Chemical Composition of Faint (I~21 mag) Microlensed Bulge Dwarf OGLE-2007-BLG-514S

Abstract

We present a high-resolution spectrum of a microlensed G dwarf in the Galactic bulge with spectroscopic temperature T_eff = 5600 +/- 180 K. This I~21 mag star was magnified by a factor ranging from 1160 to 1300 at the time of observation. Its high metallicity ([Fe/H] = 0.33 +/- 0.15) places this star at the upper end of the bulge giant metallicity distribution. Using a K-S test, we find a 1.6% probability that the published microlensed bulge dwarfs share an underlying distribution with bulge giants, properly accounting for a radial bulge metallicity gradient. We obtain abundance measurements for 15 elements and perform a rigorous error analysis that includes covariances between parameters. This star, like bulge giants with the same metallicity, shows no alpha enhancement. It confirms the chemical abundance trends observed in previously analyzed bulge dwarfs. At supersolar metallicities, we observe a discrepancy between bulge giant and bulge dwarf Na abundances. Comment: 13 pages, 8 figures, 5 tables, submitted to ApJ

Full text

Page 1

arXiv:0910.1358v1 [astro-ph.SR] 7 Oct 2009
Draft version October 7, 2009
Preprint typeset using LATEX style emulateapj v. 04/21/05
CHEMICAL COMPOSITION OF FAINT (I ∼ 21 MAG) MICROLENSED BULGE DWARF OGLE-2007-BLG-514S
Courtney R. Epstein1, Jennifer A. Johnson1, Subo Dong1,2,3, Andrzej Udalski4, Andrew Gould1,2, and George
Becker5,6
Draft version October 7, 2009
ABSTRACT
We present a high-resolution spectrum of a microlensed G dwarf in the Galactic bulge with spectro-
scopic temperature Teff= 5600±180 K. This I ∼ 21 mag star was magnified by a factor ranging from
1160 to 1300 at the time of observation. Its high metallicity ([Fe/H] = 0.33±0.15) places this star at
the upper end of the bulge giant metallicity distribution. Using a K-S test, we find a 1.6% probabil-
ity that the published microlensed bulge dwarfs share an underlying distribution with bulge giants,
properly accounting for a radial bulge metallicity gradient. We obtain abundance measurements for
15 elements and perform a rigorous error analysis that includes covariances between parameters. This
star, like bulge giants with the same metallicity, shows no alpha enhancement. It confirms the chem-
ical abundance trends observed in previously analyzed bulge dwarfs. At supersolar metallicities, we
observe a discrepancy between bulge giant and bulge dwarf Na abundances.
Subject headings:
1. INTRODUCTION
Baade’s discovery (1946, 1958) that the central regions
of the Galaxy, like other Sb spirals, contained population
II stars sparked intense interest in the Galactic bulge.
Because it is the closest galactic spheroid to the Sun, we
can study the Galactic bulge in unique detail and use it
to better understand galaxy formation and evolution.
Proposed senarios for building bulges include mergers
and secular dynamical evolution (e.g. Kormendy & Ken-
nicutt 2004). Discriminating between bulge formation
theories hinges on age determinations of bulge stars, a
process complicated by crowding, contamination by fore-
ground stars, reddening, and dispersion in both metallic-
ity and distance along the line of sight. Comparing main-
sequence photometry of bulge globular clusters with halo
globular clusters demonstrated that they are coeval (Or-
