Académique Documents
Professionnel Documents
Culture Documents
( 2000. The American Astronomical Society. All rights reserved. Printed in U.S.A.
ABSTRACT
We present rst 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 10 m Keck I telescope. Our analysis conrms previous studies of O and Mg
in the thick disk stars, which reported enhancements in excess 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 exception of V and Co, the thick disk abundance
patterns match or tend toward the values observed for halo stars with [Fe/H] B [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. (1) All 10 stars exhibit enhanced a/Fe ratios with
O, Si, and Ca showing tentative trends of decreasing overabundances with increasing [Fe/H]. In con-
trast, the Mg and Ti enhancements are constant. (2) The light elements Na and Al are enhanced in these
stars. (3) With the exception of Ni, Cr, and possibly Cu, the iron-peak elements show signicant depar-
tures from the solar abundances. The stars are decient in Mn, but overabundant in V, Co, Sc, and Zn.
(4) The heavy elements Ba and Y are consistent with solar abundances, but Eu is signicantly 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 Z1 Gyr. We argue that this formation time-scale
would rule out most dissipational collapse scenarios for the formation of the thick disk. Models which
consider the heating of an initial thin diskeither through gradual heating mechanisms or a sudden
merger eventare 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 a-rich freeze out fueled by neutrino-driven winds. Meanwhile, the conicting trends for Mg,
Ti, Ca, Si, and O pose a difficult challenge to our current understanding of nucleosynthesis in Type Ia
and Type II SN. The Ba/Eu ratios favor r-process dominated enrichment for the heavy elements, consis-
tent with the ages (t [ 10 Gyr) expected for these stars.
agethe impact of the thick disk abundances on interpretations of the abundance pat-
Finally, we discuss
terns of the damped Lya 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.
Key words : Galaxy : abundances Galaxy : formation
nuclear reactions, nucleosynthesis, abundances stars : abundances
numbers of stars bluer than the main sequence turn-os of as the thin disk stars. Similarities would favor the
similar metallicity globular clusters (Gilmore & Wyse 1987 ; evolutionary history of the thick disk ; discontinuities
Carney et al. 1989 ; Gilmore, Wyse & Kuijken 1989). The would support a merger origin.
mean metallicity of the thick disk population, S[Fe/H]T, Large stellar samples with high-precision abundance
lies between [0.5 and [0.7 (Gilmore & Wyse 1985 ; analyses have appeared over the last several years, which
Carney et al. 1989 ; Gilmore et al. 1995 ; Layden 1995a ; may, in principle, answer this question. Edvardsson et al.
Layden 1995b). The spread in metallicities of thick disk (1993) studied a large sample of F and G dwarfs and found
stars ranges from near solar to [Fe/H] B [1, although that lower metallicity stars had, in general, enhanced
claims for much lower metallicities have been made (see [a/Fe]-values, but they did not compare the thick disk and
Norris et al. 1985 ; Morrison et al. 1990 ; Allen et al. 1991 ; thin disk stellar abundance patterns in detail. Gratton et al.
Ryan & Lambert 1995 ; Beers & Sommer-Larsen 1995 ; (1996) and Gratton et al. (2000) were the rst to directly
Twarog & Anthony-Twarog 1996 ; Martin & Morrison compare the abundance ratios of a sample (B15) of thick
1998 ; Chiba & Beers 2000). The asymmetric drift of the disk stars with halo and thin disk populations. Their mea-
thick disk (the amount by which it lags the circular orbit surements of Fe/O and Fe/Mg ratios indicate a stark dier-
motion at the solar Galactocentric distance) has been esti- ence in the Fe/O and Fe/Mg abundance of the thick and
mated to vary between 20 and 50 km s~1 (Carney et al. thin disk populations with the thick disk stars exhibiting
1989 ; Morrison et al. 1990 ; Majewski 1992 ; Beers & halo-like ratios. These authors argue for an early, rapid
Sommer-Larsen 1995 ; Ojha et al. 1996 ; Chiba & Beers formation of the Galactic thick disk, prior to the thin disk
2000), although Majewski (1992), Chen (1997), and Chiba & and perhaps due to an early accretion event. Fuhrmann
Beers (2000) have argued that the value varies with distance (1998) also compared Mg/Fe ratios for a sample of thick
from the Galactic plane. Values for the vertical velocity and thin disk stars taken from both Edvardsson et al. (1993)
dispersion, p(W ), are almost all near 40 km s~1 (Norris and his own smaller kinematic sample. Although the
1986 ; Carney et al. 1989 ; Beers & Sommer-Larsen 1995 ; Edvardsson et al. sample does not provide compelling evi-
Ojha et al. 1996 ; Chiba & Beers 2000), which implies a dence for a discontinuity, the Fuhrmann sample shows
vertical scale height of order 1 kpc or less. This may be strong evidence for a disjunction, which further supports
compared to the older stars of the thin disk, which obey a the assertion that the thick disk and thin disk have not
density distribution consistent with a vertical scale height of shared the same chemical enrichment history. Most recent-
about 300 pc. Although the thin disk is B10 times more ly, Chen et al. (2000) presented results from a sample of 90 F
massive than the thick disk, at distances of 1 to 2 kpc above and G dwarfs, which show no signicant scatter in a-
the plane, the thick disk population dominates. element ratios as a function of [Fe/H], contrary to the
The properties of the thick disk thus place it between results obtained by Fuhrmann (1998).
those of the halo and the thin disk. In turn, one questions We contend, however, that the sample selection
whether it is closely related to either of them in terms of the employed by Chen et al. (2000) was awed for a program
Galaxys chemical and dynamical evolution, or if it might aimed at the study of the thick disk. They chose to study
be the result of a merger event (see Gilmore et al. 1989 and only stars with eective temperatures between 5800 and
Majewski 1993 for excellent reviews of the various models). 6400 K, so that few of their stars have life expectancies as
Most evolutionary models predict that there should be con- great as the thick disks age. As an example, consider the
tinuity in the thick disk and disk dynamical and chemical disk, probably thick disk, globular cluster 47 Tuc. Using the
histories, and that thick disks should be found in other Alonso et al. (1996a) temperature scale (employed by Chen
galaxies. A merger scenario, conversely, would require some et al. 2000), the metallicity of [Fe/H] \ [0.70 (Carretta &
degree of discontinuity in the chemical and dynamical pa- Gratton 1997), and the photometry of Hesser et al. (1987),
rameters of the thick disk and the thin disk, or observations we nd that the temperatures of main sequence turn-o
that indicate not all disk galaxies have thick disks. It is stars in the cluster is near 5970 K, only slightly hotter than
interesting in this regard that very deep surface photometry the lower limit cut-o for the Chen et al. (2000) sample. And
of edge-on spirals reveals thick disks in some cases (e.g., for more metal-rich clusters or stars whose ages are as great
NGC 891 : van der Kruit & Searle 1981 ; Morrison et al. as 47 Tuc, the turn-o temperature is even cooler. Thus if
1997) but not in all cases (e.g., NGC 5907 : Morrison, the thick disk is composed almost exclusively of ancient
Boroson, & Harding 1994 ; NGC 4244 : Fry et al. 1999). stars, the Chen et al. (2000) sample cannot contain many
In this paper we study the problem using Galactic stars thick disk stars. Fuhrmanns (1998) claimed thick disk stars,
whose motions are consistent with thick disk membership. however, are cool enough to have long enough life expec-
Our goal here is to compare their abundance patterns, tancies to be considered part of the thick disk. We believe
[X/Fe] versus [Fe/H], with those of the other major Galac- that Chen et al. (2000) have studied, primarily, the detailed
tic stellar populations : the halo, thin disk and bulge. If the chemical evolution of the thin disk, which no doubt reaches
histories of the thick disk and the disk are closely related, to quite low metallicity levels itself (see in particular the
for example, so should be the derived chemical abundances study regarding the overlap in abundances of the thick disk
patterns versus metallicity. It is well established that very and the thin disk by Wyse & Gilmore 1995). We emphasize
metal-poor (halo) stars show enhanced levels of the light that comparative studies of the thick disk and the thin disk
a -rich nuclei elements oxygen, magnesium, silicon, must employ stars with life expectancies as great as the
calcium, and even titanium, but at a metallicity of [Fe/ thick disks age lest the trace but important population, the
H] B [1 the [a/Fe]-values begin to decline from ]0.4 dex thick disk, be overlooked. We do so here. In a future paper
or so to solar values at [Fe/H] \ 0 (see Wheeler, Sneden, & (Carney et al. 2000), we will study the relation between
Truran 1989 ; McWilliam 1997). A fundamental comparison kinematics and mean metallicity for long-lived dwarf stars
then is whether thick disk stars show similar [a/Fe] and in the mid-plane, nding further evidence for two distinct
other element abundance ratios at the same [Fe/H]-values populations.
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2515
We have initiated a program to measure the chemical most of the stars. Standard ThAr arc calibrations and
abundances of a large (N D 50) sample of thick disk stars at quartz ats were taken for reduction and calibration of the
very high resolution (R B 50,000) with high signal-to-noise spectra.
ratio (S/N [ 100), and a nearly continuous wavelength The data were extracted and wavelength calibrated with
coverage from j B 44009000 A . The stars were selected the MAKEE package developed by T. Barlow specically
from the surveys by Carney et al. (1994), Carney et al. (1996) for HIRES observations. The algorithm performs an
as exhibiting disk-like kinematics with large maximum dis- optimal extraction using the observed star to trace the
tance from the Galactic plane. The current sample is com- prole, and it solves for a wavelength calibration solution
prised of 10 stars, all brighter than V \ 10.5. We present a by cross-correlating the extracted ThAr spectra with an
detailed description of our stellar abundance analysis and extensive database compiled by T. Barlow. The pairs of
give rst results on a small but meaningful sample of stars. exposures were rebinned to the same wavelength scale and
In part, our goal is to build the analysis framework for coadded conserving ux. Finally, we continuum t the
future observations. This initial sample, however, suggests a summed spectra with a routine similar to IRAF, using the
number of exciting results which we will test through a Arcturus spectrum (Griffin 1968) as a guide in the bluest
larger survey. In 2 we describe the observations and orders where the ux rarely recovers to the continuum. An
present a summary of the stellar parameters of the sample. example of a typical spectral order is presented in Figure 1
The following section presents a thorough explanation of and we identify several representative absorption-line fea-
the techniques employed to measure the chemical abun- tures.
dances of the program stars. In general, we follow standard The sample of stars were kinematically selected from the
stellar analysis procedures. A solar analysis is discussed in study of Carney et al. (1994) to be members of the thick disk
4 and 5 gives an element-by-element account of the according to the following criteria. Our large initial list
results. Finally, 6 compares the thick disk results against excluded stars with any uncertain observational param-
other stellar populations and discusses the implications for
the formation history of the Galaxy, the damped Lya
systems, and nucleosynthesis in the early universe.
TABLE 1
JOURNAL OF OBSERVATIONS
NOTE.Units of right ascension are hours, minutes, and seconds, and units of declination are degrees, arcmin-
utes, and arcseconds (J2000.0).
a Hipparcos Identier, ESA (1997).
b j \ 4325[6760 A .
c j obs \ 6810[9200 A .
d jobs \ 6380[8750 A .
obs
2516 PROCHASKA ET AL. Vol. 120
eters, known subgiants, stars whose reddenings might and an extensive literature search. The equivalent width,
exceed 0.05 mag, and all stars known to be multiple-lined or W , for each line was then measured with the GETJOB
j
double-lined. To avoid stars whose lifetimes are shorter package developed by A. McWilliam for stellar spectro-
than the age of the thick disk, we avoided stars with the scopic analysis (McWilliam et al. 1995a). In the majority of
TO ag (meaning their colors place them near the main cases, we tted the absorption lines with single Gaussian
sequence turn-o for globular clusters of similar proles, which provide an excellent match to the majority of
metallicity). We further restricted the list to stars with observations. When necessary we tted regions with multi-
[1.1 [M/H] [0.4 to probe the thick disk metallicity ple Gaussians or calculated an integrated equivalent width
regime, and likewise eliminated stars whose orbits did not using Simpsons rule. The latter approach was particularly
carry them farther than 600 pc from the plane. To further important for strong Ca I and Mg I lines. The GETJOB
maximize the probability of observing thick disk stars program yields an error estimate for each W value based on
j
within this sample, we restricted the V3 velocities to lie the goodness-of-t and signal-to-noise of the spectra. The
between [20 and [100 km s~1. While these criteria help typical 1 p error for a single component t is B2 mA . With
minimize the contamination of the thick disk sample from the exception of a few special cases, those lines with errors
metal-rich halo stars and metal-poor thin disk stars, these exceeding 20% were eliminated from the subsequent abun-
stellar populations do overlap in both metallicity and kine- dance analysis. We also focused on unsaturated lines, spe-
matic properties and the possibility of contamination exists. cically lines with W /j \ (100 m)/(5000 ). Table 3 lists
j
In general, the overlap between the thick and thin disk the W values for the measured absorption lines. We have
j
populations is small (as determined from proper motion agged those absorption lines which we believe are blended
studies ; e.g., Carney et al. 1989) but the problem deserves or have incorrect gf-values.
further observational attention. Table 2 summarizes values
of the observed stars photometric temperatures, high- 3.2. Model Stellar Atmospheres
resolution and low signal-to-noise spectroscopic metallicity, Throughout the abundance analysis we adopt Kurucz
stellar gravities determined with Hipparcos measurements stellar atmospheres (Kurucz 1988) with convection on and
(ESA 1997), and stellar kinematics and galactic orbital pa- 72 layers with optical depth steps, *q \ 0.125, ending at
rameters from Carney et al. (1994). All of the stars are G q \ 100. Depending on the application, we either inter-
dwarfs found in the solar neighborhood with distances of polated between the stellar atmosphere grids kindly provid-
d \ 50100 pc. In those cases where there are Hipparcos ed by R. Kurucz or implemented the Kurucz package
pc
parallax measurements, we calculated the stellar gravity ATLAS9 to calculate specic models. The former approach
according to the equations presented in Appendix A. The has the advantage that the interpolation can be performed
uncertainties in the parallax and photometry imply an error with minimal human intervention and at minimal computa-
in log (g/g ) of B0.1 dex. In the following section, we will tional cost. In particular, we relied upon the grids to narrow
_
compute spectroscopic physical parameters for each star in on the spectroscopic physical parameters of each star
and compare with the photometric values presented here. ( 3.4). When applicable we constructed model atmospheres
with enhanced a-elements using the ]0.4 dex enhanced
3. ABUNDANCE ANALYSIS Rosseland opacities and the appropriate opacity distribu-
In this section we outline the prescription to measure tion functions.
elemental abundances for our sample of thick disk stars. In
short, we measured equivalent widths with the package 3.3. gf-V alues
GETJOB, implemented Kurucz model stellar atmospheres, Columns (4) and (5) of Table 3 list the adopted gf-values
culled log gfvalues from the literature, and used the stellar and their references for our sample of measured absorption
analysis package MOOG to constrain the spectroscopic lines. In general, we selected the most accurate and recent
physical parameters and determine the elemental abun- laboratory measurements available, avoiding solar gf-
dances. values where possible. Even with these accurate laboratory
values, however, the gf-values pose a major source of uncer-
3.1. Equivalent W idths tainty in the analysis particularly with respect to obtaining
We rst compiled a list of nearly 1000 reasonably measurements relative to the solar meteoritic abundances
unblended lines from the solar spectrum (Moore et al. 1966) which will serve as our abundance reference frame. We
TABLE 2
STELLAR PHYSICAL PARAMETERS
U3 V3 W3 Z R R d T
max apo per phot
Star V (km s~1) (km s~1) (km s~1) (kpc) (kpc) (kpc) (pc) (K) [M/H] log g
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 8.17 5.44 89 5069 [0.44 4.32
G92-19 . . . . . . . 10.31 ] 88 [70 [67 1.08 9.10 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
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2517
TABLE 3
EW MEASUREMENTS
j EP Sun G66-51 G84-37 G88-13 G92-19 G97-45 G114-19 G144-52 G181-46 G211-5 G247-32
Ion (A ) (eV) log gf Ref (mA ) (mA ) (mA ) (mA ) (mA ) (mA ) (mA ) (mA ) (mA ) (mA ) (mA )
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
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
NOTE.Table 3 is presented in its entirety in the electronic edition of the Astronomical Journal. A portion is shown here for guidance regarding its form
and content.
REFERENCES.1 : OBrian 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 : Blackwell 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 : Bard 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 : Biemont et al. (1991) ; 33 : McWilliam et al. (1995b) ; 36 : Hannaford & Lowe (1983) ; 37 : Moity (1983) ; 38 :
Buurman et al. (1986) ; 39 : Biemont et al. (1982) ; 42 : Kock & Richter (1968) ; 45 : Lawler & Dakin (1989) ; 48 : Lambert & Luck (1978) ; 54 : Fuhrmann et al.
(1995) ; 55 : Wicklie & Lawler (1997) ; 56 : Francois (1988) ; 57 : Biemont & 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).
FIG. 2.Plots of v(Fe)-values vs. excitation potential (EP) for six dierent sets of Fe I gf-values. (a) All samples, with the priority given in the text.
