2007, ArXiv Astrophysics e-prints, 
1120: arXiv:astro-ph/0702315 
1121: \bibitem[Palla \& Stahler(1999)]{pal99} Palla, F., \& Stahler, S.~W.\ 1999, \apj, 525, 772 
1122: \bibitem[Press et al.(1992)]{pre92} Press, W.~H., Teukolsky, 
1123: S.~A., Vetterling, W.~T., \& Flannery, B.~P.\ 1992, Cambridge: University 
1124: Press, |c1992, 2nd ed., 
1125: \bibitem[Rucinski(2002)]{ruc02} Rucinski, S.~M.\ 2002, \aj, 124, 1746
1126: \bibitem[Scholz et al.(2007)]{sch07} Scholz, A., Coffey, J., 
1127: Brandeker, A., \& Jayawardhana, R.\ 2007, ArXiv e-prints, 704, 
1128: arXiv:0704.3266 
1129: \bibitem[Siess et al.(2000)]{sie00} Siess, L., Dufour, E., \& Forestini, M.\ 2000, \aap, 358, 593 
1130: \bibitem[Soderblom et al.(1993)]{sod93} Soderblom, D.~R., 
1131: Jones, B.~F., Balachandran, S., Stauffer, J.~R., Duncan, D.~K., Fedele, 
1132: S.~B., \& Hudon, J.~D.\ 1993, \aj, 106, 1059 
1133: \bibitem[Song et al.(2003)]{son03} Song, I., Zuckerman, B., 
1134: \& Bessell, M.~S.\ 2003, \apj, 599, 342 
1135: \bibitem[Stahler \& Palla(2005)]{sta05} Stahler, S.~W., \& 
1136: Palla, F.\ 2005, The Formation of Stars, by Steven W.~Stahler, Francesco 
1137: Palla, pp.~865.~ISBN 3-527-40559-3.~Wiley-VCH , January 2005., 
1138: \bibitem[Stassun et al.(2004)]{sta04} Stassun, K.~G., 
1139: Mathieu, R.~D., Vaz, L.~P.~R., Stroud, N., \& Vrba, F.~J.\ 2004, \apjs, 
1140: 151, 357 
1141: \bibitem[Stauffer et al.(1998)]{sta98} Stauffer, J.~R., 
1142: Schultz, G., \& Kirkpatrick, J.~D.\ 1998, \apjl, 499, L199
1143: \bibitem[Stauffer et al.(1999)]{sta99} Stauffer, J.~R., et 
1144: al.\ 1999, \apj, 527, 219 
1145: \bibitem[Stelzer \& Neuh{\"a}user(2000)]{ste00} Stelzer, B., 
1146: \& Neuh{\"a}user, R.\ 2000, \aap, 361, 581
1147: \bibitem[Torres et al.(2000)]{tor00} Torres, C.~A.~O., da 
1148: Silva, L., Quast, G.~R., de la Reza, R., \& Jilinski, E.\ 2000, \aj, 120, 
1149: 1410 
1150:  \bibitem[Torres-Dodgen \& Weaver(1993)]{tor93} Torres-Dodgen, 
1151: A.~V., \& Weaver, W.~B.\ 1993, \pasp, 105, 693
1152: \bibitem[Zapatero Osorio et al.(2002)]{zap02} Zapatero 
1153: Osorio, M.~R., B{\'e}jar, V.~J.~S., Pavlenko, Y., Rebolo, R., Allende 
1154: Prieto, C., Mart{\'{\i}}n, E.~L., \& Garc{\'{\i}}a L{\'o}pez, R.~J.\ 2002, 
1155: \aap, 384, 937 
1156: \bibitem[Zuckerman \& Webb(2000)]{zuc00} Zuckerman, B., \& 
1157: Webb, R.~A.\ 2000, \apj, 535, 959 
1158: \bibitem[Zuckerman et al.(2001)]{zuc01} Zuckerman, B., Song, 
1159: I., Bessell, M.~S., \& Webb, R.~A.\ 2001, \apjl, 562, L87 
1160: \bibitem[Zuckerman \& Song(2004)]{zuc04} Zuckerman, B., \& 
1161: Song, I.\ 2004, \araa, 42, 685 
1162: \bibitem[Zuckerman et al.(2004)]{zuc04b} Zuckerman, B., Song, 
1163: I., \& Bessell, M.~S.\ 2004, \apjl, 613, L65 
1164: \end{thebibliography}
1165: 
1166: \clearpage
1167: 
1168: \begin{deluxetable}{llcllll}
1169: \tablecaption{Results for stars in TW~Hydrae}
1170: \tablecolumns{7}
1171: \tablewidth{0pc}
1172: %\tabletypesize{\small}
1173: \tablehead{
1174: 	\colhead{Object ID} 			&
1175: 	\colhead{Sp.T.} 			&
1176: 	\colhead{\# Obs.}			&
1177: 	\colhead{EW$_{6708}$ (\AA)} &
1178: 	\colhead{$T_{\mathrm{eff}}$ (K)} 	&
1179: 	\colhead{$\log g$ }			&
1180: 	\colhead{$\log N_{\mathrm{Li}}$}		
1181: }
1182: \startdata
1183: \input{tab1.tex}
1184: \enddata
1185: \tablecomments{all uncertainties are internal, derived from the scatter of fitted values from individual spectra from the mean. The external uncertainties are much larger (see \S\ref{s:LiRes}).}
1186: \tablenotetext{a}{\citet{zuc04}}
1187: \tablenotetext{b}{\citet{del06}}
1188: \label{tab:TWA}
1189: \end{deluxetable}%
1190: 
1191: \begin{deluxetable}{llcllll}
1192: \tablecaption{Results for stars in $\eta$~Chamaeleontis}
1193: \tablecolumns{7}
1194: \tablewidth{0pc}
1195: \tabletypesize{\small}
1196: \tablehead{
1197: 	\colhead{Object ID} 			&
1198: 	\colhead{Sp.T.} 			&
1199: 	\colhead{\# Obs.}			&