tolani et al. 1995). Examining the color-magnitude di-
agrams of both bulge fields and clusters shows that the
bulge clusters are representative of the general bulge pop-
ulation, the majority of which are old stars (∼ 10 Gyr)
(e.g. Ortolani et al. 1995; Zoccali et al. 2003). Although
Feltzing & Gilmore (2000) agree that the bulge is mostly
old, their interpretation of the photometry permits a
young metal-rich population of stars.
Evidence for a short star formation phase comes from
the chemical composition of bulge stars. The α-elements
(O, Mg, Si, Ca, and Ti) are created in the evolution
and explosion of massive stars as core collapse super-
novae (SNcc) and almost immediately recycled in the in-
1Department
sity, 140
epstein,jaj,gould@astronomy.ohio-state.edu
2Microlensing Follow-Up Network (µFUN)
3Institute for Advanced Study, School of Natural Sciences,
Einstein Drive, Princeton, NJ, 08540; dong@ias.edu
4Warsaw University Observatory, Warszawa, Poland; udal-
ski@astrouw.edu.pl
5Carnegie Observatories, 813 Santa Barbara Street, Pasadena,
CA 91101, USA
6Fellow,Kavli Institute for Cosmology and Institute of
Astronomy,Madingley Road,
gdb@ast.cam.ac.uk
of Astronomy,
Ave.,
OhioState
43210,
Univer-
USA;W.18thColumbus,OH
Cambridge CB30HA,UK;
terstellar medium (ISM) (e.g. Woosley & Weaver 1995).
Because of their longer progenitor lifespan, Type Ia su-
pernovae (SNIa) do not enrich the ISM with substan-
tial amounts of iron until 1-1.5 Gyr after star formation
begins (Matteucci et al. 2009). The delayed SNIa iron
contribution dilutes the SNcc-enhanced α/Fe ratio until
it reaches solar levels (Tinsley 1979). Since bulge giants
maintain an elevated α/Fe ratio to high metallicity, the
bulge must have evolved faster than the solar neighbor-
hood and undergone an intense burst of star formation
early in the history of the galaxy (e.g. Matteucci & Bro-
cato 1990; McWilliam & Rich 1994).
Within this framework, the α-elements exhibit differ-
ent behavior.The explosive α-elements Si, Ca, and
Ti appear to track each other (Fulbright et al. 2007).
Surprisingly, the hydrostatic α-elements O and Mg do
not, even though both are made in SNcc progenitors
and released in SN explosions.
mains enhanced in the bulge to supersolar metallicities
(McWilliam & Rich 1994; Fulbright et al. 2007). The
bulge O abundance closely tracks the abundance trends
in disk stars and begins declining around [Fe/H]≈ −0.3
dex (e.g. Mel´ endez et al. 2008). McWilliam et al. (2008)
reconciles these trends by invoking Wolf-Rayet winds in
metal-rich, massive stars to produce lower mass SNcc
progenitors, which Woosley & Weaver (1995) predict
yield less oxygen.
Neutron-capture elements also have the potential to
act as clocks. AGB stars are associated with the pro-
duction of s-process elements, while supernovae may
be responsible for r-process elements.
tion about neutron-capture elements is available in the
bulge since the absorption lines are concentrated in the
crowded blue portion of the spectrum.
McWilliam & Rich (2004) observed an enhanced Eu
abundance and tentatively suggest a behavior similar to
Ca. McWilliam & Rich (1994) find a subsolar s-process
abundance, reminiscent of old, metal-poor halo stars, in-
dicating that the r-process dominates in the bulge.
Light elements, like Na, contain important information
In particular, Mg re-
Little informa-
Nevertheless,

Page 2

2
about the chemical evolution of the bulge. Na produc-
tion is sensitive to the neutron excess (Arnett 1971) and
its yield is therefore dependent on the metallicity of its
SNcc progenitor. Although typically interpreted as SNcc
ejectum, Na can also be synthesized in envelope burning
of AGB stars. Lecureur et al. (2007) report a flat Na
abundance trend in the bulge below solar metallicity, af-
ter which it increases abruptly with a dispersion at high-
metallicities exceeding measurement error. Lecureur et
al. determine that the elevated Na abundance cannot be
explained by the mild internal mixing seen in their sam-
ple of giants and therefore must reflect the composition
of the ISM at its formation.
Our knowledge of the bulge’s chemical evolution comes
from spectroscopic studies of K and M giants because
they are typically the only stars accessible to high-
resolution observations due to the far distance and the
high degree of interstellar extinction toward the bulge.
However, large surveys designed to identify and follow up
microlensing events, such as the Optical Gravitional Lens
Experiment7(OGLE) and the Microlensing Observations
in Astrophysics (MOA) collaboration8, have made possi-
ble the high-resolution spectroscopic study of otherwise
unobservable bulge dwarfs. Lennon et al. (1996) per-
formed the first spectroscopic abundance determination
for a microlensed bulge dwarf, 96-BLG-3, detected by the
MACHO collaboration. Based on low-resolution data,
they estimated a metallicity of between 0.3 and 0.6 dex.
Minniti et al. (1998) obtained a high-resolution spectrum
of a dwarf using Keck. Cavallo et al. (2003) presented a
preliminary abundance analysis of six microlensed bulge
stars, including the Minniti et al. star and two other
dwarfs.
There are seven published abundances analyses of mi-
crolensed bulge dwarfs based on high-resolution data:
OGLE-2006-BLG-265S (Johnson et al. 2007), OGLE-
2007-BLG-349S (Cohen et al. 2008), MOA-2006-BLG-
099S (Johnson et al. 2008), OGLE-2008-BLG-209S
(Bensby et al. 2009a), MOA-2008-BLG-310S and MOA-
2008-BLG-311S (Cohen et al. 2009), and OGLE-2009-
BLG-076S (Bensby et al. 2009b). Although OGLE-2008-
BLG-209S is actually a sub-giant, we will refer to these
seven stars collectively as bulge dwarfs. Additionally,
Bensby et al. (2009c) announced that high-resolution
spectra have been obtained for six more bulge dwarfs,
including a reanalysis of the Cavallo et al. dwarfs, whose
detailed abundances will appear in Bensby et al.
prep.) and Johnson et al. (in prep.).
Studying dwarfs offers several advantages. Dwarfs are
subject to different systematic effects than the giants.
Mixing on the red giant branch (RGB) can destroy some
elements and dredge up others, altering a star’s surface
composition. Some elements that are difficult to see in
giants, like S and Zn, can be measured in dwarfs since
their hotter temperatures strengthen these atomic lines
while weakening CN. Additionally, we can compare the
metallicity distribution function (MDF) for dwarfs and
giants.
The first-observed high-magnification bulge dwarf with
a high-resolution spectrum had a surprisingly high metal-
licity of 0.56 dex (Johnson et al. 2007).
(in
Subsequent
7http://ogle.astrouw.edu.pl/ogle3/ews/ews.html
8www.massey.ac.nz/∼iabond/alert/alert.html
Fig. 1.— Distribution in Galactic latitude and longitude of
dwarf stars in the bulge that have published high-resolution spec-
tral analyses (solid circles). A star marks the location of OGLE-
2007-BLG-514S. The heliocentric radial velocity for each star is
indicated by an arrow where up corresponds to a positive velocity
and the vectors are scaled so that one degree in length is 70 km
s−1. The crosses indicate the fields from Zoccali et al. (2008), in-
cluding Baade’s window (BW), whose MDF is discussed in §5. The
dashed circles mark the locus where the projected distance from
the Galactic Center (GC) equals that of a Zoccali et al. field.
observations revealed more metal-rich dwarfs indicating
that high metallicities are common in bulge dwarfs. By
contrast, the bulge giant MDF peaks just below solar
metallicity (e.g. Fulbright et al. 2007; Cunha & Smith
2006; Rich & Origlia 2005; Rich et al. 2007; Lecureur
et al. 2007; Zoccali et al. 2003, 2008).
(2008) suggest that high-metallicity stars experience a
sufficiently high mass-loss rate to cause them to peel off
the RGB before undergoing a He flash and become He
white dwarfs (WDs). Zoccali et al. (2008) counter that
this scenario would produce a decline in the RGB lumi-
nosity function, which is not observed in the Zoccali et al.
(2003) data. OGLE-2008-BLG-209S, the first bulge sub-
giant with a high-resolution spectrum, was found to have
a subsolar metallicity, consistent with the giants also lo-
cated in Baade’s window (Bensby et al. 2009a). The de-
tection of metal-poor dwarf OGLE-2009-BLG-076S op-
poses the idea of a strong selection effect against metal-
poor microlensed stars.
Here we present both the high-magnification bulge
dwarf MDF and a detailed chemical abundance analy-
sis of the eighth bulge dwarf, OGLE-2007-BLG-514S.
Cohen et al.
2. OBSERVATIONS AND DATA REDUCTION
OGLE-2007-BLG-514S is located toward the Galactic
bulge (J2000 R.A. = 17h58m3.s09, Dec. = −27◦31′05.′′7;
l = 2.62, b = −1.63). Figure 1 gives the position of this
star and the other seven bulge dwarfs. Of these stars,
this dwarf lies the closest to the plane of the Galaxy.
OGLE-2007-BLG-514S’s dereddened Johnson-Cousins
color and magnitude relative to the red clump (see §3.3)
are consistent with it being a main sequence turnoff star
at the distance of the red clump. Alerted to this sit-
uation, Michael Rauch and George Becker took three
high-resolution spectra of OGLE-2007-BLG-514S with

Page 3

Chemical Compostion of OGLE-2007-BLG-514S3
Fig. 2.— Spectrum of OGLE-2007-BLG-514S in the region sur-
rounding the O I triplet where the S/N∼ 50. The spectrum is
shown shifted to the rest frame.
the Magellan Inamori Kyocera Echelle (MIKE) double
spectrograph mounted on the Clay telescope at Las Cam-
panas (Bernstein et al. 2003).
giving a resolution of 25,000. One 20-minute and two
30-minute exposures were taken, in addition to the usual
Th-Ar and flatfield calibration frames. We use only the
red side of the spectrograph, covering the wavelength
range 9300 − 4900˚ A. The three exposures were com-
bined to make the final spectrum for analysis. These
data were then reduced using the procedure described in
Johnson et al. (2008).
At its peak, OGLE-2007-BLG-514S was magnified by
a factor of ∼1300 as it crossed a cusp in the caustic,
brightening this I ∼ 21 mag star to approximately 13.3
mag. This is probably the faintest star for which a high-
resolution spectrum has been obtained.
effects, in particular the magnification of the limb rela-
tive to the center occur during such crossings. We have
modeled the effect of differential limb magnification on
the spectrum, and find that in more extreme cases than
the actual observations discussed here, the change in the
derived temperature is < 40 K, in microturbulent veloc-
ity is < 0.08 km s−1, and in surface gravity is < 0.1
dex, all much smaller than the uncertainties in our at-
mospheric parameters (see §3.2) due to the S/N of the
spectrum, blending of lines, and other effects that appear
in all spectra, whether magnified or not (Johnson et al.
2009, in prep). Johnson et al. find that ignoring differ-
ential limb magnification when analyzing the spectrum
results in abundance errors < 0.05 dex.
We measure the per pixel signal-to-noise ratio (S/N) in
the continuum regions at three different wavelengths to
provide a sense how the S/N deteriorates due to redden-
ing. The spectrum has a S/N of ∼ 50 at 8000˚ A, ∼ 30 at
7000˚ A, and ∼ 20 at 6000˚ A. Bluer than that, the S/N
additionally suffers from reduced flux from the star due
to its relatively cool temperature and the strong metal-
absorption lines in this portion of the spectrum. Figure
2 gives an example of the quality of the data in the red
portion of the spectrum.
The slit was 1′′wide,
Finite source
TABLE 1
Line Parameters and Equivalent Widths
IonWavelength
(˚ A)
E.P.
(eV)
loggfEWstar
(m˚ A)
EW⊙
(m˚ A)
Ba II
Ca I
Ca I
Ca I
Ca I
Ca I
Ca I
Ca I
Ca I
Ca I
...
6496.900
5260.387
5261.704
5512.980
5581.965
5601.277
5867.562
6122.217
6166.439
6169.042
...
0.60
2.52
2.52
2.93
2.52
2.53
2.93
1.89
2.52
2.52
...
-0.377
-1.780
-0.450
-0.560
-0.530
-0.240
-1.610
-0.360
-1.170
-0.840
...
syn
36.6
126.0
100.2
144.7
153.8
54.1
260.9
111.6
138.4
...
syn
31.5
101.3
89.5
95.5
119.2
21.5
191.1
69.8
92.4
...
Note. — Table 1 is published in its entirety in the
electronic edition of The Astrophysical Journal. A por-
tion is shown here for guidance regarding its form and
content.
For OGLE-2007-BLG-514S, we measure an average ra-
dial velocity of 186.9 ± 0.4 km s−1from ten unblended
iron lines with accurate laboratory wavelengths. At the
time of observation, the heliocentric correction amounted
to -28.08 km s−1. Adding this correction, the heliocentric
radial velocity of OGLE-2007-BLG-514Sis 158.8±0.4km
s−1. The diversity of heliocentric radial velocities for all
published bulge dwarfs in Figure 1 reflects the large ve-
locity dispersion of stars in the bulge (e.g. Howard et al.
2008).
3. ABUNDANCE ANALYSIS
The linelist used for this analysis consists of a combi-
nation of the lists from Johnson et al. (2008) and Cohen
et al. (2008). Table 1 lists the transition probabilities
and measured equivalent widths (EW) for lines in both
OGLE-2007-BLG-514S and the Sun. We measure EW
using SPECTRE9for each line individually (C. Sneden,
2007, private communication).
For elements with hyperfine splitting, namely Ba, Mn,
V, and Sc, we use TurboSpectrum (Alvarez & Plez 1998),
a 1-dimensional LTE code, to create a synthetic spectrum
for the region surrounding the desired line and adjust
the elemental abundance10to match the observed spec-
trum. The hyperfine splitting A-values and, if available,
B-values for the dominant isotope of vanadium, V-51,
come from Cochrane et al. (1998), Palmeri et al. (1995),
Lef` ebvre et al. (2002), and Childs et al. (1979). The
HFS constants for the other elements are taken from the
sources listed in Johnson et al. (2006).
3.1. Atmospheric Parameters
We interpolated model atmospheres over a range in
effective temperature, microturbulent velocity, surface
gravity, and metallicity using a grid of the ATLAS9 mod-
els11with new opacity distribution functions (Castelli &
9http://verdi.as.utexas.edu/spectre.html
10
Weadopt the
[X/Y] ≡ log10(NX/NY)⋆ − log10(NX/NY)⊙ and logǫ(X) ≡
log10(NX/NH) + 12.0 for elements X and Y.
11http://wwwuser.oat.ts.astro.it/castelli/grids.html
standardspectroscopicnotationthat