(b) Hannover measurements. (c) Oxford measurements. (d) OBrian values. (e) May74]GK81. The various data sets yield systematically dierent v(Fe)
results. In particular, note the oset 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 May74]GK81 gf measurements.
et al. (1991), Heise & Kock (1990), McWilliam et al. (1995b), mic equilibrium (LTE) holds throughout the stellar atmo-
Kroll & Kock (1987), Moity (1983), and Fuhr et al. (1988). sphere which is a good assumption for G dwarf stars. To
perform the analysis, we have used the stellar line analysis
3.4. Spectroscopic Atmospheric Parameters software package MOOG (version 1997) kindly provided
We now proceed to determine the spectroscopic atmo- by C. Sneden. In the mode ABFIND, the MOOG package
spheric parameterstemperature T , gravity log g , inputs a model stellar atmosphere and a list of data on each
eff spec
microturbulence m, and metallicity [Fe/H]for each star. absorption line (W , j, EP, log gf), and then it matches the
Following standard practice, we assume local thermodyna- observed W valuesj with a computed W value by adjusting
j j
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2519
the elemental abundance. For damping, we assumed the we derived a nal stellar atmosphere with ATLAS9. This
Unsold approximation with no enhancement. As a check, nal model atmosphere was then adopted in the abundance
we experimented with other assumptions for the damping, analysis of all of the elements. As we shall see, nearly every
in particular the Blackwell correction and a factor of two star observed in this sample has enhanced a-elements (Mg,
enhancement to the Unsold approximation. To our sur- Si, O, etc.). For the stars with [Fe/H] B [0.5, Mg is a
prise, under neither of these latter assumptions were we able principal source of electrons and therefore an a-
to derive a model atmosphere which was physically reason- enhancement can signicantly modify the model atmo-
able. Specically, the enhancements suggest a lower micro- sphere. In particular, ignoring the higher electron density
turbulence which in turn require a lower temperature which associated with an enhanced magnesium abundance leads
then implies a lower microturbulence. In any case, we have to a systematic overestimate of the stellar gravity. There-
adopted the same damping approximation (no fore, we also derived an a-enhanced model atmosphere with
enhancement) for our solar analysis ( 4) and hope to have ATLAS9 (assuming [a/Fe] \ ]0.4 dex and using the
minimized the eects on our nal results. appropriate opacities) and performed a complete abun-
To measure the spectroscopic atmospheric parameters, dance analysis with this enhanced atmosphere including a
one modies the model atmosphere to satisfy three con- reanalysis of the atmospheric parameters. We have found,
straints : (1) minimize the slope of v(Fe I) versus W /j ; (2) however, that these atmospheres imply small dierences
j
minimize the slope of v(Fe I) versus EP ; and (3) require that from the results of the standard atmospheres.
the median v(Fe I) value equal the median v(Fe I) value. The Figure 3 presents (a) v(Fe I) versus W /j and (b) v(Fe I)
j in the sample,
rst constraint determines m because the adopted micro- versus EP plots for every thick disk star
turbulence value has a signicant eect on the abundances assuming the standard model atmospheres. The physical
derived from large equivalent width lines, i.e., lines which parameters of the nal model atmospheres (with and
suer from saturation. Therefore, requiring that the v(Fe I)- without a-enhancement) are listed in Table 4. As expected,
values exhibit no trend with W sets the microturbulence of the a-enhanced models tend to have lower log gvalues by
j the slope of v(Fe I)-values
the model atmosphere. Similarly, B0.050.1 dex. We will show that the typical a-
versus EP is sensitive to the eective temperature because enhancement is 0.20.3 dex so the most accurate spectro-
the predicted population of various EP levels is a function scopic gravity is more likely the average of the two values.
of the temperature of the stellar atmosphere. Finally, log g is Furthermore, the a-enhanced models require slightly higher
constrained by requiring that the Fe abundance derived T values to compensate the larger opacity implied by the
eff
from the Fe II lineswhich are sensitive to log g match the increase in electron density. With the notable exception of
Fe abundance from Fe I. This parameter is perhaps the G84-37, our spectroscopic T , log g, and [Fe/H]-values are
eff photometric values based on
most uncertain as it is dependent on systematic errors in systematically higher than the
both the Fe I and Fe II gf-values. In practice, the constraints the Carney et al. (1994) observations listed in Table 2.
are mildly degenerate in the atmospheric parameters and Excluding G84-37, the oset between the spectroscopic
one iteratively solves for the model atmosphere. Our temperatures and the Carney et al. (1994) values is ST
spec
approach was to guess values of log g and [Fe/H] and then [T T \ ]87 ^ 15 K for the standard atmospheres.
Carn94
use a s2 minimization routine AUTOAB4 developed by A. On the other hand, comparing the spectroscopic values
McWilliam to nd the m and T values which minimize the with the Alonso et al. (1995) color-temperature relations
slope of v(Fe I) versus log W /jeffand EP. This notion placed based on the infrared ux method (IRFM), we nd excellent
typical error estimates for T j and m at B50 K and 0.05 km agreement : ST [ T T \ [1 ^ 12 K for the four
eff
s~1 respectively. We then adjusted the log g and [Fe/H]- stars with uvbyspec
Alonso96The agreement is possibly the
photometry.
values and reran AUTOAB4 until Fe I and Fe II were result of small number statistics, however, particularly given
brought into agreement with one another. To minimize our the excellent overall agreement between the two photo-
eort, we calculated the stellar atmospheric models by metric techniques for low mass main sequence stars (Alonso
interpolating the grids provided by R. Kurucz.7 Once we et al. 1996b). As we increase the sample of thick disk stars, it
determined reasonable values for T , log g, m, and [Fe/H], will be important to further compare the various tem-
eff perature scales. Finally, as we will ultimately be interested
in a comparison of the Edvardsson et al. (1993) and Chen et
7 Available at http ://cfaku5.harvard.edu. al. (2000) results with our analysis, it is important to
TABLE 4
ATMOSPHERIC SPECTROSCOPIC PARAMETERS
T m Ta ma
spec spec
Star (K) [M/H] log g (km s~1) (K) [M/H]a log ga (km s~1)
FIG. 3.Plots of v(Fe) vs. the reduced equivalent width (log W /j) and excitation potential (EP) derived from the Fe I lines for the 10 program stars. The
j
v(Fe I)-values correspond to standard model atmospheres with physical parameters listed in Table 4. These parameters were determined by minimizing the
trends of v(Fe) vs. log W /j and EP simultaneously and by requiring that the median v(Fe) value from Fe I match that for Fe II.
j
examine their temperature scales. As an indirect test, we gravity oset is more difficult to explain. Taking the average
compared the 33 overlapping stars from the Alonso et al. spectroscopic gravity from the standard and a-enhanced
(1996a) and Edvardsson et al. (1993) samples which are atmospheres (and again ignoring G84-37 for now), the
nearby (minimally aected by dust) and in the appropriate average oset is Slog g [ log g T \ [0.08 ^ 0.0 1
Hippa systematic
spec error in the Fe II
temperature range (T \ 5900 K). We nd ST dex. We might suggest that
T \ [95 ^ 6 eff
Alonso96
[T K, so we will assume that our stars gf-values has biased the spectroscopic gravity, but our log g
Edvard
are B100 K cooler than the Edvardsson et al. (1993) tem- measurement for the Sun is in excellent agreement with the
perature scale. Similarly Chen et al. (2000), whose tem- known value. A fraction of the oset (0.030.05 dex) can be
peratures are based on the IRFM scale, report a explained by the dierence in assumed eective tem-
systematically lower temperature (B70 K) than Edvardsson perature, but it can not account for the entire discrepancy.
et al. (1993). Therefore, we will assume no temperature oset Nonetheless, the eect of even a 0.1 dex error in log g has a
between our analysis and that of Chen et al. (2000). minimal impact on the abundances we derive, particularly
While a dierence between the spectroscopic and the since we rely primarily on the Fe I lines to determine
color-temperature scales may not be surprising, the stellar [Fe/H]. Finally, note that the metallicity we compute from
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2521
FIG. 3.Continued
the Fe lines is systematically B0.1 dex higher than the formed a standard abundance error analysis. We calculated
[M/H]-values derived by Carney et al. (1994) from low S/N the elemental abundances for 10 atmospheric models for
spectra. In this case, the dierence is largely explained by two stars (G114-19 and G84-37) using the Kurucz atmo-
the oset in temperature and gravity as both imply a higher sphere grids : (1) the best-t model ; (2, 3) T @ \ T ^ 50 K ;
metallicity. While a 0.1 dex systematic error in the spectro- (4, 5) m@ \ m ^ 0.05 km s~1 ; (6, 7) log g@ \ log g ^ 0.05 ;
scopic metallicity measurements will not signicantly aect (8, 9) [M/H]@ \ [M/H] ^ 0.05 dex ; and (10) a ]0.4 a-
our conclusions on the abundance trends of the thick disk enhanced atmosphere. The two stars were chosen to have
stars, a systematic error in the Carney et al. (1994) [M/H] signicantly dierent stellar atmospheres and the range of
measurements could have important implications for the parameters roughly corresponds to our estimated 1 p sta-
metallicity distribution function of the thick disk. Therefore, tistical uncertainty. Tables 5 and 6 summarize the results of
we will carefully reassess this discrepancy once we have a the error analysis for the two stars. For those elements with
larger, statistically signicant sample of accurate spectro- N [ 5 absorption lines, we expect the uncertainties in the
scopic measurements. abundances to be dominated by errors in the atmospheric
parameters. For the remaining elements, the errors in W
3.5. Error Analysis (i.e., Poissonian noise, blends, continuum error) are signi-j
To assess the systematic eects of the model atmospheric cant. In 5 we remark on those elements for which errors in
parameters on the elemental abundance ratios, we have per- the W measurements are a particular problem.
j
2522 PROCHASKA ET AL. Vol. 120
TABLE 5
ERROR ANALYSIS FOR G84-37
Fe I/H . . . . . . . . 94 ]0.050 [0.042 [0.002 ]0.006 [0.006 ]0.011 ]0.000 ]0.000 ]0.025
Fe II/H . . . . . . . 17 [0.003 ]0.001 ]0.018 [0.017 [0.011 ]0.013 ]0.006 [0.005 [0.009
O I/Fe . . . . . . . . 3 [0.102 ]0.090 ]0.017 [0.023 ]0.004 [0.009 [0.003 ]0.002 [0.080
Na I/Fe . . . . . . 3 [0.023 ]0.015 [0.002 [0.001 ]0.005 [0.010 ]0.000 ]0.000 [0.011
Mg I/Fe . . . . . . 5 [0.037 ]0.030 ]0.002 [0.006 ]0.006 [0.011 ]0.001 ]0.001 [0.017
Si I/Fe . . . . . . . . 12 [0.032 ]0.025 ]0.004 [0.007 ]0.004 [0.008 ]0.000 ]0.000 [0.018
S I/Fe . . . . . . . . 1 [0.089 ]0.077 ]0.017 [0.024 ]0.005 [0.011 [0.002 ]0.001 [0.067
Ca I/Fe . . . . . . 17 [0.017 ]0.008 [0.003 ]0.000 ]0.000 [0.005 ]0.000 ]0.001 [0.014
Sc II/Fe . . . . . . 4 [0.040 ]0.031 ]0.020 [0.025 ]0.003 [0.009 ]0.005 [0.006 ]0.001
Ti I/Fe . . . . . . . 17 ]0.003 [0.012 ]0.002 [0.003 ]0.002 [0.005 ]0.000 ]0.001 ]0.000
Ti II/Fe . . . . . . 16 [0.038 ]0.030 ]0.021 [0.024 ]0.000 [0.008 ]0.006 [0.005 [0.009
Cr I/Fe . . . . . . . 12 ]0.003 [0.004 ]0.001 [0.003 ]0.003 [0.002 ]0.001 ]0.000 ]0.004
Cr II/Fe . . . . . . 4 [0.060 ]0.049 ]0.020 [0.025 ]0.001 [0.007 ]0.002 [0.003 [0.037
Mn I/Fe . . . . . . 10 [0.013 ]0.002 ]0.002 [0.006 ]0.005 [0.011 ]0.000 ]0.000 [0.005
Co I/Fe . . . . . . 1 ]0.000 [0.009 ]0.003 [0.007 ]0.005 [0.011 ]0.001 ]0.001 ]0.009
Ni I/Fe . . . . . . . 21 [0.021 ]0.012 ]0.002 [0.006 ]0.003 [0.009 ]0.000 [0.001 [0.013
Cu I/Fe . . . . . . 2 [0.010 ]0.002 ]0.005 [0.006 ]0.006 [0.010 ]0.000 ]0.000 ]0.000
Zn I/Fe . . . . . . 2 [0.037 ]0.029 ]0.010 [0.014 [0.001 [0.004 ]0.002 [0.002 [0.022
Y II/Fe . . . . . . . 5 [0.035 ]0.027 ]0.020 [0.026 [0.002 [0.002 ]0.006 [0.006 [0.003
Ba II/Fe . . . . . . 3 [0.023 ]0.017 ]0.013 [0.013 [0.023 ]0.024 ]0.006 ]0.001 [0.002
3.6. Hyperne Splitting dex. For our abundance analysis, we have included hfs cor-
Isotopes with an odd number of protons and/or neutrons rections for Mn, Ba, Sc, Co, and Cu.8 With the exception of
experience hyperne interactions between the nucleus and Ba where we have implemented the results from McWilliam
electrons. These interactions split the lines into multiple (1998), we adopt the wavelengths of the hfs transitions from
components with typical separations of 110 mA . For Kuruczs hyperne tables (R. L. Kurucz 1999, private
strong lines with large equivalent width the eect is to communication) and calculated the relative gf strengths
desaturate the absorption line, a phenomenon which must
be taken into account in order to accurately measure the
elemental abundance. In the case of several Cu I lines, for 8 The eects of hfs are negligible for the very weak Eu II 6645 line and
example, hyperne splitting leads to a correction of over 0.5 insignicant for the Y II lines.
TABLE 6
ERROR ANALYSIS FOR G114-19
Fe I/H . . . . . . . . 79 ]0.031 [0.042 [0.001 ]0.004 [0.007 ]0.006 ]0.008 [0.007 ]0.006
Fe II/H . . . . . . . 20 [0.021 ]0.019 ]0.026 [0.028 [0.005 ]0.005 ]0.011 [0.012 ]0.031
O I/Fe . . . . . . . . 3 [0.079 ]0.084 ]0.025 [0.031 ]0.006 [0.005 [0.014 ]0.011 [0.015
Na I/Fe . . . . . . 3 ]0.002 ]0.007 [0.007 ]0.005 ]0.006 [0.005 [0.006 ]0.005 ]0.011
Mg I/Fe . . . . . . 6 [0.016 ]0.025 [0.005 ]0.003 ]0.007 [0.005 [0.005 ]0.004 ]0.023
Al I/Fe . . . . . . . 1 [0.001 ]0.011 [0.003 ]0.001 ]0.006 [0.005 [0.007 ]0.007 ]0.001
Si I/Fe . . . . . . . . 14 [0.031 ]0.041 ]0.005 [0.008 ]0.005 [0.004 ]0.001 [0.002 ]0.035
S I/Fe . . . . . . . . 1 [0.068 ]0.074 ]0.025 [0.030 ]0.007 [0.005 [0.012 ]0.010 [0.012
Ca I/Fe . . . . . . 16 ]0.014 [0.003 [0.017 ]0.020 ]0.000 ]0.001 ]0.002 ]0.001 ]0.031
Sc II/Fe . . . . . . 7 [0.030 ]0.039 ]0.023 [0.026 ]0.006 [0.002 ]0.007 [0.007 ]0.049
Ti I/Fe . . . . . . . 41 ]0.030 [0.020 [0.002 ]0.002 ]0.006 [0.002 [0.006 ]0.008 ]0.003
Ti II/Fe . . . . . . 12 [0.028 ]0.038 ]0.021 [0.024 [0.003 ]0.006 ]0.008 [0.007 ]0.031
V I/Fe . . . . . . . . 15 ]0.037 [0.028 [0.002 ]0.001 ]0.006 [0.004 [0.006 ]0.007 ]0.020
Cr I/Fe . . . . . . . 13 ]0.012 [0.009 [0.015 ]0.008 [0.006 ]0.000 [0.003 [0.002 ]0.001
Cr II/Fe . . . . . . 5 [0.044 ]0.053 ]0.023 [0.027 ]0.003 [0.002 ]0.000 [0.001 ]0.002
Mn I/Fe . . . . . . 10 ]0.024 [0.027 [0.001 [0.001 ]0.005 [0.004 [0.005 ]0.005 ]0.013
Co I/Fe . . . . . . 9 ]0.003 ]0.008 ]0.007 [0.010 ]0.006 [0.005 [0.002 ]0.002 ]0.030
Ni I/Fe . . . . . . . 28 [0.008 ]0.014 [0.008 ]0.000 [0.003 ]0.001 ]0.004 [0.005 ]0.013
Cu I/Fe . . . . . . 1 [0.015 ]0.028 ]0.008 [0.010 ]0.005 [0.003 ]0.000 ]0.001 ]0.021
Zn I/Fe . . . . . . 2 [0.031 ]0.041 ]0.007 [0.010 [0.005 ]0.006 ]0.005 [0.006 ]0.002
Y II/Fe . . . . . . . 1 [0.023 ]0.030 ]0.016 [0.023 [0.010 ]0.007 ]0.007 [0.011 ]0.000
Ba II/Fe . . . . . . 2 [0.016 ]0.024 ]0.012 [0.016 [0.016 ]0.015 ]0.011 [0.013 ]0.003
Eu II/Fe . . . . . . 1 [0.029 ]0.039 ]0.023 [0.026 ]0.006 [0.004 ]0.007 [0.008 ]0.055
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2523
4. SOLAR ANALYSIS
In order to facilitate abundance comparisons between
our thick disk sample and other stellar populations or
galactic systems (e.g., the damped Lya systems), it is crucial
to compare our results with a spectroscopic solar analysis.
In this fashion, we can report our abundances relative to the
solar meteoritic abundances (Grevesse et al. 1996) by com-
paring the solar analysis with the meteoritic values.9 This
exercise also accentuates systematic errors associated with
the log gfvalues or blending of individual ions. More ambi-
tiously, by making a line by line comparison we might
eliminate errors in the model atmospheres, damping, and
the stellar analysis package, particularly given that these
thick disk stars have similar spectral types to the Sun.
Therefore, we performed an elemental abundance analysis
of the Sun applying the exact same techniques utilized for
the thick disk stars. The equivalent widths were measured
with the GETJOB package, we adopted Kurucz model
atmospheres, and constrained the atmospheric parameters
with AUTOAB4. The only dierences lie in the solar spec-
trum itself ; we have analyzed the Kurucz solar spectrum
(Kurucz et al. 1984) obtained with resolving power 522,000
and a signal-to-noise ratio in excess of 2000. Column (6) of
FIG. 4.v(Fe)-values vs. log W /j and EP derived from 112 Fe I mea-
Table 3 lists the W measurements for the Sun which were j
j sured in the Kurucz solar spectrum. By minimizing the slope of v(Fe) with
included in the abundance analysis. We estimate that the log W /j and EP, we derived the best atmospheric parameters for the
typical error of each measurement is 12 mA , the dominant j
Sun (T , log g, [M/H], m). To our surprise, these parameters agree very
eff
sources of error being line blends and poor continuum well with the known parameters.
determination. Figure 4 presents the v(Fe) versus W /j and
j spec-
EP plots for 112 Fe I lines measured from the solar
TABLE 7
trum. Somewhat to our surprise, the physical parameters
that we derive are in excellent agreement with the known SOLAR ABUNDANCES RELATIVE TO METEORITIC
values : T \ 5750 ^ 50 K, log g \ 4.44 ^ 0.03, m \ 1.00
^ 0.03 kmeffs ~1, and [M/H] \ 0.0 dex. Perhaps most aston-
Ion v(X)a N [X/H] [X/H] p
d n
ishing, the Fe abundance matches the meteoritic value to C I ........ 8.55 4 0.23 0.19 0.18
within 0.015 dex. While the nearly exact agreement is prob- O I........ 8.87 3 0.12 0.12 0.02
ably fortuitous, our analysis indicates no signicant dis- Na I . . . . . . 6.32 3 0.00 [0.01 0.02
agreement between the photometric and meteoritic solar Fe Mg I . . . . . . 7.58 3 0.03 0.04 0.02
abundance. Al I . . . . . . . 6.49 2 [0.09 [0.09 0.05
Si I . . . . . . . . 7.56 16 0.00 0.01 0.04
With the atmospheric parameters determined, we con-
S I ........ 7.20 2 0.35 0.35 0.00
structed a nal model atmosphere with ATLAS9 and mea- Ca I . . . . . . 6.35 20 [0.07 [0.05 0.10
sured the elemental abundances of the remaining Sc II . . . . . . 3.10 9 0.17 0.17 0.08
absorption lines. Table 7 lists the ion, the number of absorp- Ti I . . . . . . . 4.94 47 [0.05 [0.04 0.07
tion lines analyzed, the median and mean abundance rela- Ti II . . . . . . 4.94 19 0.11 0.14 0.13
tive to the meteoritic value, and the standard deviation of V I ........ 4.02 17 [0.12 [0.12 0.06
these measurements. While the majority of ions are consis- Cr I . . . . . . . 5.67 14 0.01 0.02 0.07
tent with the meteoritic values, there are notable excep- Cr II . . . . . . 5.67 6 0.09 0.14 0.12
tions : Ti II (]0.11 dex), S I (]0.35 dex), C I (]0.23 dex), Mn Mn I . . . . . . 5.53 9 [0.21 [0.20 0.07
I ([0.21 dex), Cu I ([0.13 dex), Cr II (]0.09 dex), and Sc II Fe I . . . . . . . 7.50 112 [0.01 [0.02 0.09
Fe II . . . . . . 7.50 33 0.00 0.01 0.10
(]0.19 dex). In each of these cases, either the absolute scale
Co I . . . . . . 4.91 10 0.03 0.03 0.11
of the gf-values is poorly determined or the abundances are Ni I . . . . . . . 6.25 33 0.00 [0.01 0.10
very sensitive to the model atmospheres. For example, the S Cu I . . . . . . 4.29 3 [0.01 [0.05 0.08
Zn I . . . . . . 4.67 2 0.00 0.00 0.06
Y II . . . . . . . 2.23 3 0.02 0.01 0.02
9 Implied in this exercise is the assumption that the solar spectroscopic Ba II . . . . . . 2.22 2 0.11 0.11 0.12
abundances must equal the meteoritic. While this is supported by the Eu II . . . . . . 0.54 1 0.10 0.10
excellent agreement between the two for many elements, there are notable
exceptions and we warn the reader that this assumption need not hold. a Meteoritic solar abundances from Grevesse et al. (1996).
2524 PROCHASKA ET AL. Vol. 120
TABLE 8
ABUNDANCES FOR G66-51
O I........ 3 0.630 0.634 0.020 0.588 0.594 0.018 3 0.490 0.486 0.019
Na I . . . . . . 4 0.088 0.119 0.053 0.091 0.118 0.061 4 0.091 0.099 0.053
Mg I . . . . . . 6 0.368 0.397 0.074 0.377 0.395 0.073 3 0.350 0.364 0.069
Al I . . . . . . . 1 0.245 0.256 ... 0.234 0.243 ... 1 0.290 0.286 ...
Si I . . . . . . . . 13 0.268 0.271 0.053 0.286 0.278 0.056 13 0.224 0.229 0.048
Ca I . . . . . . 19 0.214 0.226 0.090 0.213 0.242 0.081 19 0.254 0.249 0.069
Sc II . . . . . . 6 0.292 0.312 0.065 0.326 0.339 0.073 6 0.093 0.095 0.052
Ti I . . . . . . . 45 0.299 0.311 0.041 0.290 0.297 0.045 37 0.346 0.340 0.069
Ti II . . . . . . 11 0.392 0.410 0.065 0.412 0.419 0.059 11 0.204 0.198 0.083
V I ........ 9 0.074 0.098 0.050 0.081 0.099 0.048 8 0.204 0.186 0.038
Cr I . . . . . . . 16 0.062 0.061 0.081 0.047 0.047 0.086 14 0.000 0.007 0.081
Cr II . . . . . . 2 0.106 0.117 0.006 0.091 0.100 0.001 2 0.063 0.059 0.042
Mn I . . . . . . 10 [0.428 [0.421 0.051 [0.434 [0.424 0.056 8 [0.257 [0.247 0.074
Fe I . . . . . . . 102 [1.018 [1.029 0.065 [1.008 [1.016 0.065 90 [1.009 [1.005 0.076
Fe II . . . . . . 14 [1.031 [1.037 0.073 [1.021 [1.028 0.071 12 [1.046 [1.025 0.097
Co I . . . . . . 5 0.139 0.168 0.169 0.159 0.179 0.173 3 0.086 0.091 0.020
Ni I . . . . . . . 26 0.032 0.040 0.051 0.039 0.044 0.055 24 [0.006 0.009 0.094
Cu I . . . . . . 2 [0.036 [0.025 0.044 [0.032 [0.024 0.052 2 [0.039 [0.044 0.047
Zn I . . . . . . 2 0.203 0.214 0.035 0.198 0.206 0.034 2 0.191 0.187 0.021
Y II . . . . . . . 2 0.049 0.059 0.056 0.055 0.064 0.067 0
Ba II . . . . . . 3 0.096 0.122 0.097 0.092 0.127 0.111 2 [0.058 [0.063 0.073
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2525
TABLE 9
ABUNDANCES FOR G84-37
O I........ 3 0.801 0.805 0.043 0.721 0.727 0.044 3 0.656 0.646 0.066
Na I . . . . . . 3 0.229 0.188 0.086 0.218 0.177 0.088 3 0.204 0.162 0.070
Mg I . . . . . . 5 0.368 0.382 0.075 0.351 0.364 0.076 3 0.310 0.315 0.077
Si I . . . . . . . . 12 0.288 0.310 0.076 0.270 0.292 0.076 9 0.284 0.274 0.060
S I ........ 1 0.719 0.712 ... 0.652 0.648 ... 1 0.345 0.327 ...
Ca I . . . . . . 17 0.279 0.274 0.042 0.265 0.266 0.041 17 0.321 0.293 0.079
Sc II . . . . . . 4 0.197 0.172 0.077 0.196 0.174 0.081 4 [0.001 [0.013 0.024
Ti I . . . . . . . 17 0.220 0.217 0.031 0.220 0.223 0.030 15 0.228 0.217 0.081
Ti II . . . . . . 16 0.387 0.368 0.089 0.378 0.365 0.087 15 0.200 0.189 0.095
Cr I . . . . . . . 12 [0.055 [0.039 0.050 [0.051 [0.039 0.049 11 [0.092 [0.110 0.085
Cr II . . . . . . 4 0.122 0.148 0.096 0.085 0.114 0.093 4 [0.018 [0.012 0.077
Mn I . . . . . . 10 [0.476 [0.474 0.050 [0.481 [0.477 0.050 6 [0.317 [0.327 0.105
Fe I . . . . . . . 94 [1.213 [1.206 0.056 [1.188 [1.184 0.057 82 [1.188 [1.171 0.115
Fe II . . . . . . 17 [1.196 [1.213 0.071 [1.205 [1.216 0.071 15 [1.260 [1.224 0.077
Co I . . . . . . 1 0.256 0.249 ... 0.265 0.261 ... 0
Ni I . . . . . . . 21 0.037 0.034 0.053 0.024 0.028 0.056 19 0.011 0.007 0.085
Cu I . . . . . . 2 [0.329 [0.336 0.194 [0.329 [0.333 0.179 2 [0.279 [0.296 0.093
Zn I . . . . . . 2 0.058 0.051 0.033 0.036 0.032 0.034 2 0.030 0.013 0.024
Y II . . . . . . . 5 0.065 0.053 0.050 0.062 0.054 0.051 1 [0.056 [0.073 ...
Ba II . . . . . . 3 0.207 0.176 0.083 0.205 0.180 0.079 2 0.015 [0.002 0.042
particular attention to the possibility that we have overesti- from the Fe I lines. In this section and all further analysis,
mated T for the majority of stars ( 3.4). In the following we take [Fe/H] from the median v(Fe) value of the Fe I
eff will compare the observed trends of these thick
section, we lines. While signicant uncertainties exist for the Fe I and
disk stars with the halo (McWilliam et al. 1995b), thin disk Fe II gf-values, the excellent agreement between our solar
(Edvardsson et al. 1993 ; Chen et al. 2000), and bulge spectroscopic Fe abundance and the meteoritic abundance
(McWilliam & Rich 1994) stellar populations. raises our condence in the absolute value of our [Fe/H]
measurements. Furthermore, with the exception of one star,
5.1. Iron the [Fe/H]-values are essentially independent of a-
We computed the [Fe/H]-values for the program stars enhancement. This one exception, G88-13, has the highest
from the Fe I and Fe II measurements under the constraint metallicity of all of our program stars but is otherwise
that the adopted stellar gravity gives a median v(Fe) value unpeculiar. While the dierence in [Fe/H] is signicant for
from the Fe II lines within 0.03 dex of the median v(Fe) value this star, the majority of elemental abundances relative to
TABLE 10
ABUNDANCES FOR G88-13
O I........ 3 0.250 0.250 0.032 0.217 0.227 0.042 3 0.094 0.106 0.052
Na I . . . . . . 3 0.096 0.093 0.027 0.083 0.088 0.060 3 0.097 0.078 0.034
Mg I . . . . . . 6 0.329 0.309 0.048 0.315 0.296 0.045 2 0.252 0.245 0.043
Al I . . . . . . . 2 0.228 0.214 0.033 0.194 0.181 0.043 2 0.290 0.283 0.018
Si I . . . . . . . . 14 0.147 0.123 0.045 0.164 0.143 0.052 12 0.093 0.096 0.051
S I ........ 1 0.216 0.202 ... 0.187 0.173 ... 1 [0.161 [0.168 ...
Ca I . . . . . . 12 0.111 0.089 0.115 0.107 0.098 0.103 12 0.145 0.125 0.083
Sc II . . . . . . 6 0.339 0.345 0.080 0.377 0.384 0.085 6 0.164 0.139 0.063
Ti I . . . . . . . 42 0.254 0.237 0.063 0.215 0.199 0.066 33 0.270 0.271 0.069
Ti II . . . . . . 14 0.303 0.292 0.096 0.313 0.315 0.097 12 0.155 0.153 0.061
V I ........ 15 0.154 0.163 0.134 0.148 0.140 0.133 12 0.288 0.285 0.084
Cr I . . . . . . . 9 0.056 0.061 0.107 0.023 0.040 0.118 9 0.040 0.026 0.070
Cr II . . . . . . 4 0.144 0.115 0.099 0.142 0.109 0.098 4 [0.013 [0.022 0.088
Mn I . . . . . . 6 [0.282 [0.280 0.087 [0.305 [0.303 0.094 6 [0.083 [0.079 0.054
Fe I . . . . . . . 74 [0.391 [0.377 0.080 [0.330 [0.316 0.088 64 [0.364 [0.357 0.089
Fe II . . . . . . 16 [0.396 [0.383 0.088 [0.311 [0.299 0.089 16 [0.362 [0.373 0.100
Co I . . . . . . 6 0.068 0.059 0.052 0.068 0.055 0.060 5 0.096 0.087 0.032
Ni I . . . . . . . 31 0.048 0.063 0.102 0.067 0.079 0.100 28 0.024 0.030 0.078
Cu I . . . . . . 1 0.071 0.057 ... 0.070 0.056 ... 1 0.047 0.040 ...
Zn I . . . . . . 2 0.139 0.125 0.090 0.155 0.141 0.088 2 0.108 0.101 0.033
Y II . . . . . . . 3 [0.007 0.056 0.137 [0.002 0.061 0.140 0
Ba II . . . . . . 2 0.106 0.093 0.063 0.150 0.136 0.105 2 [0.030 [0.037 0.061
2526 PROCHASKA ET AL. Vol. 120
TABLE 11
ABUNDANCES FOR G92-19
O I........ 3 0.386 0.390 0.008 0.413 0.410 0.014 3 0.249 0.242 0.031
Na I . . . . . . 4 0.080 0.106 0.074 0.089 0.111 0.085 4 0.071 0.085 0.075
Mg I . . . . . . 4 0.306 0.304 0.028 0.315 0.309 0.032 2 0.235 0.239 0.010
Al I . . . . . . . 2 0.204 0.209 0.014 0.195 0.200 0.016 2 0.271 0.274 0.036
Si I . . . . . . . . 15 0.177 0.173 0.056 0.208 0.198 0.060 12 0.129 0.140 0.054
S I ........ 1 0.365 0.370 ... 0.381 0.386 ... 1 [0.008 [0.004 ...
Ca I . . . . . . 19 0.145 0.150 0.074 0.143 0.159 0.074 19 0.173 0.172 0.075
Sc II . . . . . . 7 0.276 0.297 0.063 0.318 0.353 0.066 7 0.087 0.097 0.029
Ti I . . . . . . . 44 0.197 0.205 0.036 0.177 0.188 0.037 38 0.221 0.231 0.074
Ti II . . . . . . 14 0.315 0.310 0.098 0.356 0.353 0.097 13 0.145 0.133 0.065
V I ........ 13 0.041 0.066 0.057 0.028 0.056 0.056 12 0.118 0.132 0.059
Cr I . . . . . . . 16 0.046 0.057 0.076 0.040 0.042 0.078 15 [0.003 0.009 0.063
Cr II . . . . . . 5 0.092 0.102 0.101 0.118 0.129 0.100 5 [0.041 [0.037 0.039
Mn I . . . . . . 10 [0.333 [0.333 0.055 [0.344 [0.341 0.056 8 [0.188 [0.154 0.076
Fe I . . . . . . . 95 [0.634 [0.639 0.064 [0.641 [0.646 0.065 86 [0.611 [0.615 0.081
Fe II . . . . . . 20 [0.662 [0.668 0.091 [0.629 [0.636 0.092 18 [0.689 [0.687 0.079
Co I . . . . . . 6 0.100 0.090 0.097 0.107 0.095 0.099 5 0.097 0.069 0.058
Ni I . . . . . . . 31 0.039 0.041 0.062 0.046 0.050 0.062 29 [0.008 0.007 0.081
Cu I . . . . . . 3 [0.023 [0.039 0.046 [0.011 [0.032 0.048 3 [0.032 [0.011 0.035
Zn I . . . . . . 2 0.044 0.049 0.044 0.071 0.076 0.042 2 0.018 0.022 0.013
Y II . . . . . . . 4 0.045 0.048 0.061 0.090 0.093 0.065 1 0.039 0.043 ...
Ba II . . . . . . 3 0.123 0.136 0.079 0.180 0.196 0.094 2 [0.042 [0.039 0.077
Eu II . . . . . . 1 0.525 0.530 ... 0.588 0.593 ... 1 0.398 0.402 ...
Fe are insensitive to the a-enhancement and we will gener- 5.2.1. Silicon, Calcium, T itanium
ally not include this approach in our discussion. For the abundances of Si, Ca, and Ti, we have measured
over 15 absorption lines and have reasonably accurate
5.2. Alpha ElementsO, Mg, Si, S, Ca, Ti laboratory gf-values. In the case of silicon, we have adopted
Our observations include measurements on a number of the gf-values from Garz (1973) adjusted by ]0.1 dex as
a-elements. In this subsection, we describe the results. We recommended by Becker et al. (1980) as well as the solar
have grouped the elements into two subsets, one with rela- gf-values from Fry & Carney (1997) and McWilliam & Rich
tively robust results and the other with poorly constrained (1994) which give v(Si)-values in good agreement with the
measurements. adjusted Garz (1973) abundances. For calcium, we rely
TABLE 12
ABUNDANCES FOR G97-45
O I........ 3 0.545 0.548 0.019 0.566 0.563 0.010 3 0.416 0.400 0.006
Na I . . . . . . 4 0.131 0.153 0.078 0.181 0.190 0.099 4 0.136 0.132 0.081
Mg I . . . . . . 5 0.390 0.394 0.046 0.410 0.417 0.062 2 0.377 0.363 0.026
Al I . . . . . . . 2 0.334 0.335 0.013 0.345 0.340 0.010 2 0.414 0.400 0.063
Si I . . . . . . . . 13 0.218 0.215 0.053 0.245 0.245 0.053 10 0.180 0.179 0.074
S I ........ 1 0.607 0.608 ... 0.620 0.615 ... 1 0.248 0.233 ...
Ca I . . . . . . 14 0.191 0.169 0.102 0.221 0.205 0.103 14 0.191 0.195 0.091
Sc II . . . . . . 6 0.398 0.408 0.054 0.440 0.440 0.063 6 0.183 0.184 0.069
Ti I . . . . . . . 43 0.244 0.249 0.051 0.233 0.238 0.055 34 0.279 0.273 0.073
Ti II . . . . . . 14 0.388 0.379 0.061 0.391 0.388 0.058 14 0.234 0.230 0.086
V I ........ 14 0.078 0.079 0.061 0.095 0.085 0.058 13 0.189 0.159 0.052
Cr I . . . . . . . 10 0.017 0.008 0.108 0.024 0.001 0.108 10 0.011 [0.029 0.099
Cr II . . . . . . 6 0.175 0.200 0.091 0.180 0.193 0.096 6 0.066 0.041 0.048
Mn I . . . . . . 11 [0.331 [0.336 0.064 [0.307 [0.336 0.065 9 [0.166 [0.154 0.062
Fe I . . . . . . . 77 [0.401 [0.402 0.069 [0.413 [0.408 0.072 69 [0.392 [0.377 0.085
Fe II . . . . . . 20 [0.383 [0.389 0.062 [0.405 [0.393 0.068 19 [0.365 [0.371 0.081
Co I . . . . . . 11 0.197 0.203 0.145 0.224 0.221 0.147 8 0.131 0.117 0.055
Ni I . . . . . . . 28 0.076 0.057 0.074 0.094 0.067 0.075 27 0.042 0.041 0.071
Cu I . . . . . . 2 0.076 0.077 0.011 0.077 0.072 0.002 2 0.142 0.127 0.112
Zn I . . . . . . 2 0.239 0.239 0.104 0.252 0.247 0.101 2 0.226 0.212 0.047
Y II . . . . . . . 1 0.215 0.216 ... 0.246 0.241 ... 0
Ba II . . . . . . 2 0.134 0.135 0.021 0.163 0.158 0.038 1 [0.087 [0.102 ...
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2527
TABLE 13
ABUNDANCES FOR G114-19
O I........ 3 0.513 0.525 0.033 0.498 0.512 0.032 3 0.355 0.369 0.034
Na I . . . . . . 3 0.148 0.176 0.039 0.159 0.189 0.057 3 0.148 0.150 0.043
Mg I . . . . . . 6 0.452 0.470 0.079 0.475 0.478 0.080 2 0.372 0.377 0.033
Al I . . . . . . . 1 0.374 0.383 ... 0.373 0.379 ... 1 0.471 0.477 ...
Si I . . . . . . . . 14 0.229 0.231 0.058 0.264 0.255 0.057 11 0.184 0.189 0.064
S I ........ 1 0.166 0.175 ... 0.154 0.160 ... 1 [0.212 [0.206 ...
Ca I . . . . . . 16 0.167 0.199 0.096 0.198 0.221 0.094 16 0.168 0.195 0.072
Sc II . . . . . . 7 0.345 0.380 0.068 0.394 0.420 0.073 7 0.146 0.161 0.055
Ti I . . . . . . . 41 0.365 0.370 0.061 0.368 0.360 0.068 33 0.370 0.372 0.074
Ti II . . . . . . 12 0.373 0.410 0.112 0.404 0.423 0.116 10 0.204 0.184 0.092
V I ........ 15 0.165 0.184 0.072 0.185 0.191 0.069 12 0.278 0.279 0.056
Cr I . . . . . . . 13 0.073 0.076 0.083 0.074 0.065 0.091 13 0.008 0.029 0.063
Cr II . . . . . . 5 0.135 0.158 0.082 0.137 0.152 0.080 5 0.006 0.011 0.077
Mn I . . . . . . 10 [0.280 [0.281 0.035 [0.267 [0.280 0.045 8 [0.124 [0.107 0.054
Fe I . . . . . . . 79 [0.595 [0.604 0.065 [0.589 [0.595 0.069 71 [0.567 [0.573 0.084
Fe II . . . . . . 20 [0.622 [0.616 0.063 [0.591 [0.597 0.066 20 [0.598 [0.597 0.086
Co I . . . . . . 9 0.164 0.175 0.059 0.194 0.199 0.065 7 0.147 0.138 0.050
Ni I . . . . . . . 28 0.058 0.061 0.072 0.071 0.071 0.074 27 0.032 0.039 0.077
Cu I . . . . . . 1 0.014 0.023 ... 0.035 0.041 ... 1 [0.006 0.000 ...
Zn I . . . . . . 2 0.105 0.114 0.041 0.107 0.113 0.037 2 0.074 0.080 0.016
Y II . . . . . . . 1 [0.020 [0.011 ... [0.020 [0.014 ... 0
Ba II . . . . . . 2 [0.019 [0.009 0.054 [0.016 [0.010 0.050 2 [0.155 [0.150 0.070
Eu II . . . . . . 1 0.475 0.484 ... 0.530 0.536 ... 1 0.343 0.349 ...
solely on the laboratory measurements by Smith & Raggett 1997), theoretically it is expected to be synthesized predomi-
(1981) while the titanium gf-values were gleaned from a nantly in moderate mass (B20 M ) Type II SN (Woosley &
number of sources with the Oxford measurements given Weaver 1995). In Figure 5, we_plot the solar-corrected
highest priority and the Grevesse et al. (1988) correction silicon abundances versus [Fe/H] for the thick disk stars.
included. The majority are signicantly enhanced and there is an
Silicon is a prototypical a-element. In addition to indication of higher [Si/Fe] at lower metallicity as found in
exhibiting an enhancement in metal-poor stars (McWilliam most metal-poor stellar abundance studies. In terms of the
TABLE 14
ABUNDANCES FOR G144-52
O I........ 2 0.462 0.467 0.010 0.450 0.451 0.015 2 0.341 0.324 0.023
Na I . . . . . . 4 0.111 0.140 0.058 0.151 0.170 0.074 4 0.112 0.111 0.058
Mg I . . . . . . 7 0.374 0.388 0.090 0.387 0.407 0.079 3 0.275 0.264 0.056
Al I . . . . . . . 2 0.245 0.250 0.024 0.253 0.254 0.025 2 0.324 0.307 0.026
Si I . . . . . . . . 15 0.190 0.201 0.043 0.224 0.221 0.046 12 0.173 0.159 0.048
S I ........ 1 0.756 0.761 ... 0.748 0.749 ... 1 0.395 0.379 ...
Ca I . . . . . . 18 0.144 0.155 0.075 0.166 0.182 0.071 18 0.169 0.163 0.071
Sc II . . . . . . 8 0.360 0.383 0.059 0.382 0.396 0.066 8 0.192 0.169 0.060
Ti I . . . . . . . 43 0.261 0.261 0.053 0.250 0.254 0.055 37 0.288 0.277 0.087
Ti II . . . . . . 12 0.411 0.419 0.077 0.404 0.414 0.076 11 0.263 0.248 0.056
V I ........ 13 0.085 0.090 0.044 0.097 0.103 0.044 11 0.165 0.164 0.048
Cr I . . . . . . . 15 0.044 0.048 0.064 0.056 0.039 0.073 14 0.018 0.000 0.044
Cr II . . . . . . 6 0.145 0.178 0.085 0.134 0.152 0.089 6 0.048 0.011 0.051
Mn I . . . . . . 9 [0.371 [0.379 0.034 [0.367 [0.374 0.043 7 [0.209 [0.214 0.073
Fe I . . . . . . . 87 [0.558 [0.563 0.067 [0.570 [0.571 0.068 78 [0.547 [0.531 0.088
Fe II . . . . . . 19 [0.555 [0.550 0.061 [0.575 [0.579 0.065 19 [0.548 [0.535 0.090
Co I . . . . . . 7 0.188 0.220 0.135 0.208 0.234 0.138 6 0.189 0.180 0.047
Ni I . . . . . . . 26 0.046 0.057 0.057 0.062 0.063 0.054 25 0.030 0.030 0.067
Cu I . . . . . . 2 [0.018 [0.013 0.055 [0.017 [0.016 0.058 2 0.046 0.029 0.046
Zn I . . . . . . 2 0.107 0.112 0.046 0.093 0.094 0.044 2 0.093 0.076 0.011
Y II . . . . . . . 4 [0.004 0.026 0.103 0.005 0.026 0.114 1 [0.060 [0.076 ...
Ba II . . . . . . 3 0.032 0.064 0.105 0.009 0.051 0.120 2 [0.118 [0.135 0.081
Eu II . . . . . . 1 0.361 0.366 ... 0.392 0.393 ... 1 0.246 0.230 ...
2528 PROCHASKA ET AL. Vol. 120
TABLE 15
ABUNDANCES FOR G181-46
O I........ 3 0.444 0.465 0.047 0.445 0.467 0.040 3 0.298 0.309 0.023
Na I . . . . . . 3 0.064 0.061 0.016 0.055 0.065 0.014 3 0.028 0.033 0.028
Mg I . . . . . . 4 0.396 0.418 0.063 0.396 0.425 0.063 2 0.352 0.351 0.070
Al I . . . . . . . 2 0.307 0.310 0.048 0.290 0.294 0.053 2 0.369 0.367 0.002
Si I . . . . . . . . 18 0.250 0.252 0.045 0.286 0.276 0.045 15 0.197 0.214 0.063
S I ........ 1 0.419 0.422 ... 0.413 0.417 ... 1 0.041 0.040 ...
Ca I . . . . . . 16 0.204 0.217 0.092 0.206 0.238 0.090 16 0.219 0.216 0.078
Sc II . . . . . . 9 0.385 0.395 0.091 0.420 0.424 0.094 9 0.200 0.196 0.052
Ti I . . . . . . . 50 0.340 0.347 0.044 0.318 0.323 0.045 41 0.376 0.365 0.073
Ti II . . . . . . 10 0.438 0.446 0.066 0.449 0.463 0.067 9 0.265 0.271 0.063
V I ........ 16 0.181 0.186 0.086 0.157 0.165 0.083 12 0.250 0.262 0.068
Cr I . . . . . . . 13 0.098 0.078 0.074 0.100 0.064 0.080 12 0.005 0.027 0.064
Cr II . . . . . . 5 0.136 0.188 0.132 0.143 0.189 0.134 5 0.033 0.016 0.041
Mn I . . . . . . 10 [0.328 [0.327 0.061 [0.340 [0.339 0.061 8 [0.179 [0.154 0.054
Fe I . . . . . . . 94 [0.524 [0.527 0.073 [0.508 [0.512 0.072 85 [0.496 [0.495 0.084
Fe II . . . . . . 18 [0.525 [0.526 0.081 [0.497 [0.495 0.082 16 [0.534 [0.513 0.090
Co I . . . . . . 8 0.175 0.164 0.066 0.175 0.163 0.071 6 0.147 0.147 0.020
Ni I . . . . . . . 31 0.084 0.086 0.071 0.087 0.096 0.070 30 0.046 0.057 0.072
Cu I . . . . . . 2 [0.032 [0.029 0.016 [0.033 [0.029 0.013 2 0.014 0.013 0.086
Zn I . . . . . . 2 0.162 0.165 0.053 0.181 0.184 0.047 2 0.130 0.129 0.004
Y II . . . . . . . 2 0.133 0.136 0.071 0.146 0.150 0.088 1 0.127 0.126 ...
Ba II . . . . . . 2 0.184 0.187 0.077 0.243 0.247 0.100 1 [0.096 [0.097 ...
uncertainty in the atmospheric parameters, a decrease in enhanced Ti and Mg. These trends might be indicative of an
T of 100 K would further enhance the Si/Fe ratio by overestimate of temperature for these two stars because Si,
eff
0.06[0.08 dex and the ratio is largely insensitive to the Ca, and O are most sensitive to T in G dwarf stars.
eff in metal-poor stars
other parameters. As discussed in the following section, we Like silicon, Ca is enhanced
contend that the majority of stars are even more enhanced (McWilliam 1997) and is predicted to be produced in inter-
than their thin disk counterparts at the same metallicity. mediate mass Type II SN along with silicon (Woosley &
The obvious exceptions are the two highest metallicity stars Weaver 1995). Not surprisingly, then, the abundance trends
(G88-13, G211-5) which exhibit relatively low [Si/Fe]- that we observe for silicon are well matched by calcium.
values. We shall note, however, that these two stars also Figure 6 presents the solar-corrected Ca abundances rela-
show lower values of calcium and oxygen yet signicantly tive to Fe versus [Fe/H]. All of the stars exhibit enhanced
TABLE 16
ABUNDANCES FOR G211-5
O I........ 2 0.265 0.265 0.023 0.233 0.234 0.013 2 0.112 0.114 0.012
Na I . . . . . . 2 0.051 0.051 0.021 0.073 0.075 0.043 2 0.037 0.040 0.002
Mg I . . . . . . 4 0.289 0.288 0.037 0.295 0.288 0.036 1 0.188 0.191 ...
Al I . . . . . . . 2 0.146 0.146 0.011 0.138 0.140 0.015 2 0.208 0.211 0.040
Si I . . . . . . . . 13 0.115 0.110 0.050 0.137 0.126 0.051 10 0.061 0.083 0.046
S I ........ 1 0.525 0.525 ... 0.501 0.502 ... 1 0.147 0.150 ...
Ca I . . . . . . 12 0.093 0.088 0.110 0.106 0.109 0.111 12 0.116 0.100 0.076
Sc II . . . . . . 7 0.252 0.279 0.096 0.275 0.295 0.104 7 0.032 0.063 0.083
Ti I . . . . . . . 46 0.176 0.160 0.072 0.160 0.146 0.077 37 0.193 0.182 0.089
Ti II . . . . . . 12 0.209 0.231 0.078 0.208 0.237 0.084 11 0.092 0.082 0.066
V I ........ 14 0.096 0.080 0.071 0.102 0.084 0.071 11 0.189 0.188 0.045
Cr I . . . . . . . 11 0.051 0.034 0.091 0.039 0.028 0.096 10 0.002 [0.001 0.068
Cr II . . . . . . 4 0.145 0.116 0.117 0.128 0.092 0.118 4 [0.027 [0.026 0.093
Mn I . . . . . . 8 [0.237 [0.247 0.083 [0.245 [0.251 0.091 8 [0.077 [0.076 0.058
Fe I . . . . . . . 68 [0.402 [0.402 0.074 [0.382 [0.383 0.079 60 [0.374 [0.377 0.081
Fe II . . . . . . 18 [0.407 [0.409 0.068 [0.393 [0.401 0.070 18 [0.402 [0.395 0.096
Co I . . . . . . 6 0.011 0.053 0.084 0.016 0.062 0.090 5 0.052 0.047 0.048
Ni I . . . . . . . 28 0.021 0.016 0.090 0.033 0.027 0.090 26 [0.008 [0.002 0.073
Cu I . . . . . . 2 0.000 0.000 0.033 [0.002 [0.001 0.028 2 0.046 0.049 0.069
Zn I . . . . . . 2 0.009 0.009 0.073 0.000 0.001 0.071 2 [0.022 [0.019 0.016
Y II . . . . . . . 4 [0.034 [0.021 0.076 [0.035 [0.024 0.083 1 [0.054 [0.051 ...
Ba II . . . . . . 2 0.016 0.016 0.012 0.012 0.013 0.025 2 [0.121 [0.118 0.112
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2529
TABLE 17
ABUNDANCES FOR G247-32
O I........ 3 0.586 0.548 0.066 0.560 0.524 0.061 3 0.407 0.396 0.048
Na I . . . . . . 3 0.097 0.101 0.013 0.101 0.107 0.038 3 0.087 0.077 0.025
Mg I . . . . . . 6 0.434 0.435 0.046 0.439 0.440 0.051 2 0.365 0.366 0.057
Al I . . . . . . . 2 0.309 0.308 0.030 0.300 0.295 0.038 2 0.369 0.369 0.020
Si I . . . . . . . . 17 0.260 0.261 0.059 0.304 0.275 0.062 14 0.193 0.224 0.073
S I ........ 1 0.581 0.580 ... 0.555 0.550 ... 1 0.200 0.201 ...
Ca I . . . . . . 13 0.187 0.196 0.105 0.204 0.224 0.104 13 0.216 0.197 0.089
Sc II . . . . . . 7 0.382 0.410 0.071 0.393 0.420 0.076 7 0.216 0.206 0.029
Ti I . . . . . . . 48 0.288 0.295 0.059 0.279 0.283 0.059 40 0.326 0.312 0.077
Ti II . . . . . . 15 0.394 0.409 0.089 0.397 0.408 0.088 14 0.242 0.229 0.080
V I ........ 15 0.154 0.194 0.126 0.159 0.188 0.124 12 0.264 0.269 0.064
Cr I . . . . . . . 14 0.075 0.064 0.108 0.067 0.062 0.115 13 0.008 0.014 0.084
Cr II . . . . . . 5 0.175 0.192 0.104 0.164 0.168 0.105 5 0.046 0.048 0.076
Mn I . . . . . . 11 [0.313 [0.303 0.062 [0.300 [0.312 0.065 9 [0.153 [0.129 0.059
Fe I . . . . . . . 82 [0.419 [0.418 0.077 [0.394 [0.389 0.081 74 [0.388 [0.389 0.084
Fe II . . . . . . 16 [0.414 [0.405 0.059 [0.396 [0.385 0.063 16 [0.396 [0.381 0.094
Co I . . . . . . 14 0.189 0.222 0.129 0.194 0.222 0.131 9 0.159 0.151 0.062
Ni I . . . . . . . 28 0.078 0.065 0.064 0.098 0.076 0.064 26 0.043 0.039 0.071
Cu I . . . . . . 2 0.084 0.082 0.016 0.086 0.080 0.019 2 0.058 0.059 0.020
Zn I . . . . . . 2 0.112 0.110 0.076 0.120 0.115 0.073 2 0.078 0.079 0.019
Y II . . . . . . . 2 [0.015 [0.017 0.040 [0.015 [0.021 0.028 1 [0.103 [0.102 ...
Ba II . . . . . . 2 0.059 0.057 0.036 0.082 0.076 0.064 2 [0.081 [0.080 0.088
Eu II . . . . . . 1 0.537 0.536 ... 0.558 0.553 ... 1 0.402 0.403 ...
Ca and, similar to Si, there is a mild trend to higher [Ca/Fe] (Woosley & Weaver 1995). Therefore, it would not be sur-
at lower [Fe/H]. Also similar to the silicon results, the two prising if the Ti abundances dier from the results for Si and
highest metallicity stars show somewhat lower [Ca/Fe]. In Ca. In almost every star (G114-19 is an exception), the v(Ti)
contrast to most of the other well measured elements, the abundance derived from Ti II exceeds that from Ti I by
Ca measurements for a given star exhibit a fairly large B0.100.15 dex. This oset has been discussed in the liter-
scatter. We expect this is due to a greater uncertainty in the ature (e.g., Luck & Bond 1985) and has been attributed to
equivalent width measurements for the Ca I lines which non-LTE eects and other possible systematic errors. We
often have signicant damping wings. nd a similar oset in our solar analysis such that the solar-
Titanium is traditionally referred to as an a-element corrected v(Ti)-values from Ti I and Ti II abundances are in
because it exhibits enhanced abundances in metal-poor good agreement for the majority of stars. Because the Ti I
stars (Gratton & Sneden 1991), but it is unclear if the results are statistically more robust, we restrict our further
nucleosynthesis of Ti is related to the other a-elements analysis to the Ti I results. In Figure 7 we present the solar-
corrected [Ti/Fe] measurements as a function of metallicity
FIG. 7.Solar-corrected [Ti/Fe] abundance ratios vs. [Fe/H] metal- FIG. 8.Standard (uncorrected) [O/Fe] abundance ratios vs. [Fe/H]
licity for the 10 thick disk stars. The observations show a clear enhance- metallicity for the 10 thick disk stars. The observations show a clear
ment yet no trend with metallicity. The dashed line at [Ti/Fe] \ 0 enhancement and a likely trend with metallicity. Note the error bars only
indicates the solar meteoritic Ti/Fe ratio. reect the scatter in the v(O)-values derived from the three O I lines at
j B 7775 A for which the systematic uncertainties are very signicant
([0.1 dex). The dashed line at [O/Fe] \ 0 indicates the solar meteoritic
for the 10 thick disk stars. All of the observations are consis- O/Fe ratio.
tent with a single-valued enhancement, S[Ti/Fe]T \
0.29 ^ 0.02, and there is no indication of a trend with metal- to improve on the oxygen measurements, either through the
licity. The latter observation contradicts the general picture forbidden O I [6300] line or perhaps the near-IR OH lines.
described by Si and Ca. As noted in 3.4, measurements of magnesium are partic-
5.2.2. Oxygen, Magnesium, Sulfur ularly important because Mg is a signicant contributor of
For our choice of observational setup the forbidden O I electrons in the stellar atmospheres of our stars. Unfor-
j6300 line lies within the inter-order gaps of HIRES. There- tunately, there are very few Mg I lines with reported gf-
fore, we rely on the triplet of O I lines at j B 7775 A , which values that we do not nd saturated. Therefore, we are
have very high excitation potential and are very sensitive to compelled to include a few lines with solar gf-values taken
the eective temperature (Tables 5 and 6) and non-LTE from the literature (Edvardsson et al. 1993 ; McWilliam et
eects. We account for CO, CH, and OH molecule forma- al. 1995b). In general, we nd good agreement between the
tion in deriving our nal oxygen abundances which leads to various lines and have reasonable condence in our results.
an enhancement of B]0.025 dex over the abundance Figure 9 plots the [Mg/Fe]-values versus [Fe/H] for the 10
derived without molecules. For the Sun, adopting the labor- program stars. We observe enhanced Mg in every case,
atory gf-values from Beveridge & Sneden (1994) and Butler S[Mg/Fe]T \ ]0.37 ^ 0.02, with no suggestion of a trend
& Zeippen (1991) we nd a median v(O) value of [O/H] \ with metallicity. It should be noted, however, that the Mg
]0.10 dex with very small scatter. Given the large uncer- results for the two most metal-poor stars are derived from a
tainty associated with using the O I triplet for an oxygen dierent set of Mg I lines than the other stars. Given the
abundance analysis, the agreement between our solar
analysis and the meteoritic value is surprisingly good.
Figure 8 plots the [O/Fe]-values for the thick disk stars
against the stellar metallicity without the solar-correction.
We do not apply the solar-correction here because we
believe the uncertainty in the solar measurement from the
O I triplet is at least as large as 0.1 dex and therefore we
would be more likely to introduce an error in the nal
abundances. The error bars reect the scatter in the individ-
ual measurements for each star which in each case is signi-
cantly smaller than the uncertainties from the atmospheric
parameters. In particular, even a 50 K error in T results in
nearly a 0.1 dex uncertainty in [O/Fe]. Weeffpoint out,
however, that with the exception of the most metal-poor
star, the spectroscopic T values are at least as high as the
eff
photometric values, implying if anything that we have
underestimated the [O/Fe] ratios. Examining the gure,
one notes that the majority of stars exhibit enhanced
[O/Fe] abundances with tentative evidence for an increas-
FIG. 9.Standard (uncorrected) [Mg/Fe] abundance ratios vs. [Fe/H]
ing oxygen abundance at lower metallicity. The two stars metallicity for the 10 thick disk stars. The observations show a clear
with the highest metallicity, however, have nearly solar enhancement and no trend with metallicity. The dashed line at
oxygen abundance. It will be imperative in the future for us [Mg/Fe] \ 0 indicates the solar meteoritic Mg/Fe ratio.
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2531
uncertainty in the gf-values we adopted and the fact that we 5.3. L ight ElementsAl, Na
could not perform a solar analysis it is possible that there is With the exception of the extremely metal-poor stars
a systematic error in comparing against the two metal-poor (Gratton & Sneden 1988 ; McWilliam et al. 1995b ; Shetrone
stars although there is no evidence of any oset. The Mg/Fe 1996), Al is mildly enhanced in metal-poor stars (Tomkin et
ratio is insensitive to uncertainties in the atmospheric pa- al. 1985 ; Edvardsson et al. 1993). As such, Al is sometimes
rameters and we believe the observed enhancement is classied as an a-element. Contrary to the majority of Al
robust aside from possible errors in the gf-values. Because studies, however, Chen et al. (2000) found an enhancement
Mg is a principal source of electrons in the stellar atmo- in Al in disk stars with [Fe/H] D 0 and no enhancement in
spheres of our stars and we observe an enhancement of their metal-poor F and G dwarfs. The Chen et al. (2000)
Mg/Fe in every case, it is important to consider a-enhanced analysis primarily focuses on the pair of Al I lines at 7830 A ,
stellar atmospheres as we have done throughout our while Edvardsson et al. (1993) examined the pair near 8773
analysis. A . Unfortunately, all of these transitions fall into the inter-
The difficulties associated with sulfur are even more dire order gasps of our echelle setup. Instead our observations
than the problems associated with oxygen and magnesium : cover three other Al I absorption lines, all with measured
there is only one useful transition (S I j8694) ; it lies toward laboratory gf-values (Buurman et al. 1986) : jj \ 5577,
the red end of the spectrum where the sensitivity of HIRES 6696, 6698 A . In general, we nd good agreement for the Al
is markedly reduced ; it has a high excitation potential with abundances derived from the two lines at j B 6700 A , but
a correspondingly large temperature sensitivity ; it is very the j5577 line yields systematically lower values. Unfor-
weak (only 30 mA in the Sun) ; and there is no reliable tunately, this line is too badly blended in the Sun to deter-
laboratory gf value so a solar analysis is required. We rst mine a solar-correction. As such, we have decided to
adopted the gf value from Francois (1988) for the standard remove the j5577 line from our analysis. For the Sun, the
and a-enhanced values reported in Tables 717, but our two remaining lines yield a solar Al abundance B0.1 dex
solar analysis shows [S/Fe] B ]0.2 dex indicating a signi- lower than the meteoritic value therefore we have applied a
cant correction to the gf value. Figure 10 plots the [S/Fe] 0.1 dex oset in the solar-corrected values. In Figure 11 we
abundances for the nine stars with a measured sulfur equiv- plot the solar-corrected [Al/Fe]-values for the nine stars
alent width with the solar-corrected values plotted. In a few where we measured at least one of the lines at j B 6700 A ;
cases, we have included W values less than 10 mA . For the Al I lines are too weak in the most metal-poor star. All
j error even exceeds the uncer-
these stars, the measurement of the stars exhibit signicant Al enhancements, S[Al/
tainties due to errors in the atmospheric parameters. We Fe]T \ ]0.334 ^ 0.028 and there is no trend with metal-
estimate the total 1 p uncertainty to be 0.3 dex and have licity. Our error analysis indicates the Al/Fe ratio is
plotted the error bars accordingly. Given the large errors insensitive to the atmospheric parameters ; the principal
associated with the [S/Fe] measurements, it is difficult to uncertainty lies in the paucity of Al I lines. There is a further
make any meaningful statements about the sulfur abun- systematic error associated with these Al I lines, however. A
dance. It is somewhat surprising, however, that the mean non-LTE analysis by Baumuller & Gehren (1997) suggests a
ratio S[S/Fe]T \ 0.11 ^ 0.08 is consistent with the solar further enhancement to [Al/Fe] of B0.15 dex for stars with
abundance. Given the importance of sulfur in quasar metallicity [Fe/H] B [0.5 dex. We have chosen not to
absorption-line studies ( 6.4), a more careful and extensive include this non-LTE correction in our analysis, but warn
stellar abundance analysis of sulfur in metal-poor stars is the reader that the reported Al/Fe-values may be a lower
warranted. limit to the true ratio.
The observational picture for Na is complicated. The
majority of studies on sodium (Tomkin et al. 1985 ; McWil-
liam et al. 1995b ; Chen et al. 2000) report that Na scales underabundance observed for Cr in very metal-poor halo
with Fe at all metallicities, although Edvardsson et al. stars (e.g., McWilliam et al. 1995b). We note in passing that
(1993) found a mild Na enhancement at [Fe/H] B [1 and the Cr abundance based on the Cr II absorption lines
Pilachowski et al. (1996) report a Na/Fe deciency in a suggest enhanced [Cr/Fe], typically by B]0.1 dex. The
sample of 60 stars with [Fe/H] \ [1. Recently, Baumuller solar measurements of Cr II, however, exhibit the same
et al. (1998) performed a non-LTE analysis of Na and report enhancement and therefore we propose that the Cr II gf-
a trend of decreasing Na with decreasing metallicity. We values of Martin et al. (1988) may need to be revised upward
have measured Na in our thick disk stars based on the by 0.1 dex.
observations of four Na I transitions : jj5682, 5688, 6154, The majority of iron-peak elements, however, exhibit sig-
6160. None of the gf-values are secure for these lines ; the nicant departures from the solar ratios and some have
latter two are from theoretical work by Lambert & Luck clear trends with the stellar metallicity. As these departures
(1978) and we adopt solar gf-values for jj5682, 5688. There- oer insights into the nucleosynthetic processes and may
fore, the analysis relative to our solar analysis is essential. distinguish the thick disk stars from other stellar popu-
Figure 12 presents the solar-corrected [Na/Fe]-values lations, we consider each element in greater detail.
versus metallicity for the 10 stars. Every star exhibits mildly
enhanced Na with an average S[Na/Fe]T \ 5.4.1. Scandium
]0.087 ^ 0.014. The overabundance is statistically signi- Previous studies of scandium have disagreed on the
cant but the systematic uncertainty (e.g., in the gf-values) is metallicity dependence of [Sc/Fe]. Zhao & Magain (1990)
on the order of the enhancement. The Na/Fe ratio is rst suggested that Sc was enhanced by B]0.25 in metal-
remarkably insensitive to the atmospheric parameters poor stars based on their analysis of four Sc II lines includ-
therefore the results are probably limited by the small ing several of the lines that we have analyzed. Gratton &
number statistics of measuring only three Na I lines. If we Sneden (1991), however, argued that this enhancement was
were to apply the results of the non-LTE analysis by Bau- primarily due to inaccuracies in the Sc II gf-values. Our
muller et al. (1998), it is also possible that the observed Na solar analysis agrees with this assessment ; we nd
enhancement would vanish. In summary, we consider the [Sc/H] B ]0.2 dex using the gf-values from Martin et al.
Na/Fe results to be rather poorly constrained yet consistent _ Lawler & Dakin (1989). Most recently, Nissen et
(1988) and
with no signicant departure from the solar abundance. al. (2000) published an analysis of Sc in over 100 G and F
dwarf stars with metallicities ranging from [Fe/H] B [1.4
5.4. Iron-Peak ElementsSc, V, Cr, Mn, Co, Ni, Cu, Zn 0. Their study focused on ve Sc II lines (jj5239, 5526, 5657,
We now turn our attention to the elements with atomic 6245, 6604) compared against a solar analysis. Their results,
number near Fe, the so-called iron-peak elements. For based on an hfs analysis from Steen (1985), indicate an
several of these elements we observe no variations from enhancement of [Sc/Fe] with a trend that resembles the
solar abundance irrespective of the thick disk stellar metal- a-enhancement of metal-poor stars. Compared to our hfs
licity. In particular, the relative nickel and chromium abun- analysis (Table 18), however, the Steen (1985) compilation
dances show no signicant departure from solar overestimates the hfs correction for these Sc II lines, which
abundances for any of the thick disk stars. There is a small may account for a signicant part if not all of the reported
oset (]0.03 dex) from solar for the [Ni/Fe]- and [Cr/Fe]- trend (Prochaska & McWilliam 2000). Figure 13 presents
values in the standard analysis, but we observe a similar [Sc/Fe] for our full sample as a function of metallicity for
dierence in the solar analysis such that the corrected abun- the standard (plus sign) and solar-corrected (times cross)
dances are within 0.02 dex of solar for every star but one. values ; the latter reveal a somewhat puzzling picture. While
The most metal-poor star (G84-37) exhibits a low [Cr/Fe] the eight stars at [Fe/H] B [0.5 exhibit enhanced scan-
value which is probably signicant and is suggestive of the dium, S[Sc/Fe]T \ ]0.14 ^ 0.02, the most metal-poor
stars show only a minor enhancement. This dierence could
be explained by unidentied blends, but it seems very
unlikely given the reasonably close agreement of more than
ve Sc II lines in each star and because it would be difficult
to reconcile blends with the solar v(Sc) results. An underesti-
TABLE 18
HFS TABLES
j EP
Ion (A ) (eV) log gf
FIG. 12.Solar-corrected [Na/Fe] abundance ratios vs. [Fe/H] metal- NOTE.Table 18 is presented in its entirety
licity for the 10 thick disk stars. All of the stars show enhanced Na/Fe and in the electronic edition of the Astronomical
there is a mild trend with metallicity. The dashed line at [Zn/Fe] \ 0 Journal. A portion is shown here for guidance
indicates the solar meteoritic Zn/Fe ratio. regarding its form and content.
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2533
dances against other stellar populations, namely the metal- 5. For the O abundances, we have restricted our com-
poor halo stars, the thin disk stars, and a small sample of parisons to those measurements based on the O I lines at
bulge stars. In the following subsections, we comment on 7770 A as there are concerns over systematic dierences
the implications for the formation history of the thick disk between the various approaches to measuring oxygen
and Galaxy as well as the nucleosynthesis of these elements (Tomkin et al. 1992 ; Abia & Rebolo 1989) ;
in the early universe. 6. The bulge abundances are taken from McWilliam &
Figures 22a22c present abundances for 18 elements with Rich (1994) ;
our sample of thick disk stars plotted as large black stars. 7. The open squares represent values taken from the
We do not include S in this subsection because of the very remaining literature sources (Brown & Wallerstein 1992 ;
large uncertainties associated with these measurements in McWilliam et al. 1992 ; Shetrone 1996).
all stellar populations. In each subpanel, we overplot abun-
For all of the data sets we have normalized the measure-
dance measurements taken from a number of literature
ments to the updated Grevesse et al. (1996) solar meteoritic
sources which describe the metal-poor halo (green), the thin
abundance scale as necessary.
disk (red), the thick disk (black), and the bulge (blue). Abun-
It is difficult to digest all of the abundance trends present-
dance trends are sensitive to the stellar atmosphere, spec-
ed in Figures 22a22c, yet several comparisons stand out. In
trum synthesis algorithm, hyperne splitting corrections,
terms of the a-elements, the thick disk stars exhibit signi-
etc., adopted by each group and therefore systematic uncer-
cantly larger overabundances than the thin disk stars. The
tainties exist in comparing various data sets. It is important
O and Mg results agree with the overabundances claimed
at the least to keep this uncertainty in mind and if possible
by previous thick disk studies (Gratton et al. 2000 ; Fuhr-
to correct for identiable systematic osets. The plot
mann 1998) while the Ca and Ti ratios lend even further
symbols distinguish the various data sets as follows.
support that the thick disk is chemically distinct from the
thin disk. The Si/Fe ratios are the least enhanced over the
1. The Edvardsson et al. (1993) data are plotted as open
thin disk, yet other than the two lowest Si/Fe-values even
circles. We have noted ( 3.4) a systematic oset between
these values are higher than the average thin disk star at
their temperature scale and ours of *T D 100 K. We
eff in Table 9 of [Fe/H] B [0.5. Therefore, the a-elements oer convincing
correct for this oset by applying the values
evidence that the thick disk population is chemically dis-
Edvardsson et al. (1993) which gives the sensitivity of
crete from the thin disk. This point contradicts the interpre-
various element ratios on T . We have identied thick disk
eff(1993) sample according to the tation of Chen et al. (2000) who comment that the thick disk
stars in the Edvardsson et al.
abundances extend smoothly from the thin disk. As noted
same kinematic criteria imposed for our sample (Z [ 600
max We above, however, the Chen et al. (2000) sample of thick disk
pc ; [100 \ V3 \ 20 km s~1 ; [1.1 \ [M/H] \ [0.4).
stars are younger than the ducial age of the thick disk
also identied a few halo stars which have V3 \ [1000 km
casting their conclusions into serious doubt. This point not
s~1 and large eccentricity. The remaining stars are plotted
withstanding, carefully examining Figure 22 one notes that
as thin disk stars if [Fe/H] [ [0.7 and as halo stars
the thick disk stars from Chen et al. (2000) all have metal-
otherwise. We warn, however, that many of these stars at
licity [Fe/H] \ [0.6 where a comparison with the thin
[Fe/H] B [0.7 could be thick disk members.
disk is difficult owing to the paucity of measurements of
2. The triangles represent the chemical abundances from
thin disk stars with [Fe/H] \ [0.6. In contrast with the
Chen et al. (2000). We believe Chen et al. (2000) adopted a
thin disk, the halo star values are in much better agreement
similar temperature scale to ours so we make no correction
with the thick disk stars. The Mg and Ti abundances show
to this data set. Chen et al. (2000) identied a small sub-
similar enhancements for the two populations while the
sample of thick disk stars according to kinematic and
remaining a-elements approach the halo star values in the
metallicity criteria ; these are plotted as open red triangles in
most metal-poor thick disk stars. Finally, we nd an
the gure. As noted in the introduction, these stars have
impressive match between the bulge and thick disk a-
implied ages younger than that typically assumed for the
element abundances.10 The Ca, Mg, and Ti trends are all in
thick disk and therefore we contend that they are actually
agreement, while the thick disk Si abundances show some-
thin disk stars ;
what smaller enhancements. To summarize our results on
3. The solid circles depict the values taken from Gratton
the a-elements, of the four stellar populations the thick disk
& Sneden (1988), Gratton & Sneden (1991), Sneden et al.
and bulge star abundance patterns most resemble one
(1991), and Gratton & Sneden (1994). Because these authors
another, the most metal-poor thick disk stars closely match
did not characterize these eld stars according to specic
the halo values, and all of these populations show a-
stellar populations, we have plotted them under the
enhancements which are signicantly larger than the values
assumption that those stars with [Fe/H] \ [0.7 are halo
observed for thin disk stars.
stars and the remaining are thin disk stars. For the few stars
As with the a-elements, the light elementssodium and
near [Fe/H] \ [0.7 it is possible that there is some con-
aluminumare enhanced in the thick disk stars. The Na/Fe
tamination from thick disk stars ;
results are in good accord with the Edvardsson et al. (1993)
4. The Sc- and Mn-values ( lled squares) are taken from
analysis yet contradict the results from Chen et al. (2000).
Prochaska & McWilliam (2000) who reanalyzed the abun-
Chen et al. (2000) noted this discrepancy between their
dances reported by Nissen et al. (2000). As Prochaska &
results and Edvardsson et al. (1993), but did not determine
McWilliam (2000) emphasize, their values may not rep-
the cause. With respect to the halo and bulge, the thick disk
resent the true Mn and Sc abundances, but they should be
more accurate than the Nissen et al. (2000) values which are
based on an incorrect hfs treatment. For these measure- 10 There is the notable exception of oxygen, yet A. McWilliam &
ments, we have adopted the stellar population identica- R. Rich (2000, private communication) now contend that their original
tions given by Nissen et al. (2000) ; O/Fe ratios are in error.
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 dierent symbols refer to the various literature sources indicated in the text. There are three principal points to take from the
gure : (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] B [ 1, and (3) the thick disk abundances are in good agreement with the bulge values for essentially every element.
THICK DISK STELLAR ABUNDANCES 2539
FIG. 22.Continued
values are somewhat larger than the typical halo values vation that none of the populations show signicant Na/Fe
(either solar or subsolar) and the Na/Fe-values from the departures from the solar ratio. The Al measurements, on
metal-poor bulge stars (based on only one or two Na I the other hand, oer more compelling comparison. The
lines). Given the large uncertainties associated with measur- thick disk Al abundances are enhanced over the thin disk
ing Na in each of these stellar populations, we do not feel observations by both Chen et al. (2000) and Edvardsson et
condent in drawing any conclusions aside from the obser- al. (1993) and are in reasonable agreement with the halo star
2540 PROCHASKA ET AL. Vol. 120
FIG. 22.Continued
observations. Finally the majority of Al measurements from the solar abundance. In several ways, the thick disk stars
the bulge stars match the thick disk, although there are signicantly distinguish themselves from the halo and thin
several bulge measurements with [Al/Fe] [ ]0.6 dex. disk stellar populations. First, unlike the thin disk the V,
With respect to the iron-peak measurements of the thick Co, and Zn abundances are all enhanced in the thick disk.
disk, the Ni, Cu, and Cr abundances generally track Fe, yet Second, all but two of the thick disk stars exhibit signi-
the remaining elements all show signicant departures from cantly lower Mn/Fe ratios than the thin disk. Whereas the
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2541
thin disk and halo appear to exhibit plateaus in [Mn/Fe] at enhancement demonstrated in the thick disk indicates the
[0.05 and [0.3 dex, respectively, the thick disk values thin disk stars formed from gas more signicantly polluted
show a trend with metallicity which connects the other two by Type Ia SN requiring a signicant delay between the
populations. In the case of Zn, V, and Co, the thick disk formation epochs. As described below, however, this asser-
ratios are even enhanced over the halo abundances for halo tion is now complicated by the tentative evidence for Type
stars with [Fe/H] [1.8 (the extremely metal-poor halo Ia contributions in the thick disk abundance patterns.
stars also have overabundances of Co and Zn, McWilliam Nevertheless, the fact that the thick disk abundance pat-
et al. 1995b ; Johnson 1999). As discussed in 5.4.1, the terns do not smoothly transition to the thin disk does
thick disk Sc/Fe results are very peculiar. The stars with require a signicant delay in the star formation rate follow-
[Fe/H] B [0.5 dex show a marked Sc/Fe overabundance ing the formation of the thick disk stars. The over-
while the two metal-poor stars have nearly solar Sc/Fe abundance of the a-elements in the thick disk as well as the
ratios. These results are in surprisingly good agreement underabundance of Mn/Fe indicate that the thick disk
with the Sc/Fe results from Prochaska & McWilliam (2000). formed prior to the thin disk on the grounds that there was
The Sc abundances are very unusual and require further less pollution from the long-lived Type Ia SN. In addition,
investigation before one can meaningfully comment on the observed halo-like Ba/Eu ratio (Fig. 23) indicates
their implications. Given the signicant number of distinc- r-process dominated enrichment which supports a great age
tions between the thick disk and the halo/thin disk stellar for these stars.
populations, the agreement between the iron-peak abun- In the abundance studies by Gratton et al. (2000) and
dance patterns of the thick disk and bulge is stunning. In Fuhrmann (1998) there is no indication for a trend of
particular, the metal-poor bulge stars show enhancements [O/Fe] or [Mg/Fe] with metallicity for the thick disk stars.
in V and Co at levels consistent with the thick disk results Gratton et al. (2000) therefore asserted that the formation of
and there is even an indication of enhanced Sc. It must be the thick disk occurred on time-scales shorter than the Type
noted, however, that the iron-peak bulge values are pri- Ia SN, i.e., t \ 1 Gyr. If conrmed, the tentative trends
marily based on one to two lines and that McWilliam & form
of Ca, Si, and O with metallicity in our sample of thick disk
Rich (1994) did not take particular care to account for line stars contradict this assertion. Traditionally, the decrease of
blending. The latter point, in particular, gives pause because an a-element enhancement with increasing [Fe/H] metal-
with the exception of Mn the iron-peak abundances we are licity is attributed to the onset of Type Ia SN. This explana-
considering are enhanced. We anxiously await conrmation tion is well motivated by our theoretical understanding of
of these trends by McWilliam & Rich (2000) who have Type Ia and Type II SN. If it explains the Si, Ca, and O
recently obtained a sample of bulge stars observed with trends for the thick disk, it will be difficult to reconcile our
HIRES on Keck I at signicantly higher resolution than observations with the notion of a brief thick disk formation
their previous study. time. Instead the observed a-element trends may indicate
Finally, we can compare the various stellar populations that the formation time of the thick disk exceeds the Type Ia
with respect to the heavy element abundance trends. This is SN time scale, i.e., t Z 1 Gyr. To reconcile the nearly
somewhat ambitious, however, given the large uncertainties constant Mg/Fe trend formunder this scenario, however, one
of our Ba, Y, and Eu measurements. Nevertheless, the Ba may require ne tuning in the form of enhanced formation
and Y trends are in reasonable agreement for all of the of very massive Type II SN, an enhanced production factor
stellar populations in this metallicity range ; in general, they for Mg in metal-poor Type Ia SN, or increased Mg pro-
are all consistent with the solar composition. Europium, duction in more metal-rich Type II SN. Another possibility
however, presents a dierent story. In agreement with the altogether is that the a-element trends are the sole result of
a-element observations and the notion that Eu is formed Type II SN with an evolving IMF ; specically, the impor-
predominantly in Type II SN, the thick disk Eu/Fe ratios
are well in excess of the thin disk measurements and compa-
rable to the halo and bulge values.
6.2. Implications for the Formation of the T hick Disk
In the previous subsection, we drew comparisons
between the abundance patterns of the thick disk stars with
the halo, thin disk, and bulge stellar components. For the
majority of elementsthe a-elements, Co, V, Zn, Al, Mn,
Euthe thick disk stars show abundance patterns which
clearly distinguish them from the thin disk. These results
provide compelling conrmation that the chemical history
of the two stellar components is distinct as rst suggested by
the Mg and O analysis of Gratton et al. (2000) and Fuhr-
mann (1998). In turn, our analysis implies the thick disk is a
physically distinct stellar component from the thin disk
with its own specic formation history.
Gratton et al. (2000) and Fuhrmann (1998) have further
argued that the discrete separation of the O/Fe and Mg/Fe
ratios between the two stellar populations implies a signi-
FIG. 23.[Ba/Eu] ratios vs. [Fe/H] metallicity for the four stars with
cant time delay between the formation of the thick disk and measured Eu abundance. Overplotted in the gure are the ducial values
the onset of star formation in the thin disk. This assertion for the s- and r-process Ba/Eu ratios. The results suggest a mix of the two
appears well supported by our results. The larger a/Fe processes with the r process being dominant.
2542 PROCHASKA ET AL. Vol. 120
tance of the moderate mass Type II SN decreases as [Fe/H] icts and pursue the formation of both the thick and thin
increases. In this case, t > 1 Gyr, but it is difficult to disks.
understand the O/Fe trend form with [Fe/H]. In the following,
What does a comparison of the thick disk with the halo
we will adopt the hypothesis that the O, Si, and Ca trends chemical abundances reveal ? It is informative that with the
imply t Z 1 Gyr but warn that further investigation into exception of Co and V (and maybe Zn) the thick disk abun-
form
the nucleosynthetic history of the thick disk is crucial. dances for all of the elements match or approach the abun-
To date, several scenarios have been put forward to dances of halo stars with [Fe/H] B [1.3. This implies that
explain the formation of the thick disk. In particular, these the metal-poor thick disk stars formed from similar
include dissipative collapse models (Larson 1976 ; Jones & material as that for the most metal-rich halo stars. This
Wyse 1983), dynamical heating (Noguchi 1998), and merger naturally follows from the notion that the halo stars formed
scenarios (Quinn et al. 1993 ; Huang & Carlberg 1997 ; Sell- rst, i.e., before the majority of gas had dissipated to form a
wood et al. 1998). Our observations place important con- disk (thin or thick). Subsequently, those stars which formed
straints on these models. Consider rst the formation of the rst in the disk exhibit abundances similar to the metal-rich
thick disk during the dissipative collapse of a gaseous halo. halo stars while the later disk stars show higher metallicity
In order to maintain the chemical (and kinematic) distinc- and evolved abundance patterns. This explanation does not
tion between the thick and thin disks, the thick disk must account for the discordant Co and V trends. Perhaps these
form prior to the thin disk, in situ from a gaseous disk with abundances are the result of a specic type of nucleo-
a velocity dispersion consistent with the thick disk stars synthesis event (e.g., supernova) whose time scale exceeds
(p B 40 km s ~1). As suggested by Jones & Wyse (1983), the formation of the halo. Alternatively, the overproduction
W may be the natural evolution of a gaseous halo which
this of these elements may require specic physical conditions
has undergone some initial dissipation. The dissipational (i.e., surface density, binary fraction, metallicity) which did
time scale, however, is signicantly shorter than 1 Gyr, i.e., not exist in the halo. Therefore, by identifying the mecha-
t > 1 Gyr in the dissipational collapse scenario. One nisms responsible for the Co and V enhancements, one
form invoke additional energy sources such as primordial
could might gain further insight into the time-sequence of the
magnetic elds (Jedamzik et al. 1999) or an enhanced super- formation of the Galactic halo and thick disk.
novae rate and demand that they maintain the 40 km s~1 The similarities of the thick disk and bulge chemical
velocity dispersion of the gas disk for 1 Gyr while allowing abundance patterns are striking. With the exception of the
the dissipational collapse and formation of molecular O/Fe (which can be disregarded) and possibly the Si/Fe
clouds and stars, but this is unlikely. On the grounds that ratios, there is no signicant dierence in the abundance pat-
the formation time of the thick disk might well exceed 1 terns of any element. The obvious interpretation of this
Gyr, we contend it is unlikely that the thick disk formed in agreement is that the two stellar components formed at
situ as envisioned in an ELS-like scenario. Therefore, the essentially the same time and from the same gas reservoir. It
thick disk must have been preceded by a stellar thin disk is also interesting to note that the presence of thick disk in
which was heatedeither gradually over the course of B1 external galaxies has been linked to systems with signicant
Gyr or suddenlyto the current thick disk. The clumpy bulges (Burstein 1979). The most straightforward explana-
star-forming region model envisioned by Noguchi (1998) tion is that a single merger event sparked the formation of
provides a gradual heating mechanism (t B 1 Gyr) for both stellar components, although this is not the only possi-
the formation of the thick disk. In this form model, 109 M bility. The two components may have formed together
clumps form during the dissipational collapse of the Galaxy _
through a lengthy dynamical process as described by
which initiate star formation, spiral to the galaxy center via Noguchi (1998). It is also worth noting that the dissi-
dynamical friction, scatter the stars in the initial thin disk to pational collapse model of Jones & Wyse (1983) also pre-
form a thick disk, and eventually merge to form the bulge. dicts a close association between the metal-poor bulge stars
The model does not describe the formation of the current and the thick disk. Nevertheless, the claim that the two
thin disk ; presumably it will form after the clumps have components shared a similar gas reservoir and formation
merged through the accretion of a new reservoir of gas, but epoch is largely independent of the exact physical processes
this may require some ne tuning to insure that the involved in their formation. We eagerly look forward to
resulting thin disk is chemically distinct from the thick disk. future analyses of the bulge (McWilliam & Rich 2000), in
This complication aside, the model accounts for the major- particular the metal-poor tail. If these observations conrm
ity of our observations. Finally, consider merging scenarios the current picture (in particular the Co and V
where the initial thin disk is heated via the accretion of one enhancements), the two stellar components will be irrefut-
or more satellites. Unfortunately, it is difficult to assess this ably wed and all viable Galactic formation scenarios will
model as the various numerical studies conict on the need to account for them simultaneously.
eects of a merger event (e.g., heating efficiency, Quinn et al.
1993 ; Huang & Carlberg 1997 ; Sellwood et al. 1998). None- 6.3. Implications for Nucleosynthesis
theless, this scenario should naturally allow for the forma- Irrespective of comparisons with other stellar popu-
tion of the initial thin disk over the course of 1 Gyr and the lations, the chemical abundances of the thick disk stars
merger event(s) would erase all trace of the initial thin disk provide unique constraints on the processes of nucleo-
providing the discrepancy between the resulting thick disk synthesis in the early universe. Consider rst the a-elements
and the future thin disk. Similar to the clumpy region which exhibit an overabundance relative to Fe for all of the
model, however, the formation of the thin disk after the thick disk stars ; the enhancements indicate that the gas
merger event has not been considered in these numerical from which these stars formed was primarily enriched by
simulations. Furthermore, they disagree on the robustness Type II SN. Furthermore, by comparing the various abun-
of the thick disk to future accretion events. Hopefully, dance trends of the a-elements one gains insight into the
future numerical simulations will resolve the current con- yields from dierent mass and/or metallicity Type II SN. As
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2543
noted in the previous subsection, the most striking compari- however, we do not observed enhanced Ni or Cu, yet we
son is that the Mg and Ti abundances show no dependence note that the Zn enhancement (also seen in the extremely
on [Fe/H] metallicity whereas the Ca, Si, and O ratios metal-poor stars ; Johnson 1999) could be related to the Co
appear to decrease steadily with increasing [Fe/H] for the pattern. It will be important to focus on the Co/Zn ratio in
thick disk stars. While this statement is largely dependent future studies. The vanadium overabundance is not predict-
on the observations of the two metal-poor stars in our ed by any current theoretical model of nucleosynthesis.
sample and therefore suers from small number statistics, According to Woosley & Weaver (1995), the dominant
we speculate on the implications. Dierences between the vanadium isotope 51V is primarily produced during the
trends of Mg and Si or Ca may be expected on theoretical a-rich freeze out, yet the leading theories predict only a
grounds because these elements are believed to be synthe- fraction of the observed solar vanadium abundance
sized in dierent mass Type II SN (Woosley & Weaver (Woosley 1986). In fact, to the best of our knowledge this
1995). In particular, the SN models suggest that Mg is pro- marks the rst signicant evidence for enhanced V in any
duced primarily in the highest mass Type II SN whereas the stellar population. The V overabundance poses an excellent
production of Si and Ca is dominated by moderate mass challenge for nucleosynthesis research on an element which
SN. The observed trends, however, are difficult to explain until now has been largely ignored. Finally, note that the
under the assumption of a constant IMF. If anything, one unusual Sc/Fe trends agree surprisingly well with the rea-
might expect signicant evolution in Mg as it arises in the nalysis of Nissen et al. (2000) by Prochaska & McWilliam
more massive Type II SN. Therefore, either nucleosynthesis (2000). The trend is very peculiar with the [Fe/H] B [1
occurred in Type II SN with an evolving IMF or Type Ia stars exhibiting essentially no Sc/Fe enhancement, most
SN are playing a signicant role in the observed abundance stars at [Fe/H] B [0.5 enhanced above solar, and the
trends. If the latter explanation is adopted (as assumed in nearly solar metallicity stars show solar Sc/Fe. As with Co
the previous subsection), then it becomes a challenge to and V, scandium is believed to be produced primarily
explain the constancy of the Mg/Fe ratio. Perhaps at through the a-rich freeze out process. Given the very
[Fe/H] B [0.5 there is an increase in the number of very unusual behavior of Sc/Fe, a combination of metallicity
massive Type II SN or maybe the rst generation of Type Ia dependent yields and various supernovae appears likely.
SN overproduces Mg. It is also difficult to reconcile the Perhaps the overabundance at [Fe/H] B [0.5 is due to the
dierence in the Mg and O abundance trends as both ele- onset of Type Ia SN whose relative Sc/Fe production
ments are expected to be produced primarily in massive decreases with increasing metallicity. Altogether, the Co, V,
Type II SN and therefore should track one another reason- Sc, and Zn enhancements point toward a further investiga-
ably well. It is possible the O trend is the result of a system- tion of the a-rich freeze out nucleosynthesis. In turn, one
atic error (i.e., non-LTE eects) in measuring O from the may learn about the IMF of the thick disk stars. Finally, we
O I triplet at j B 7775 A . If this is not the case, the obser- note that the Mn/Fe-values for the thick disk are generally
vations pose a meaningful challenge to the current models lower than the thin disk values. We consider this evidence
of Type II SN. Finally, while the similarity between the Mg for the overproduction of Mn in Type Ia SN (Gratton
and Ti trends may be a coincidence, it does suggest the 1989). This assertion is further supported by the fact that
possibility that the two elements are produced in similar the stars with the highest Mn/Fe ratios show the lowest
nucleosynthetic sites. At the very least, while searching for Si/Fe, Ca/Fe, and O/Fe-values and argue against a metal-
the processes which yield enhanced Ti one may wish to rst licity dependent yield for Mn.
focus on those mechanisms which produce Mg. We conclude our discussion of nucleosynthetic implica-
The behavior of the light elements in the thick disk stars tions with a few comments on the heavy element results.
diers considerably. The Na/Fe ratios follow the a-element Figure 23 presents the [Ba/Eu] ratio against [Fe/H] for the
enhancementsalbeit with a more mild overabundance four stars with wavelength coverage of the Eu II j6645 line.
in a fashion possibly consistent with some Type II SN con- Overplotted on the gure are two lines indicating the du-
tribution. The enhancement is small, however, and could be cial values for the solar s- and r-process values of the Ba/Eu
the result of a systematic error in the Na analysis. In con- ratio. The thick disk ratios lie in between the two ducial
trast, aluminum is signicantly enhanced over the solar values suggesting that both processes are important in the
ratio which clearly points to a signicant production of Al nucleosynthesis of the heavy elements, although the r
in massive stars. Like Mg and Ti, Al exhibits no obvious process appears dominant. As noted in the previous section,
trend with [Fe/H] metallicity in our sample. Again, we can this is consistent with the very large age believed for the
speculate that Al is formed in similar sites as Mg and Ti. thick disk stars. Lastly, the Ba/Y ratio is approximately
With respect to nucleosynthesis, the abundance trends of solar in line with the halo composition (Gratton & Sneden
several of the iron-peak elements may oer the most star- 1994).
tling results. In particular, the thick disk stars show
enhanced Co and V where if anything one predicts a de- 6.4. Comparisons with the Damped L ya Systems
ciency for these odd nuclei. Furthermore, the Zn This nal section describes the implications of our results
enhancementwhile mildappears to contradict the pre- on interpretations of the abundance patterns of the damped
vious empirical belief that Zn/Fe is solar at all metallicities. Lya systems. The damped Lya systems are absorption-line
The Co enhancement brings to mind the overabundance systems identied along the sight lines to distant quasars
observed for extremely metal-poor stars (McWilliam et al. and have neutral hydrogen column densities N(H I)
1995b). Cobalt is believed to be synthesized during the 2 ] 1020 cm~2. Owing to their very large N(H I)-values,
a-rich freeze out fueled by a neutrino-driven wind (Woosley these systems dominate the neutral hydrogen content
& Homan 1992). An enhancement of Co might then be of the universe at all epochs (Wolfe et al. 1995 ; Rao
expected to be correlated with other elements slightly more & Turnshek 2000). Furthermore, the very large column
massive than Fe. Similar to the extremely metal-poor stars, densities imply overdensities suggestive (do/o ? 100) of pro-
2544 PROCHASKA ET AL. Vol. 120
togalaxies. Finally, the comoving baryonic mass density in winds of Type II SN (Homan et al. 1996). Under this
gas inferred from the damped systems at z \ 23 coincides scenario, Zn may be expected to behave like an a-element
with the current comoving baryonic mass density in stars and therefore exhibit an enhancement relative to Fe in
today (Wolfe et al. 1995 ; Storrie-Lombardi & Wolfe 2000). metal-poor stars.
For these reasons the damped Lya systems are widely Our observations suggest such an enhancement ([Zn/
believed to be the progenitors of modern galaxies. The Fe] D ]0.1), albeit at a level below the majority of
majority of studies on the chemical abundances of these a-elements andmore importantlybelow the Zn/Fe
protogalaxies has focused on damped systems with absorp- enhancement observed in the damped Lya systems. The
tion redshift z [ 1.7 (i.e., t [ 10.5 Gyr for a universe 15 thick disk abundances, however, do reect a more compli-
abs
Gyr old), where the Lya absorption line is observable with cated origin for Zn than the one readily adopted in quasar
optical spectrographs. Therefore, with the exception of Mn absorption-line studies. Furthermore, as noted in 5.4.6, the
(whose transitions lie at large rest wavelength and are most Zn/Fe ratio is quite sensitive to the eective temperature
easily observed in lower z systems), our discussion will adopted for the stellar atmosphere. In fact, we may have
abs
focus on these very old systems. overestimated T in the majority of stars ( 3.4), such that
A comparison of the damped Lya systems with the eff
the true S[Zn/Fe]T-value in the thick disk stars would be
Galactic thick disk is motivated by the following consider- closer to ]0.15 dex. Even this enhancement would not fully
ations : (1) The damped systems are believed to be the pro- account for the [Zn/Fe] B ]0.4 enhancement observed for
genitors of modern galaxies like the Milky Way. (2) The the damped Lya systems (Prochaska & Wolfe 1999), but it
thick disk is believed to have formed at an epoch consistent does emphasize that a dust explanation alone is not accu-
with z of the damped systems. (3) The Galactic thick disk rate for the damped Lya observations. Furthermore, note
abs are surprisingly consistent with a model intro-
kinematics that the typical [Fe/H] of the thick disk stars is signicantly
duced by Prochaska & Wolfe (1997a) to explain the higher than that of the damped Lya systems ([Fe/
absorption-line proles of the damped Lya systems. And (4) H] B [1.5). Recent studies of the extremely metal-poor
the metal-rich damped systems contain enough baryons at stars indicate a [Zn/Fe] enhancement of B]0.2 to ]0.3
the thick disk metallicity to account for the stellar mass of dex at [Fe/H] \ [2.5 (Johnson 1999). An extensive survey
the Galactic thick disk (Wolfe & Prochaska 1998). While for Zn in stars with [Fe/H] B [1.5 is clearly well moti-
these similarities are present, we should note that the thick vated.
disk component may not be a generic component of disk The prospect of nucleosynthetically enhanced Zn/Fe may
galaxies and that the majority of damped Lya systems actually be a welcome sight to the interpretations of the
exhibit signicantly lower metallicity than the Galactic damped Lya abundance patterns. The current controversy
thick disk. Therefore, the two systems may not have a regarding the abundance patterns is the following (see Pro-
one-to-one correspondence. These points not withstanding, chaska & Wolfe 1999 for a more detailed discussion). Figure
our observations of Zn in the thick disk have immediate 24 plots the abundance patterns for a compilation of
impact on the damped Lya abundance studies. For the damped Lya systems (Lu et al. 1996 ; Prochaska & Wolfe
damped Lya systems, Zn currently plays the most pivotal 1999) with a range of metallicity.11 When a stellar abun-
role in interpreting abundance patterns in the early uni- dance researcher examines the plot, s/he sees a classic
verse. The key point is that measurements of the damped example of halo abundances with enhanced a-elements,
Lya systems are based on observations of Fe, Si, Zn, Cr, Ni, decient Mn and unpeculiar Ni, Cr, and Al.12 The over-
etc., in the gas phase (analogous to abundance measure- abundance of Zn relative to Fe would be puzzling yet the
ments made of the Galactic ISM ; e.g., Savage & Sembach overall implication of Type II SN enriched gas would be
1996), where elements like Fe, Ni, and Cr can be signi- clear. In fact, one might go so far as to claim that the gure
cantly depleted onto dust grains. Concerns over the eects depicts a natural evolutionary sequence in the abundances
of dust depletion in the damped systems are well motivated from the damped systems to the thick disk stars. In con-
by Pettini et al. (1994) who demonstrated an over- trast, if an expert in the interstellar medium studies the
abundance of Zn/Cr relative to solar of B0.4 dex. This gure, he or she would identify a warm, halo gas dust-
enhancement suggests dust depletion because Zn, unlike Cr, depletion pattern to explain the enhanced Si, Zn, and S.
is not heavily depleted onto dust grains in the ISM. The This interpretation is consistent with the Ni, Cr, and Al
implication, therefore, is that in order to assess even the abundances, too, but fails to account for the Mn trend or
[Fe/H] metallicity of a damped Lya system one must rst the two systems with enhanced Ti as neither under-
account for the depletion of Fe onto dust grains. An alter- abundant Mn nor enhanced Ti is observed in dust depleted
nate approachthe one typically implemented in damped gas in the ISM, SMC, or LMC (Savage and Sembach 1996 ;
Lya researchis to utilize Zn as a surrogate for Fe because : Welty et al. 1997 ; Welty et al. 1999). The difficulty in inter-
(1) Zn is not expected to be signicantly depleted onto dust preting the damped Lya abundance patterns, therefore, has
grains, and (2) Zn was found to track Fe at essentially all been that two explanations exist which account for the
metallicities in stars (Sneden et al. 1991). It is on the second majority of the observations but not all. Because the two
point that our observations play a meaningful role. interpretations are degenerate in the observed elements, it
In the majority of studies on the damped Lya systems, may even be possible to allow for both Type II SN
researchers have assumed that [Zn/Fe] B 0 irrespective of
[Fe/H]. While empirically this assertion has the support of
stellar abundance analysis (Sneden et al. 1991), it is difficult 11 The Cr and Ni observations at the lowest metallicities in the damped
(if not impossible) to motivate theoretically. Zinc is an iron- systems are at the detection limit and are likely to have been biased to
larger [Ni/Fe]- and [Cr/Fe]-values.
peak element, but it is not expected to be synthesized in a 12 Note that eld stars with metallicities similar to the damped Lya
similar fashion to Fe. The leading theory of Zn nucleo- systems do not show signicantly enhanced Al/Fe (Shetron 1996 ; Gratton
synthesis contends that Zn forms in the neutrino-driven & Sneden 1988).
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2545
FIG. 24.A comparison of the abundance patterns for our sample of thick disk star with those for a sample of damped Lya systems taken from Lu et al.
(1996), Lu et al. (1997), and Prochaska & Wolfe (1999).
a-enrichment and a dust depletion pattern (although the liam 1997 ; Trager et al. 2000). If the damped Lya systems do
S/Fe ratio may set an upper limit to the combination ; Pro- not contain the gas which gave rise to these systems, then
chaska & Wolfe 1999). where is that gas ? We reemphasize that the damped systems
The implications of a Type II SN enrichment pattern for dominate the neutral hydrogen gas content of the universe
the damped Lya abundances are reasonable. These systems at all observed epochs (z \ 04.5), i.e., that gas which is
have low metallicity and presumably young ages and one most likely responsible for star formation. Second, if the gas
would expect them to have been enriched primarily by Type in the damped systems has been signicantly enriched by
II SN. The difficulty remains, however, in explaining the Type Ia SN, then the stars which polluted these systems
large Zn/Fe enhancements. On the other hand, if one must have formed B1 Gyr prior to the observed epoch. For
adopts the ISM perspective then the underlying nucleo- a damped system at z B 3, this implies a formation epoch
abs the big bang ! For these reasons
synthetic pattern must more closely resemble a Type Ia SN which quickly approaches
yield (Vladilo 1998). The implications for this scenario are it is difficult to accept a pure dust depletion explanation for
troubling. First, we know that a signicant number of Zn/Fe.
stellar systems exhibit enhanced a-elements (e.g., McWil- Now consider that Zn/Fe may be enhanced in metal-
2546 PROCHASKA ET AL. Vol. 120
poor stars and therefore presumably in the metal-poor less, the results suggest that [S/Zn] is not a reliable indica-
damped Lya systems. In this case, a scenario with both dust tor of nucleosynthesis. In fact, our observations may even
depletion and Type II SN enhancement is valid, if not help to explain the puzzling subsolar S/Zn ratios observed
favored. While one must introduce dust to account for some by Centurion et al. (2000) in a few damped systems.
of the observed Zn/Fe overabundance, there could still be
an underlying Type II SN pattern. The exact level of
a-enhancement would depend on the dust depletion pattern 6.5. Concluding Summary
one assumes, particularly for Si. Dust-corrected enhance- We have presented a detailed chemical abundance
ments of [Si/Fe] B ]0.2 dex are viable and even enhance- analysis of 10 kinematically selected thick disk stars with
ments of ]0.3 dex are conceivable. Unfortunately, there metallicity ranging from [1.2 to [0.4 dex. The majority of
will always be uncertainties in interpreting Si owing to the [X/Fe] ratios for the elements studied exhibit signicant
exact depletion pattern one adopts. Oxygen would be an ([0.1 dex) departures from the solar abundance. These
ideal prospect for removing this degeneracy if not for the include (1) overabundances for all of the a-elements except
difficulties in measuring O in the damped Lya systems S, (2) enhancements above solar for the light elements Na
(Prochaska 2000). and Al, (3) overabundances of the iron-peak elements Sc,
As a nal point regarding the damped Lya systems, con- Co, Zn, and V and decient Mn, and (4) signicantly
sider the S/Zn ratio observed for the thick disk stars. In enhanced Eu. For S, Ni, Cr, Cu, Ba, and Y we nd essen-
several recent studies on the damped systems, researchers tially solar abundances relative to Fe.
have suggested that the S/Zn ratio may provide the best We compared our results with abundance studies of the
indication of nucleosynthesis in these systems (Centurion et halo, bulge, and thin disk taken from the literature. These
al. 2000). The argument follows from the fact that neither S comparisons reveal that in the majority of cases the thick
nor Zn are signicantly aected by dust depletion and S is disk stars exhibit X/Fe-values distinct from the thin disk. In
expected to be a Type II SN tracer while Zn is a surrogate particular, O, Ca, Mg, Ti, Eu, Al, V, and Co all show
for Fe. In Figure 25, we plot the [S/Zn] measurements for enhancements in the thick disk well in excess of those found
the nine thick disk stars with a sulfur measurement. Recall in the majority of thin disk stars with comparable [Fe/H]
that the sulfur abundances are very sensitive to the assumed metallicity. We argue, therefore, that the thick disk has a
T values and are based on weak, single line measure- distinct chemical history from the thin disk. In general, the
eff
ments, i.e., the v(S)-values are very uncertain. This uncer- thick disk abundances either match or at low [Fe/H]
tainty is somewhat tempered by the fact that the Zn I lines tend toward the patterns observed for halo stars with
have a reasonably signicant dependence on T such that [Fe/H] [ [1.5. Most impressive, however, is the excellent
S/Zn is less aected by errors in T than S/Fe.effExamining agreement between the thick disk and metal-poor bulge star
eff
Figure 25 one notes that the [S/Zn]-values suggest no sig- abundance patterns. With the possible exception of Si, the
nicant enhancement in these stars ([S/Zn] \ 0.03 ^ 0.08) chemical histories of these two stellar populations appear
even though the a-elements Ca, Ti, Mg, O, and Si are all remarkably similar.
enhanced relative to Fe. This result is due to the fact that we Unlike previous elemental abundance studies of the thick
nd only a mild S/Fe enhancement in the thick disk stars (in disk (Gratton et al. 2000 ; Fuhrmann 1998), we nd tentative
contradiction with the observations of Francois 1988) and evidence that several a-elements (O, Si, Ca) show trends of
the observed Zn/Fe enhancement. Of course, it would be declining enhancements with increasing [Fe/H]. If these
most informative to repeat this analysis at metallicities trends are interpreted as evidence for the onset of Type Ia
more representative of the damped Lya systems. Nonethe- SN (we caution that this is not the only viable explanation),
then the thick disk formed over the course of Z1 Gyr.
Because this conclusion has signicant impact on formation
scenarios for the thick disk, future eorts to conrm the
a-element trends and investigate other explanations are
essential.
We discussed the implications of our observations on the
leading formation models of the thick disk. The dierences
between thick and thin disk abundance patterns imply that
the thick disk formed prior to the thin disk and that there
exists a signicant delay between its formation and that of
the thin disk. We argued that if the formation time of the
thick disk is D1 Gyr then most dissipational collapse
models are ruled out. On the other hand, the clumpy region
model of Noguchi (1998) and the merger scenarios (e.g.,
Quinn et al. 1993) are viable scenarios, although none of
these studies considered the formation of the thin disk in a
self-consistent manner.
The excellent agreement between the thick disk and the
metal-poor bulge abundance patterns provides a strong
argument that the two populations shared a common gas
reservoir and formation epoch. A connection between the
FIG. 25.Solar-corrected [S/Zn] ratios vs. [Fe/H] metallicity for the two populations has previously been suggested through
nine thick disk stars with S measurements. All of the data points scatter imaging studies of edge-on spirals (Burstein 1979) and can
around the solar meteoritic ratio. be explained by several of the formation scenarios (Jones &
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2547
Wyse 1983 ; Noguchi 1998). Our observations tighten the thick disk stars at greater radial distances to examine radial
association between the two populations, hinting they are dependencies of the abundance trends. To minimize poten-
intimate from the onset of their formation. tial systematic errors of comparing our results against
In the nal sections, we discussed implications of our abundance studies taken from the literature, we will observe
analysis on nucleosynthesis in the early universe and inter- a signicant sample of halo stars with [Fe/H] B [1, thin
pretations of the damped Lya abundance patterns. In terms disk stars with [Fe/H] B [0.5, and metal-poor bulge stars.
of nucleosynthesis, the conicting a-element trends of Mg, The results presented in this paper demonstrate that these
O, Si, Ti, and Ca present a challenge to our current under- stellar populations oer remarkable insight into the Galac-
standing of nucleosynthetic production in Type Ia and tic formation history.
Type II SN. We also noted that the Sc, V, Co, and Zn
overabundances hint at an overproduction through an We wish to acknowledge J. Fulbright, J. Johnson, M.
enhanced a-rich freeze-out process. For the damped Lya Albada, B. Weiner, and R. Bernstein for insightful dis-
systems, we described the implications of enhanced Zn/Fe cussion and comments. We thank T. Barlow for providing
in the thick disk stars. In particular, this enhancement the HIRES reduction package. We acknowledge the very
allows for interpretations of the damped Lya abundance helpful Keck support sta for their eorts in performing
patterns which include a combination of dust depletion and these observations. We thank C. Sneden for providing and
Type II SN enrichment patterns. Finally, we noted that the updating the package MOOG and J. Omeara and D. Tytler
S/Zn ratio is not a good nucleosynthetic indicator. for taking observations of G92-19. Finally, we thank R.
Our future observational eorts will include obtaining Kurucz for help in using his codes and databases. J. X. P.
additional spectra of kinematically selected thick disk stars. acknowledges support from a Carnegie postdoctoral fellow-
In particular, we will further investigate the trends of O, Si, ship. B. C. acknowledges support from an NSF grant AST
and Ca by increasing the sample of thick disk stars with 96-19381. A. M. acknowledges support from NSF grant
[Fe/H] B [1 and by adjusting our setups to include the AST 96-18623. A. M. W. was partially supported by NASA
O I [j6300] forbidden transition. We also intend to observe grant NAG W-2119 and NSF grant AST 86-94 20443.
APPENDIX A
STELLAR GRAVITY
In those cases where there are Hipparcos parallax measurements, we calculated the stellar gravity according to the
following equations :
log (g/g ) \ log (L /L ) ] log (M/M ) ] 4 log (T /T ) (A1)
_ _ _ _
by assuming the following mass relation for these main sequence stars,
log (M/M ) \ 0.480.105 M , (A2)
_ bol
and
log (L /L ) \ [(M [ 4.64)/2.5 . (A3)
_ bol
To estimate the bolometric correction (BC 4 M [ M ), we interpolated between the tabulated values of Alonso et al.
V bol
(1995). Given the uncertainties in the parallax and BC-values, the error in log (g/g ) from this approach is B0.1 dex.
_
APPENDIX B
HYPERFINE SPLITTING
The coupling of the nuclear angular momentum I with the angular momentum of the outer electrons J is known as the
hyperne interaction. These interactions lead to the splitting of absorption lines by typically 110 mA and can have the eect
of desaturating strong features. As such, it is important to account for hyperne splitting in order to accurately measure
elemental abundances.
Analogous to spin-orbit coupling, one can dene the vector sum of the angular momenta as F which has quantum numbers
ranging from o I[J o to o I]J o . The number of energy states is 2J ] 1 when I J and 2I ] 1 for J I. The allowed
transitions must satisfy the Wigner-Eckart theorem, namely *F \ 0, ^1 with 0 ] 0 forbidden. While one could calculate the
wavelength splittings from rst principles, it is usually more accurate to determine them empirically. The energy splitting for a
given energy level is typically represented as
1 (3/4)K(K ] 1) [ J(J ] 1)I(I ] 1)
*E \ AK ] B , (B1)
2 2I(2I [ 1)(2J [ 1)
where
K 4 F(F ] 1) [ I(I ] 1) [ J(J ] 1) . (B2)
2548 PROCHASKA ET AL. Vol. 120
By determining the A and B constants empirically from very high resolution laboratory experiments for both the upper and
lower energy levels, one can calculate the wavelength shifts of hyperne lines to the precision required for stellar spectroscopy.
Kurucz (1999) has compiled the most accurate A and B constants from the literature and calculated the hyperne lines for
nearly all of the absorption lines identied in our spectra. With the exception of Ba (for which we adopt the values provided
by McWilliam 1998), we take the values presented by Kurucz.