1200: 	\colhead{EW$_{6708}$ (\AA)} &
1201: 	\colhead{$T_{\mathrm{eff}}$ (K)} 	&
1202: 	\colhead{$\log g$ }			&
1203: 	\colhead{$\log N_{\mathrm{Li}}$}		
1204: }
1205: \startdata
1206: \input{tab2.tex}
1207: \enddata
1208: \tablecomments{all uncertainties are internal, derived from the scatter of fitted values from individual spectra from the mean. The external uncertainties are much larger (see \S\ref{s:LiRes}).}
1209: \tablenotetext{a}{\citet{zuc04}}
1210: \tablenotetext{b}{\citet{luh04}}
1211: \label{tab:etaCha}
1212: \end{deluxetable}%
1213: 
1214: \begin{deluxetable}{llcllll}
1215: \tablecaption{Results for stars in the $\beta$~Pictoris moving group (BPMG)}
1216: \tablecolumns{7}
1217: \tablewidth{0pc}
1218: \tabletypesize{\small}
1219: \tablehead{
1220: 	\colhead{Object ID} 			&
1221: 	\colhead{Sp.T.} 			&
1222: 	\colhead{\# Obs.}			&
1223: 	\colhead{EW$_{6708}$ (\AA)} &
1224: 	\colhead{$T_{\mathrm{eff}}$ (K)} 	&
1225: 	\colhead{$\log g$ }			&
1226: 	\colhead{$\log N_{\mathrm{Li}}$}		
1227: }
1228: \startdata
1229: \input{tab3.tex}
1230: \enddata
1231: \tablecomments{all uncertainties are internal, derived from the scatter of fitted values from individual spectra from the mean. The external uncertainties are much larger (see \S\ref{s:LiRes}). All spectral types from \citet{zuc04}}
1232: \label{tab:BPMG}
1233: \end{deluxetable}%
1234: 
1235: \begin{deluxetable}{llcllll}
1236: \tablecaption{Results for stars in Tucanae--Horologium (TUCHOR)}
1237: \tablecolumns{7}
1238: \tablewidth{0pc}
1239: \tabletypesize{\small}
1240: \tablehead{
1241: 	\colhead{Object ID} 			&
1242: 	\colhead{Sp.T.} 			&
1243: 	\colhead{\# Obs.}			&
1244: 	\colhead{EW$_{6708}$ (\AA)} &
1245: 	\colhead{$T_{\mathrm{eff}}$ (K)} 	&
1246: 	\colhead{$\log g$ }			&
1247: 	\colhead{$\log N_{\mathrm{Li}}$}		
1248: }
1249: \startdata
1250: \input{tab4.tex}
1251: \enddata
1252: \tablenotetext{a}{Lithium abundance could not be measured due to high rotational broadening in HIP\,108422 (v$\sin$\,i = 139.80 km/s)}
1253: \tablecomments{all uncertainties are internal, derived from the scatter of fitted values from individual spectra from the mean. The external uncertainties are much larger (see \S\ref{s:LiRes}). All spectral types from \citet{zuc04}}
1254: \label{tab:TH}
1255: \end{deluxetable}%
1256: 
1257: \begin{deluxetable}{llcllll}
1258: \tablecaption{Results for stars in AB Doradus (ABD)}
1259: \tablecolumns{7}
1260: \tablewidth{0pc}
1261: \tabletypesize{\small}
1262: \tablehead{
1263: 	\colhead{Object ID} 			&
1264: 	\colhead{Sp.T.} 			&
1265: 	\colhead{\# Obs.}			&
1266: 	\colhead{EW$_{6708}$ (\AA)} &
1267: 	\colhead{$T_{\mathrm{eff}}$ (K)} 	&
1268: 	\colhead{$\log g$ }			&
1269: 	\colhead{$\log N_{\mathrm{Li}}$}		
1270: }
1271: \startdata
1272: \input{tab5.tex}
1273: \enddata
1274: \tablecomments{all uncertainties are internal, derived from the scatter of fitted values from individual spectra from the mean. The external uncertainties are much larger (see \S\ref{s:LiRes}). All spectral types from \citet{zuc04}}
1275: \label{tab:ABD}
1276: \end{deluxetable}%
1277: 
1278: \begin{deluxetable}{llcllll}
1279: \tablecaption{Results for the field stars (radial velocity standards)}
1280: \tablecolumns{7}
1281: \tablewidth{0pc}
1282: \tabletypesize{\small}
1283: \tablehead{
1284: 	\colhead{Object ID} 			&
1285: 	\colhead{Sp.T.} 			&
1286: 	\colhead{\# Obs.}			&
1287: 	\colhead{$T_{\mathrm{eff}}$ (K)} 	&
1288: 	\colhead{$\log g$ }			&
1289: 	\colhead{$\log N_{\mathrm{Li}}$}		
1290: }
1291: \startdata
1292: \input{tab6.tex}
1293: \enddata
1294: \tablecomments{all uncertainties are internal, derived from the scatter of fitted values from individual spectra from the mean. The external uncertainties are much larger (see \S\ref{s:LiRes}).}
1295: \label{tab:RVstd}
1296: \end{deluxetable}%
1297: 
1298: \end{document}
1299: