arXiv:astro-ph/0008075v1 4 Aug 2000
Accepted to the Astronomical Journal July 21, 2000
Preprint typeset using L
A
T
E
X style emulateapj v. 20/04/00
THE GALACTIC THICK DISK STELLAR ABUNDANCES
JASON X. PROCHASKA
1,2
, SERGEI O. NAUMOV
3,4
, BRUCE W. CARNEY
3
, ANDREW
McWILLIAM
2
, & ARTHUR M. WOLFE
1,5
Accepted to the Astronomical Journal July 21, 2000
ABSTRACT
We present first results from a program to measure the chemical abundances of a large (N > 30)
sample of thick disk stars with the principal goal of investigating the formation history of the Galactic
thick disk. We have obtained high resolution, high signal-to-noise spectra of 10 thick disk stars with the
HIRES spectrograph on the 10m K e ck I telescope. Our a nalysis confir ms previous studies of O and Mg
in the thick disk stars which reported enhancements in ex cess of the thin disk population. Furthermore,
the observations of Si, Ca, Ti, Mn, Co, V, Zn, Al, and Eu all argue that the thick disk population has a
distinct chemical history from the thin disk. With the exce ption of V and Co, the thick disk abundance
patterns match or tend towards the values observed for halo stars with [Fe/H] ≈ − 1. This suggests that
the thick disk stars had a chemical enrichment history similar to the metal-rich halo stars. With the
possible exception of Si, the thick disk abundance patterns are in excellent agreement with the chemical
abundances observed in the metal-poor bulge stars suggesting the two populations formed from the same
gas reservoir at a common epoch.
The principal results of our analysis are as follows. (i) All 10 stars exhibit enhanced α/Fe ratios with
O, Si, and Ca showing tentative trends of decreasing overabundances with increasing [Fe/H]. In contrast,
the Mg and Ti enhancements are constant. (ii) The light elements Na and Al are enhanced in these stars.
(iii) With the exception of Ni, Cr and possibly Cu, the iron-peak elements show significant departures
from the s olar abunda nces. The stars are de ficient in Mn, but overabundant in V, Co, Sc, and Zn. (iv)
The heavy elements Ba and Y are consistent with solar abundances but Eu is significa ntly enhanced.
If the trends of decreasing O, Si, and Ca with increasing [Fe/H] are explained by the onset of Type Ia
SN, then the thick disk stars formed over the course of & 1 Gyr. We argue that this formation time-sca le
would rule out most dissipational collapse scenarios for the formation of the thick disk. Models which
consider the heating of an initial thin disk – either through ’gradual’ heating mechanisms or a sudden
merger event – are favored.
These observations provide new tests of theories of nucleosynthesis in the early universe. In particular,
the enhancements of Sc, V, Co, and Zn may imply overproduction during an enhanced α-rich freeze out
fueled by neutrino-driven winds. Meanwhile, the conflicting trends for Mg, Ti, Ca, Si, and O pose
a difficult challenge to our cur rent unders tanding of nucleosynthesis in Type Ia and Type II SN. The
Ba/Eu ratios favor r-process dominated enrichment for the heavy elements, consistent with the ages
(t
age
> 10 Gyr) expected for these stars.
Finally, we discuss the impact of the thick disk abundances on interpretations of the abundance
patterns of the damped Lyα systems. The observations of mildly enhanced Zn/Fe imply an interpretation
for the damped systems which includes a dust depletion pattern on top of a Type II SN enrichment
pattern. We also argue that the S/Zn ratio is not a good indicator of nucleosynthetic processes.
Subject headings: Gala xy: abundances — Galaxy:formation — stars: abundances — nuclear reactions,
nucleosynthesis, abundances
1. INTRODUCTION
The history of our Galaxy may be read through the long-
lived stellar relics of its past. In their landmark study,
Eggen et al. (1962) employed dynamica l and chemical
data to argue that the dynamica lly hot and metal-poor
halo stellar po pulation was the precursor of the dynam-
ically cool and metal-rich disk population. While this
1
Visiting Astronomer, W.M. Keck Telescope. The Keck Obser-
vatory is a joint facility of the University of California and the Cali-
fornia Institute of Technology.
2
The Observatories of the Carnegie Institute of Washington, 813
Santa Barbara St., Pasadena, CA 91101
3
Department of P hysics and Astronomy, University of North Car-
olina, Chapel Hill, North Carolina 27599-3255
4
Present address: Rostov State University, Russia
5
Department of Physics, and Center for Astrophysics and Space
Sciences, University of California, San Diego, C–0424, La Joll a, CA
92093-0424
conclusion has been subjected to considerable debate, the
comparative study of the halo and the disk populations
is certainly the primary means by which we learn of the
Galaxy’s earliest history.
Gilmore & Reid (1983) offered the best evidence for the
existence of another Galactic stellar population, the thick
disk. Their data consolidated earlier but less well-formed
views of the “intermediate population II” class described
in the 1957 Vatican Conference (O’Connell 1958). The
reality of the thick disk population was in its turn hotly
debated (Bachall & Soneira 1984), but it is now gener-
ally regar ded as a separ ate population. The key historical
question is whether the thick disk is related to any or all of
the other Galactic s tellar populations: the halo, the bulge,
and the disk. T he first steps have been to determine the
basic parameters of the thick disk, including its age, its
chemical composition, and its dynamics/distribution. The
thick disk appears to be very old, based on the abrupt cut-
1
2 THE GALACTIC THICK DISK STELLAR ABUNDANCES
off in the numbers of stars bluer than the main sequence
turn-offs of similar metallicity globular clusters (Gilmore &
Wyse 19 87; Carney et a l. 1989; Gilmore, Wyse & Kuijken
1989). The mean metallicity of the thick disk population,
<[Fe/H]>, lies between −0.5 and −0.7 (Gilmore & Wyse
1985; C arney et al. 1989; Gilmore et al. 1995; Layden
1995a,b). The spread in metallicities o f thick disk stars
ranges from near solar to [Fe/H] ≈ −1, although claims
for much lower metallicities have been made (cf. Norris et
al. 1985; Morrison et al. 1990; Allen et al. 1991; Ryan &
Lambert 1995; Beers & Sommer-La rsen 1995; Twarog &
Anthony-Twarog 1996; Martin & Morrison 1998; Chiba &
Beers 2000). The “asymmetric drift” of the thick disk (the
amount by which it lags the circular orbit motion at the
solar Galactocentric distance) has been estimated to vary
between 20 and 50 km s
−1
(Carney et al. 1989; Morrison et
al. 1990; Majewski 1992; Beers & Sommer- L arsen 1995;
Ojha et al. 1996; C hiba & Beers 2000), although Ma-
jewski (1992), Chen (19 97), and Chiba & Beers (2000)
have argued that the value var ies with distance from the
Galactic plane. Values for the vertical velocity dispersion,
σ(W), are almost all near 40 km s
−1
(Norris 1986; Car-
ney et a l. 1989; Beers & Sommer-Larsen 1995; Ojha et
al. 1996; Chiba & Beers 2000), w hich implies a vertical
scale height of order 1 kpc or less. This may be compared
to the older stars of the thin disk, which obe y a density
distribution consistent with a vertical s cale height of about
300 pc. Although the thin disk is ≈ 10 times more massive
than the thick disk, at distances of 1 to 2 kpc above the
plane, the thick disk population dominates.
The properties of the thick disk thus place it between
those of the halo and the thin disk. In turn, one questions
whether it is closely related to either of them in terms of
the Galaxy’s chemical and dynamical evolution, or if it
might be the result of a merger event (se e Gilmore et a l.
1989 and Majewski 1993 for excellent reviews of the vari-
ous models). Most evolutionary models pr e dic t that there
should be continuity in the thick disk and disk dynami-
cal and chemical histories, and that thick disks should be
found in other galaxies. A merger scenario, conversely,
would require some degree of discontinuity in the chem-
ical and dynamical parameters of the thick disk and the
thin disk, or observations that indicate not all disk galax-
ies have thick disks. It is interesting in this regard that
very deep surface photometry of edge-on spirals reveals
thick disks in some cases (e.g., NGC 891: van der Kruit
& Searle 1981; Morriso n et al. 1997) but not in all cases
(e.g., NGC 5907: Morrison, Boroson, & Harding 1994;
NGC 4244: Fry et a l. 1999).
In this paper we study the problem using Galactic stars
whose motions ar e consistent with thick disk membership.
Our goal here is to compare their abundance patterns,
[X/Fe] vs. [Fe/H], with those of the other major Galactic
stellar populations: the halo, thin disk and bulg e. If the
histories o f the thick disk and the disk are closely related,
for example, so should be the derived chemical abunda nces
patterns vs. metallicity. It is well e stablished that very
metal-poor (halo) stars show enhanced levels of the light
“α”-rich nuc lei elements oxygen, magnesium, silic on, cal-
cium, and even titanium, but at a metallicity of [Fe/H] ≈
−1 the [α/Fe] va lue s begin to decline from +0.4 dex or so
to solar values at [Fe/H] = 0 (see Wheeler, Sneden, & Tr u-
ran 1989; McWilliam 1997). A fundamental compar ison
then is whether thick disk stars show similar [α/ Fe] and
other element abundance ratios at the same [Fe/H] values
as the thin disk stars. Similarities would favo r the “evo-
lutionary” history of the thick disk; discontinuities would
support a merger origin.
Large stellar samples with high-precision abundance anal-
yses have appeared over the last several years, which may,
in principle, answer this question. Edvardsson et al. (1993)
studied a large sample of F and G dwarfs and found that
lower metallicity stars had, in general, enhanced [α/Fe]
values, but they did not compare the thick disk and thin
disk stellar abundance patterns in detail. Gratton et al.
(1996, 2000) were the first to directly compare the abun-
dance ratios of a sample (≈ 15) of thick disk sta rs with
halo and thin disk populations. The ir measure ments of
Fe/O and Fe/Mg ratios indicate a stark difference in the
Fe/O and Fe/Mg abundance of the thick and thin disk
populations with the thick disk stars exhibiting halo-like
ratios. These authors argue for an early, rapid formation
of the Galactic thick disk, prior to the thin disk and per -
haps due to an early accretion event. Fuhrmann (1998)
also co mpared Mg/Fe ratios for a sample of thick and thin
disk stars taken from both Edvards son et al. (1993) and
his own smaller kinematic sample. Although the Edvards-
son et al. sample doe s not provide compelling evidence for
a discontinuity, the Fuhrmann sample shows strong evi-
dence for a disjunction, which further supports the as ser-
tion that the thick dis k and thin disk have not shared the
same che mical enrichment history. Most recently, Chen et
al. (2000 ) presented results from a sample of 90 F and
G dwarfs, which show no significant scatter in α-element
ratios as a function of [Fe/H], contrary to the results ob-
tained by Fuhrmann (1998). We contend, however, that
the sample selection employed by Chen et al. (2000) was
flawed for a program aimed at the study of the thick disk.
They chose to s tudy only stars with effective temperatures
between 5800 and 6400 K, so that few of their stars have
life expectancies as great as the thick disk’s a ge. As an
example, consider the disk, probably thick disk, globular
cluster 47 Tuc. Us ing the Alonso et al. (1996a) temp era-
ture scale (employed by Chen et al. 2000), the metallicity
of [Fe/H] = −0.70 (Carretta & Gratton 1997), and the
photometry of Hess er et al. (1987), we find that the tem-
peratures of main sequence turn-off stars in the cluster is
near 5970 K, only slightly hotter than the lower limit cut-
off for the Chen et al. (200 0) sample. And for more metal-
rich clusters or stars whose ages are as great as 47 Tuc, the
turn-off temper ature is even coole r. Thus if the thick disk
is composed almost exclusively of ancient stars, the Chen
et al. (200 0) sample cannot contain many thick disk stars.
Fuhrmann’s (1998) claimed thick disk stars, however, are
cool enough to have long enough life expectancies to be
considered part o f the thick disk. We believe that Chen et
al. (2000) have studied, primarily, the detailed chemical
evolution of the thin disk, which no doubt reaches to quite
low metallicity levels itself (see in particular the study r e-
garding the overlap in abundances of the thick disk and
the thin disk by Wyse & Gilmore 1995). We emphasize
that comparative studies of the thick disk and the thin
disk must employ stars with life expectancie s as grea t as
the thick disk’s age lest the trace but important popula-
PROCHASKA ET AL. 3
tion, the thick disk, be overlooked. We do so here. In a
future paper (Carney et al. 2000), we will study the re-
lation between kinematics and mean metallicity for long-
lived dwarf stars in the mid-plane, finding further evidence
for two distinct populations.
We have initia ted a program to measure the chemical
abundances of a large (N ∼ 5 0) sample of thick disk stars
at very high resolution (R ≈ 50, 000) with high signal-to-
noise ratio (S/N > 100), and a near ly continuous wave-
length coverage from λ ≈ 4400 − 90 00
˚
A. The stars were
selected from the surveys by Carney et al. (1994, 1996)
as exhibiting disk-like kinematics with la rge maximum dis-
tance from the Galactic plane. The current sample is co m-
prised of 10 stars, all brig hter than V = 10.5. We present a
detailed description of our stellar abundance a nalysis and
give first results on a small but meaningful s ample of stars.
In part, our goal is to build the analys is framework for fu-
ture observations . This initial sample, however, suggests
a number of exciting results which we will test through
a larger survey. In § 2 we describe the observations and
present a summary of the stellar parameters of the sam-
ple. The following section presents a thorough expla nation
of the techniques employed to measure the chemical abun-
dances of the program stars. In genera l, we follow standard
stellar analysis procedures. A solar analysis is discussed
in § 4 and § 5 gives an element-by-element account of the
results. Finally, § 6 compares the thick disk results against
other stellar populations and discusses the implications for
the formation history of the Galaxy, the damped Lyα sys-
tems, and nucleosynthesis in the ear ly universe.
2. OBSERVATIONS AND DATA REDUCTION
All of the observations presented here were carried out
in twilight time during an ongoing program to study high
redshift damped Lyα systems with the high resolution
echelle spectrograph (HIRES; Vogt et al. 1 994) on the
10m Keck I telescope. Table 1 summarizes the current
sample of program stars and pres ents our journal of obser-
vations. For each star we took multiple exposures at two
settings to achie ve nearly continuous wavelength coverage
from λ ≈ 44 00 − 900 0
˚
A with the exception of the inter-
order gaps longward of 5250
˚
A. The blue setup consisted
of the C1 decker which affords FWHM ≈ 6 km s
−1
resolu-
tion (≈ 2 km s
−1
per pixel) and the kv380 filter to block
second order light. For the red settings, we implemented
the longer C2 decker for improved sky subtraction and the
og530 filter to eliminate second order light. The typical
signal-to-noise is in excess of 100 per pixe l for all of the
spectra and > 200 for most of the stars. Standard ThAr
arc calibrations and quartz flats were taken for reduction
and calibration of the spectra.
The data were ex tracted and wavelength c alibrated with
the makee package deve lop ed by T. Barlow specifically for
HIRES observations. The algorithm perfor ms an optimal
extraction using the observed star to trace the profile, and
it solves for a wavelength calibration solution by cross-
correlating the extracted ThAr spectra with an extensive
database compiled by Dr. Barlow. The pairs of e xposures
were rebinned to the same wavelength scale and c oadded
conserving flux. Finally, we continuum fit the summed
spectra with a routine similar to IRAF, using the Arc-
turus spectrum (Griffin 1968) as a guide in the bluest
orders where the flux rare ly recovers to the continuum.
An example of a typical spectral order is presented in Fig-
ure 1 and we identify several representative absorption line
features.
The s ample of stars were kinematically selected from the
study of Carney et al. (1994) to be members of the thick
disk according to the following criteria. Our large initial
list excluded stars with any uncertain observational pa-
rameters, known subgiants, stars whose reddenings might
exceed 0.05 mag, and all stars known to be multiple-lined
or double-lined. To avoid stars whose lifetimes are shorter
than the age of the thick disk, we avoided stars with the
“TO” flag (meaning their colors place them near the main
sequence turn-off for globular clusters of similar metallic-
ity). We further restricted the list to stars with
−1.1 ≤ [M/H] ≤ −0.4 to probe the thick disk metallic-
ity regime, and likewise eliminated stars whose orbits did
not carry them farther than 600 pc from the plane. To
further maximize the probability of observ ing thick disk
stars within this sample, we restricted the
˜
V velocities to
lie betwe en −20 and −100 km s
−1
. While these criteria
help minimize the contamination of the thick disk sam-
ple from metal-rich halo stars and metal-poor thin disk
stars, these stellar populations do overlap in both metal-
licity and kinematic pr op erties and the possibility of con-
tamination exists. In general, the overlap between the
thick and thin disk populations is small (as determined
from proper motion studies; e.g. Carney et al. 1989) but
the problem deserves further obs e rvational attention. Ta-
ble 2 summarizes values of the obser ved stars’ photomet-
ric temperatures, high-resolution and low signal-to-noise
spectroscopic metallicity, stellar gravities determined with
Hipparcos meas urements (ESA 1997), and stellar kine-
matics and galactic orbital parameters from Carney et al.
(1994). All of the stars ar e G dwarfs found in the solar
neighborhood with distances of d
pc
= 50 − 100 pc. In
those cases where there are Hipparcos parallax measure-
ments, we calculated the stellar gravity according to the
Fig. 1.— Example of a single echelle order from the star G114-19.
The data is straight sum of two exposures taken with HIRES on the
Keck I telescope and normalized to unit flux. The dotted l ine in the
figure marks the 1σ error array.
4 THE GALACTIC THICK DISK STELLAR ABUNDANCES
Table 1
JOURNAL OF OBSERVATIONS
Star Alt Name HIP ID
a
RA (2000) DEC (2000) V Exp(blue)
b
Exp(red)
G66-51 15:00:50.0 +02:07:37 10.63 380 600
d
G84-37 HD 241253 24030 05:09:56.9 +05:33:26 9.72 350 500
d
G88-13 B+17 1524 34902 07:13:17.4 +17:26:01 10.10 800 800
d
G92-19 B-02 5072 96673 19:39:14.7 −02:36:44 10.31 500 600
d
G97-45 HD 36283 25860 05:31:13.7 +15:46:24 8.64 430 500
d
G114-19 HD 75530 43393 08:50:21.0 −05:32:09 9.19 500 600
c
G144-52 B+19 4601 103812 21:02:12.1 +19;54:03 9.07 600 300
d
G181-46 B+31 3025 85373 17:26:41.4 +31:03:34 9.68 400 600
d
G211-5 B+33 4117 103691 21:00:43.2 +33:53:20 9.62 600 600
d
G247-32 B+66 343 21921 04:42:50.2 +66:44:08 8.28 350 400
c
a
Hipparcos Identifier, ESA (1997)
b
λ
obs
= 4325 − 6760
˚
A
c
λ
obs
= 6380 − 8750
˚
A
d
λ
obs
= 6810 − 9200
˚
A
Table 2
STELLAR PHYSICAL PARAMETERS
Star V
˜
U
˜
V
˜
W Z
max
R
apo
R
per
d T
phot
[M/H] log g
(km/s) (km/s) (km/s) (kpc) (kpc ) (kpc) (pc) (K)
G66-51 10.63 +102 −72 −54 0.79 9.39 3.69 5196 −1.09
G84-37 9.72 −19 −61 +88 1.77 8.15 5.01 97 5898 −0.92 4.47
G88-13 10.10 −20 −41 −51 0.70 8.1 7 5.44 89 5069 −0.44 4.32
G92-19 10.31 +88 −70 −67 1.08 9.1 0 3.80 121 5433 −0.81 4.33
G97-45 8.64 +20 −52 −49 0.68 8.10 4.94 53 5429 −0.53 4.37
G114-19 9.19 −27 −85 −69 1.10 8.15 3.67 54 5218 −0.52 4.43
G144-52 9.07 +54 −11 +56 0.87 9.13 6.36 58 5497 −0.67 4.46
G181-46 9.68 +50 −80 +54 0.72 8.32 3.65 71 5277 −0.64 4.43
G211-5 9.62 −69 −21 −46 0.63 9.24 5.78 67 5196 −0.50 4.42
G247-32 8.28 −49 −52 +47 0.61 8.45 4.75 36 5270 −0.43 4.45
equations presented in Appendix A. The uncertainties in
the parallax and photometry imply an error in log(g/g
⊙
)
of ≈ 0.1 dex. In the following section, we will compute
spectroscopic physical parameters for each star and com-
pare with the photometric values presented here.
3. ABUNDANCE ANALYSIS
In this section we outline the prescription to mea sure
elemental abundances for our sample of thick dis k stars.
In short, we measured equivalent widths with the pack-
age getjob, implement Kurucz model stellar atmospheres,
culled log gf values from the literature, and used the stel-
lar analysis package MOOG to constrain the spectroscopic
physical parameters and determine the elemental abun-
dances.
3.1. Equivalent Widths
We first compiled a list of nearly 1000 reasonably un-
blended lines from the solar spectrum (Moore et al. 1966)
and an extensive literature search. The equivalent width,
W
λ
, for each line was then measured with the getjob pack-
age developed by A. McWilliam for stellar spectroscopic
analysis (McWilliam et al. 1995a). In the majority of
cases, we fit the absorption lines with single Gaussian pro-
files which provide an excellent match to the major ity of
observa tions. When necessary we fit regions with multi-
ple Gaussians or calculated an integrated e quivalent width
using Simpson’s Rule. The latter approach was particu-
larly important for strong Ca I and Mg I lines. The getjob
program yields an error estimate for each W
λ
value based
on the goodness-of-fit and signal-to-noise of the spectra.
The typical 1σ error for a single component fit is ≈ 2 m
˚
A.
With the exception of a few special cases , those lines with
errors exceeding 20% were eliminated fro m the subsequent
abundance analysis. We also focused on unsaturated lines,
specifically lines with W
λ
/λ < (100m
˚
A)/(5000
˚
A). Table 3
lists the W
λ
values for the measured absorption lines. We
have flagged those absorption lines which we believe are
blended or have incorrect gf values.
3.2. Model Stellar Atmospheres
PROCHASKA ET AL. 5
Table 3
EW MEASUREMENTS
Ion λ EP log gf Ref Sun G66-51 G84-37 G88-13 G92-19 G97-45 G114-19 G144-52 G181-46 G211-5 G247-32
(
˚
A) (eV) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A) (m
˚
A)
O I 7771.954 9.140 0.360 62 71.5 20.4 48.6 30.3 42.6 60.7 36.5 50.5 39.8 33.3 46.0
O I 7774.177 9.140 0.210 62 63.7 17.2 37.6 23.0 35.7 52.8 32.8 41.8 33.8 44.2
O I 7775.395 9.140 −0. 010 62 53.2 11.8 26.8 17.0 26.0 42.5 23.2 27.9 21.2 34.3
Na I 5682.647 2.100 −0.890 99 90.0 51.1 22.6 108.4 66.2 93.4 87.5 74.0 91.1 89.0
Na I 5688.210 2.100 −0.580 99 119.1 72.5 37.2 >120 86.1 113.2 95.5 102.8 115.2 115.0
References. — Key to References – 1: O’Brian et al. (1991); 2: Fuhr et al. (1988); 3: Blackwell et al. (1979a); 4: Blackwell et al. (1979b);
5: Blackwell et al. (1980); 6: Blackwell et al. (1982a); 8: Blackwell et al. (1982d); 9: Blackwell et al. (1982b); 10: Blackwel l et al. (1982c);
11: Blackwell et al. (1983); 12: Blackwell et al. (1984); 13: Blackwell et al. (1986a); 14: Blackwell et al. (1986b); 15: Blackwell et al. (1986c);
17: Martin et al. (1988); 18: Fry & Carney (1997); 19: B ard et al. (1991); 21: Bard & Kock (1994); 25: McWilliam & Rich (1994); 26:
Cardon et al. (1982); 27: Wiese & Martin (1980); 28: Garz (1973); 29: Bizzarri et al. (1993); 30: Smith & Raggett (1981); 31: Meylan et
al. (1993); 32: Bi´emont et al. (1991); 33: McWilliam et al. (1995b); 36: Hannaford & Lowe (1983); 37: Moity (1983); 38: Buurman et al.
(1986); 42: Kock & Richter (1968); 45: Lawler & Dakin (1989); 48: Lambert & Luck (1978); 54: Fuhrmann et al. (1995); 55: Wickliffe &
Lawler (1997); 56: Francois (1988); 57: Bi´emont & Godefroid (1980); 58: Blackwell et al. (1976); 60: Schnabel et al. (1999); 61: Kroll &
Kock (1987); 62: Butler & Zeippen (1991); 66: Booth et al. (1984a); 67: Savanov et al. (1990); 99: Solar gf (this work); 104: Edvardsson et
al. (1993); 106: Beveridge & Sneden (1994)
Note. — The complete version of this table is in the electronic edition of the Journal. The printed edition contains only a sample
Throughout the abundance analysis we adopt Kurucz
stellar atmospheres (Kurucz 1988) with convection on and
72 layers with optical depth steps, ∆τ = 0.125, ending at
τ = 100. Depending on the application, we either inter-
polated between the stellar atmos phere g rids kindly pro-
vided by R. Kurucz or implemented the Kurucz package
atlas9 to calculate specific models. The former approach
has the advantage that the interpolation can be performed
with minimal human intervention and a t minima l com-
putational cost. In particula r, we relied upo n the grids
to narrow in on the spectroscopic physical parameters of
each s tar (§ 3.4). When applicable we constructed mo de l
atmospheres with enhanced α-elements using the + 0.4 dex
enhanced Rosseland opacities and the appropriate opacity
distribution functions.
3.3. gf Values
Columns 4 and 5 of Table 3 list the ado pted gf values
and their references for our sample of measured absorp-
tion lines. In general, we selected the most accurate and
recent laboratory measurements available, avoiding solar
gf values where p ossible. Even with these accurate labo-
ratory values, however, the gf values pose a major source
of uncertainty in the analysis particularly w ith respect to
obtaining measurements relative to the solar meteoritic
abundances which will serve as our abunda nce refer e nc e
frame. We address this issue in § 4 by pe rforming an anal-
ysis of the solar spectrum. In the following, we discuss the
criteria established to select the Fe I and Fe II gf values
which are critical in determining the spectroscopic atmo-
spheric parameters of each star. We reserve comments on
the remaining elements to § 5.
To minimize the uncertainties and systematic errors as-
sociated with solar gf values, we restricted the Fe I analy-
sis to laboratory gf measurements. The principal sources
that we considered are: (1) the Hannover measurements
(Bard et al. 1991; Bard & Kock 1994), (2) the Oxford
gf values (Blackwell et al. 1995a), (3) the O’Brian values
Fig. 2.— Plots of ǫ(Fe) values versus Excitation Potential (EP) for
six different sets of Fe I gf values: (a) all samples with the priority
given in the text, (b) Hannover measurements, (c) Oxford measure-
ments, (d) O’Brian values, and (e) May74+GK81. The various data
sets yield systematically different ǫ(Fe) r esults. In par ticular , note
the offset between the Hannover and May74+GK81 values which
cover nearly the same range in EP space. In our analysis, we have
chosen to discount the M ay74+GK81 gf measurements.
![Fig. 12.— Solar-corrected [Na/Fe] abundance ratios vs. [Fe/H] metallicity for the 10 thick disk stars. All of the stars show enhanced Na/Fe and there is a mild trend with metallicity. The dashed line at [Zn/Fe] = 0 indicates the solar meteoritic Zn/Fe ratio.](/figures/fig-12-solar-corrected-na-fe-abundance-ratios-vs-fe-h-142jn06s.webp)
![Fig. 22.— Abundance patterns for the thick disk stars (large black stars) for 18 of the elements analyzed in this paper. For comparison, we plot the abundance patterns for halo stars (green points), thin disk stars (red points), other thick disk measurements (small black stars), and a small sample of bulge stars (light blue points). The different symbols refer to the various literature sources indicated in the text. There are three principal points to take from the figure: (1) for many of the elements, the thick disk abundances are distinct from the thin disk; (2) the thick disk abundances tend toward the halo star values at [Fe/H] ≈ −1; (3) the thick disk abundances are in good agreement with the bulge values for essentially every element.](/figures/fig-22-abundance-patterns-for-the-thick-disk-stars-large-1ga7vdmh.webp)