To calculate the relative line strength g of each hyperne line, one can adopt the Russell-Saunders terms (Condon &
Shortly 1935, 238). In the case where J \ J@ the following equations apply :
(2F ] 1)[F(F ] 1) [ I(I ] 1) ] J(J ] 1)]2
g\ [F \ F@] (B3)
4F(F ] 1)
(F [ I ] J)(F ] I [ J)(I ] J ] F ] 1)(I ] J ] 1 [ F)
\ [F [ F@] (B4)
4F
(F [ I ] J ] 1)(F ] I [ J ] 1)(I ] J ] F ] 2)(I ] J [ F)
\ [F \ F@] . (B5)
4(F ] 1)
If J is greater than J@, then
(2F ] 1)(F ] J [ I)(F ] I [ J ] 1)(F ] I ] J ] 1)(I ] J [ F)
g\ [F \ F@] (B6)
4F(F ] 1)
(F ] I [ J ] 1)(I ] J [ F)(F ] I [ J ] 2)(I ] J [ 1 [ F)
\ [F \ F@] (B7)
4(F ] 1)
(F [ I ] J [ 1)(F ] J [ I)(F ] I ] J ] 1)(I ] J ] F)
\ [F [ F@] . (B8)
4F
Finally, if J is less than J@,
(2F ] 1)(F ] J@ [ I)(L ] I [ J@ ] 1)(F ] I ] J@ ] 1)(I ] J@ [ F)
g\ [F \ F@] (B9)
4F(F ] 1)
(F@ ] I [ J@ ] 1)(I ] J@ [ F@)(F@ ] I [ J@ ] 2)(I ] J@ [ 1 [ F@)
\ [F \ F@] (B10)
4(F@ ] 1)
(F@ [ I ] J@ [ 1)(F@ ] J@ [ I)(F@ ] I ] J@ ] 1)(I ] J@ ] F@)
\ [F [ F@] . (B11)
4F@
We then normalize the relative line strengths to give the total log gfvalue listed in Table 3. We present the actual values
that we have adopted in Table 19 in the event that the Kurucz table is updated after publication.
REFERENCES
Abia, C., & Rebolo, R. 1989, ApJ, 347, 186 Blackwell, D. E., Ibbetson, P. A., Petford, A. D., Shallis, M. J., & Simmons,
Allen, C., Poveda, A., & Schuster, W. J. 1991, A&A, 244, 280 G. J. 1980, MNRAS, 191, 445
Alonso, A., Arribas, S., & Martinez-Roger, C. 1995, A&A, 297, 197 Blackwell, D. E., Ibbetson, P. A., Petford, A. D., & Simmons, G. J. 1982c,
. 1996a, A&AS, 117, 227 MNRAS, 201, 595
. 1996b, A&A, 313, 873 Blackwell, D. E., Ibbetson, P. A., Petford, A. D., & Willis, R. B. 1976,
Bachall, J. N., & Soneira, R. M. 1984, ApJS, 55, 67 MNRAS, 177, 219
Bard, A., & Kock, M.1994, A&A, 282, 1014 Blackwell, D. E., Lynas-Gray, A. E., & Smith, G. 1995, A&A, 296, 217
Bard, A., Kock, A., & Kock, M. 1991, A&A, 248, 315 Blackwell, D. E., Menon, S., & P. A., Petford, A. D. 1983, MNRAS, 204,
Baumuller, D., Butler, K., & Gehren, T. 1998, A&A, 338, 637 883
Baumuller, D., & Gehren, T. 1997, A&A, 325, 1088 . 1984, MNRAS, 207, 533
Becker, U., Zimmermann, P., & Holweger, H. 1980, Geochim. Cosmochim. Blackwell, D. E., Menon, P. A., Petford, A. D., & Shallis, M. J. 1982d,
Acta, 44, 2145 MNRAS, 201, 611
Beers, T. C., & Sommer-Larsen, J. 1995, ApJS, 96, 175 Blackwell, D. E., Petford, A. D., Shallis, M. J., & Leggett, S. 1982a,
Beveridge, C. R., & Sneden, C. 1994, AJ, 108, 285 MNRAS, 199, 21
Biemont, E., Baudoux, M., Kurucz, R. L., Ansbacher, W., & Pinnington, Blackwell, D. E., Smith, G., & Lynas-Gray, A. E. 1995, A&A, 303, 575
E. H. 1991, A&A, 249, 539 Booth, A. J., Blackwell, D. E., Petford, A. D., & Shallis, M. J. 1984a,
Biemont, E., & Godefroid, M. 1980, A&A, 84, 361 MNRAS, 208, 147
Biemont, E., Karner, C., Meyer, G., Trager, F., & zu Putlitz, G. 1982, A&A, Booth, A. J., Blackwell, D. E., & Shallis, M. J. 1984b, MNRAS, 209, 77
107, 166 Brown, J. A., & Wallerstein, G. 1992, AJ, 104, 1818
Bizzarri, A., Huber, M. C. E., Noels, A., Grevesse, N., Bereson, S. D., Burstein, D. 1979, ApJ, 234, 829
Tsekeris, P., & Lawler, J. E. 1993, A&A, 273, 707 Butler, K., & Zeaippen, C. J. 1991, J. Physique IV, Colloque, C1, 141
Blackwell, D. E., Booth, A. J., Haddock, D. J., Petford, A. D., & Leggett, Buurman, E. P, Donszelmann, A., Hansen, J. E., & Snoek, C. 1986, A&A,
S. K. 1986c, MNRAS, 220, 549 164, 224
Blackwell, D. E., Booth, A. J., Menon, S. L. R., & Petford, A. D. 1986a, Cardon, B. L., Smith, P. L., Scalo, J. M., Testerman, L., & Whaling, W.
MNRAS, 220, 289 1982, ApJ, 260, 395
. 1986b, MNRAS, 220, 303 Carney, B. W., Laird, J. B., Latham, D. W., & Aguilar, L. A. 1996, AJ, 112,
Blackwell, D. E., Ibbetson, P. A., Petford, A. D., & Shallis, M. J. 1979a, 668
MNRAS, 186, 633 Carney, B. W., Latham, D. W., & Laird, J. B. 1989, AJ, 97, 423
. 1979b, MNRAS, 186, 657 Carney, B. W., Latham, D. W., Laird, J. B., & Aguilar, L. A. 1994, AJ, 107,
. 1982b, MNRAS, 199, 43 2240
No. 5, 2000 THICK DISK STELLAR ABUNDANCES 2549
Carney, B. W., Naumov, S., Laird, J. B., Latham, D. W., Andersen, J., & Martin, G. A., Fuhr, J. R., & Weise, W. L. 1988, J. Phys. Chem. Ref. D, Vol.
Nordstrom, B. 2000, in preparation 17, No. 3
Carretta, E., & Gratton, R. G. 1997, A&AS, 121, 95 Martin, J. C., & Morrison, H. L. 1998, AJ, 116, 1724
Centurion, M., Bonifacio, P., Molaro, P., & Vladilo, G. 2000, ApJ, 536, 540 May, M., Richter, J., & Wichelmann, J. 1974, A&AS, 18, 405
Chen, B. 1997, ApJ, 491, 181 McWilliam, A. 1997, ARA&A, 35, 503
Chen, Y. Q., Nissen, P. E., Zhao, G., Zhang, H. W., & Benoni, T. 2000, . 1998, AJ, 115, 1640
A&AS, 141, 491 McWilliam, A., Geisler, D., & Rich, R. M. 1992, PASP, 104, 1193
Chiba, M., & Beers, T. C. 2000, AJ, 119, 2843 McWilliam, A., Preston, G. W., Sneden, C., & Searle, L. 1995a, AJ, 109,
Condon, E. U., & Shortly, G. H. 1935, The Theory of Atomic Spectra 2736
(Cambridge : Cambridge Univ. Press), 238 . 1995b, AJ, 109, 2757
Edvardsson, B., Anderson, J., Gutasfsson, B., Lambert, D. L., Nissen, P. E., McWilliam, A., & Rich, R. M. 1994, ApJS, 91, 749
& Tompkin, J. 1993, A&A, 275, 101 . 2000, in preparation
Eggen, O. J., Lynden-Bell, D., & Sandage, A. 1962, ApJ, 136, 748 Meylan, T., Furenlid, I., Wiggs, M. S., & Kurucz, R. L. 1993, ApJS, 85, 163
ESA. 1997, The Hipparcos and Tycho Catalogues (ESA SP-1200) Moity, J. 1983, A&AS, 52, 37
(Noordwijk : ESA) Moore, C. E., Minnaert, G. J., & Houtgast, J. 1966, The Solar Spectrum,
Francois, P. 1988, A&A, 195, 226 NBS Monograph 61, (2d ed. ; Washington : GPO)
Fry, A. M., & Carney, B. W. 1997, AJ, 113, 1073 Morrison, H. L., Boroson, T. A., & Harding, P. 1994, AJ, 108, 1191
Fry, A. M., Morrison, H. L., Harding, P., & Boroson, T. A. 1999, AJ, 118, Morrison, H. L., Flynn, C., & Freeman, K. C. 1990, AJ, 100, 1191
1209 Morrison, H. L., Miller, E. D., Harding, P., Stinebring, D. R., & Boroson,
Fuhr, J. R., Martin, G. A., & Wiese, W. L. 1988, J. Phys. Chem. Ref. Data, T. A. 1997, AJ, 113, 2061
17, Suppl. 4 Nissen, P. E., Chen, Y. Q., Schuster, W. J., & Zhao, G. 2000, A&A, 353, 722
Fuhrmann, K. 1998, A&A, 338, 161 Noguchi, M. 1998, Nature, 392, 253
Fuhrmann, K., Axer, M., & Gehren, T. 1995, A&A, 301, 492 Norris, J. 1986, ApJS, 61, 667
Garz, T. 1973, A&A, 26, 471 Norris, J., Bessell, M. S., & Pickles, A. J. 1985, ApJS, 58, 463
Gilmore, G., & Reid, N. 1983, MNRAS, 202, 1025 OBrian, T. R., Wicklie, M. E., Lawler, J. E., Whaling, W., & Brault, J. W.
Gilmore, G., & Wyse, R. F. G. 1985, AJ, 90, 2015 1991, J. Opt. Soc. Am. B, 8, 1185
. 1987, in The Galaxy, ed. G. Gilmore & B. Carswell (Dordrecht : OConnell, D. J. K. 1958, in Stellar Populations, ed. D. J. K. OConnell
Reidel), 247 (New York : Interscience), 57
Gilmore, G., Wyse, R. F. G., & Jones, J. B. 1995, AJ, 109, 1095 Ojha, D. Bienayme, O., Robin, A. C., Creze, M., & Mohan, V. 1996, A&A,
Gilmore, G., Wyse, R. F. G., Kuijken, K. 1989, ARA&A, 27, 555 311, 456
Gratton, R. G. 1989, A&A, 208, 171 Pettini, M., Smith, L. J., Hunstead, R. W., & King, D. L. 1994, ApJ, 426, 79
Gratton, R. G., Carretta, E., Matteucii, F., Sneden, C. 1996, in ASP Con- Pilachowski, C. A., Sneden, C., & Kraft, R. P. 1996, AJ, 111, 1689
ference Series 92, Formation of the Galactic Halo. . .Inside and Out, ed. Prochaska, J. X. 2000, Proc. SPIEE, 4005, 65
H. Morrison & A. Sarajedini (San Francisco : ASP), 307 Prochaska, J. X., & McWilliam, A. 2000, ApJ, 537, L57
. 2000, A&A, 358, 671 Prochaska, J. X., & Wolfe, A. M. 1997, ApJ, 474, 140
Gratton, R. G., & Sneden, C. 1988, A&A, 204, 193 . 1999, ApJS, 121, 369
. 1991, A&A, 241, 501 Quinn, P. J., Hernquist, L., & Fullagar, D. P. 1993, ApJ, 403, 74
. 1994, A&A, 287, 927 Rao, S. M., & Turnshek, D. A. 2000, ApJS, 130, in press
Grevesse, N., Blackwell, D. E., & Petford, A. D. 1989, A&A, 208, 157 Ryan, S., & Lambert, D. L. 1995, AJ, 109, 2068
Grevesse, N., Noels, A., & Sauval, A. J. 1996, in ASP Conf. Ser. 99, Cosmic Savage, B. D., & Sembach, K. R. 1996, ARA&A, 34, 279
Abundances, ed. S. Holt & G. Sonneborn (San Francisco : ASP), 117 Savanov, I. S., Huovelin, J., & Tuominen, I. 1990, A&AS, 86, 531
Grevesse, N., & Sauval, A. J. 1999, A&A, 347, 348 Schnabel, R., Kock, M., & Holweger, H. 1999, A&A, 342, 610
Griffin, R. F. 1968, A Photometric Atlas of the Spectrum of Arcturus Sellwood, J. A., Nelson, R. W., & Tremaine, S. 1998, ApJ, 506, 590
(Cambridge : Univ. Printing House) Shetron, M. D. 1996, AJ, 112, 1517
Gurtovenko, E. A., & Kostik, R. I. 1981, A&AS, 46, 239 Smith, G., & Raggett, D. St. J. 1981, J. Phys. B : Atom. Molec. Phys., 14,
Hannaford, P., & Lowe, R. M. 1983, Opt. Eng., 22, 532 4015
Hannaford, P., Lowe, R. M., Grevesse, N., Biemont, E., & Whaling, W. Sneden, C., & Crocker, D. A. 1988, ApJ, 335, 406
1982, ApJ, 261, 736 Sneden, C., Gratton, R. G., & Crocker, D. A. 1991, A&A, 246, 354
Hesser, J. E., Harris, W. E., VandenBerg, D. A., Allwright, J. W. B., Shott, Steen, M. 1985, A&AS, 59, 403
P., & Stetson, P. B. 1987, PASP, 99, 739 Storrie-Lombardi, L. J., & Wolfe, A. M. 2000, ApJ, in press
Heise, C., & Kock, M. 1990, A&A, 230, 244 Timmes, F. X., Woosley, S. E., & Weaver, T. A. 1995, ApJS, 98, 617
Helfer, H. L., Wallerstein, G., & Greenstein, J. L. 1959, ApJ, 129, 700 Tomkin, J., Lambert, D. L., & Balachandran, S. 1985, ApJ, 290, 289
Homan, R. D. et al. 1996, ApJ, 460, 478 Tomkin, J., Lemke, M., Lambert, D. L., & Sneden, C. 1992, AJ, 104, 1568
Holweger, H., Kock, M., & Bard, A. 1995, A&A, 296, 233 Trager, S. C., Faber, S. M., Worther, G., & Gonzalez, J. J. 2000, AJ, 119,
Holweger, H., & Muller, E. A. 1974, Solar Phys., 39, 19 1645
Huang, S., & Carlberg, R. G. 1997, ApJ, 480, 503 Twarog, B. A., & Anthony-Twarog, B. J. 1996, AJ, 111, 220
Jedamzik, K., Katalinic, V., & Olinto, A. 1999, Phys. Rev. Lett., 85, 700 van der Kruit, P. C., & Searle, L. 1981, A&A, 95, 116
Johnson, J. 1999, Ph.D. thesis, Univ. California at Santa Cruz Vladilo, G. 1998, ApJ, 493, 583
Jones, B. J. T., & Wyse, R. F. G. 1983, A&A, 120, 165 Vogt, S. S., et al. 1994, SPIE, 2198, 362
Kappeler, F., Zhao, W. R., Beer, H., & Ratzel, U. 1990, ApJ, 355, 348 Wallerstein, G. 1962, ApJS, 6, 407
Karner, C., Meyer, G., Trager, F., zu Putlitz, G. 1981, in Upper Main Welty, D. E., Frisch, P. C., Sonneborn, G., & York, D. G. 1999, ApJ, 512,
Sequence Chemically Peculiar Stars (Liege : Inst. dastrophys.), 223 636
Kock, M., & Richter, J. 1968, Z. Astrophys., 69, 180 Welty, D. E., Lauroesch, J. T., Blades, J. C., Hobbs, L. M., & York, D. G.
Kroll, S., & Kock, M. 1987, A&AS, 67, 225 1997, ApJ, 489, 672
Kurucz, R. L., Furenlid, I., Brault, J., & Testerman, L. 1984, Solar Flux Whaling, W., Hannaford, P., Lowe, R. M., Biemont, & Grevesse, N. 1985,
Atlas from 296 to 1300 nm (Sunspot, N. M. : Natl. Solar Obs.) A&A, 153, 109
Kurucz, R. L. 1988, Trans. IAU 20B, ed. D. McNally (Dordrecht : Kluwer), Wheeler, J. C., Sneden, C., & Truran, J. W., Jr. 1989, ARA&A, 27, 279
168 Wicklie, M. E., & Lawler, J. E. 1997, ApJS, 110, 163
Lambert, D. L., & Luck, R. E. 1978, MNRAS, 183, 79 Wiese, W. L., & Martin, G. A. 1980, Wavelengths and Transition Probabil-
Larson, R. B. 1976, MNRAS, 176, 31 ities for Atoms and Atomic Ions, NSRDS-NBS 68 (Washington : NBS)
Lawler, J. E., & Dakin, J. T. 1989, J. Opt. Soc. Am., B6, 1457 Wolfe, A. M., Lanzetta, K. M., Foltz, C. B., & Chaee, F. H. 1995, ApJ, 454,
Layden, A. C. 1995a, AJ, 110, 2288 698
. 1995b, AJ, 110, 2312 Wolfe, A. M., & Prochaska, J. X. 1998, ApJ, 494, 15L
Lu, L., Sargent, W. L. W., & Barlow, T. A. 1997, in ASP Conf. Ser. 156, Woosley, S. E. 1986, in Nucleosynthesis and Chemical Evolution, ed.
Highly Redshifted Radio Lines, ed. C. Carilli, S. Radford, & K. Menten B. Hauck, A. Maeder, & G. Meyent (Genevea : Geneva Obs.), 1
(San Francisco : ASP), 132 Woosley, S. E., & Homan, R. D. 1992, ApJ, 395, 202
Lu, L., Sargent, W. L. W., Barlow, T. A., Churchill, C. W., & Vogt, S. 1996, Woosley, S. E., & Weaver, T. A. 1995, ApJS, 101, 181
ApJS, 107, 475 Wyse, R. F. G., & Gilmore, G. 1995, AJ, 110, 2771
Luck, R. E., & Bond, H. E. 1985, ApJ, 292, 559 Zhao, G., & Magain, P. 1990, A&A, 238, 242
Majewski, S. R. 1992, ApJS, 78, 87 . 1991, A&A, 244, 425
Majewski, S. R. 1993, ARA&A, 31, 575