Page 4

4
Fig. 3.— Top: Absolute Fe I abundance verses excitation poten-
tial demonstrating excitation equilibrium. Bottom: Absolute Ca I
abundance verses reduced EW. Solid lines give the weighted-least-
squares fit to the abundances.
Kurucz 2003). TurboSpectrum is used to derive elemen-
tal abundances from the measured EWs for each model
atmosphere.
Adhering to standard practice, the effective tempera-
ture (Teff) is found by ensuring excitation equilibrium for
Fe I lines, pictorially represented by a zero slope in abun-
dance with respect to excitation potential (EP). Micro-
turbulent velocity (ξ) is determined by requiring a zero
slope in the diagram of Ca I abundance verses reduced
equivalent width (log(EW/λ)). Both of these relation-
ships are shown in Figure 3. The surface gravity (logg) is
determined by requiring ionization equilibrium, i.e. that
the average abundance of Fe I lines equals the average
abundance from Fe II lines to within 0.02 dex. We then
set the input model atmosphere metallicity to the [Fe
I/H] determined relative to our measured solar value and
iteratively adjust logg and [Fe/H] until the metallicities
match. To summarize our method, we determine the
atmospheric model parameters
mj= (Teff,ξ,logg,[Fe/H])
using the following observables oi:
o1: slope of log(Fe I) v. E.P.
o2: slope of log(Ca I) v. log(EW/λ).
o3: logǫ(Fe I) − logǫ(Fe II) for the model.
o4: [Fe I/H]model− [Fe I/H]input.
where mjand oiare both four dimensional vectors. The
best fit model m0
logg = 4.2 dex, and [Fe/H] = 0.33 dex.
jgives Teff= 5600 K, ξ = 1.5 km s−1,
3.2. Error Analysis
We present a rigorous error analysis that propagates
the uncertainties in the observables to uncertainties in
atmospheric parameters and elemental abundances. We
TABLE 2
Partial Derivatives of Observables with
Respect to Atmospheric Model Parameters
bij
∂o1
∂mj
∂o2
∂mj
∂o3
∂mj
∂o4
∂mj
∂oi
∂m1
−0.03
160
∂oi
∂m2
+0.02
0.3
∂oi
∂m3
+0.00
0.4
∂oi
∂m4
+0.00
0.2
+0.04
160
−0.06
0.3
−0.23
0.4
+0.07
0.2
+0.20
160
−0.04
0.3
−0.22
0.4
−0.04
0.2
+0.11
160
−0.10
0.3
−0.04
0.4
−0.17
0.2
begin by writing each observable as a linear combination
of deviations from the best fit model:
oi= o0
i+
4
?
j=1
bij(mj− m0
j)(1)
where bij = ∂oi/∂mj.
rameter at a time, we create a series of models with
∆m1 = ±160 K, ∆m2 = ±0.3 km s−1, ∆m3 = ±0.4
dex, and ∆m4= ±0.2 dex. These ∆mj are of order the
size of the model atmosphere grid since the grid is ap-
proximately linear on those scales. Adjusting a model
atmosphere by ∆mj will change each of the observables
by (oi− o0
i). These partial derivatives bij are given in
Table 2.
Since we have four observables, Equation (1) produces
a system of equations, which can be solved for the at-
mospheric model parameters. Inverting the matrix of bij
yields a matrix with elements cij.
Let σi be the error in the measurement oi. We treat
the errors σias independent. Then the uncertainties in
the model parameters are given by
Varying one atmospheric pa-
σ(mi) =
?
?
?
?
4
?
k=1
c2
ikσ2
k. (2)
The errors σ1and σ2are the uncertainty in the slope
of the linear least squares fit to logǫ(Fe I) v. E.P. and
logǫ(Ca I) v. log(EW/λ), respectively. Uncertainties in
EW and oscillator strengths produce a standard error of
the mean abundance of 0.03 dex for Fe I and 0.10 dex
for Fe II. We added in quadrature the standard error of
the mean for Fe I and Fe II average abundances to find
σ3. We take the standard error of the mean Fe I abun-
dance to be σ4. Substituting these values into Equation
(2) yields the uncertainties in atmospheric parameters:
σ(Teff) = 180 K, σ(ξ) = 0.5 km s−1, σ(logg) = 0.3 dex,
and σ([Fe/H]) = 0.15 dex. These uncertainties, along
with the values of the model atmosphere parameters, are
given in Table 3.
The next step is to find the uncertainty in the abun-
dance of some element X. As before, we can write our
measured quantity, in this case X, as
X = X0+
4
?
j=1
κj(mj− m0
j) = X0+
4
?
j=1
αj(oj− o0
j) (3)

Page 5

Chemical Compostion of OGLE-2007-BLG-514S5
TABLE 3
Stellar Parameters and Uncertainties
for OGLE-2007-BLG-514S
Teff
(K)
ξlogg
(dex)
[Fe/H]
(dex)(km s−1)
mj
5600
180
1.5
0.5
4.2
0.3
0.33
0.15σ(mj)
where κj= ∂X/∂mjand
αj≡
4
?
k=1
κkckj. (4)
We found κj by running the series of models with ∆mj
through TurboSpectrum or, for elements with hyper-
fine splitting, by creating a synthetic spectrum for each
model. Thus, the error in logǫ(X) is
σ(X) =
?
?
?
?σ2(X0) +
4
?
k=1
α2
kσ2
k
(5)
where σ(X0) is the standard error of the mean abundance
of element X at fixed stellar model parameters. For el-
ements with either one measured line or a synthesized
spectrum, we take the change in abundance required to
compensate for a generous offset in the continuum place-
ment as the error in measuring X0. Otherwise, we use
the rms in abundance for elements with many measured
lines.
Often, however, we are interested in the ratio of two
elements X and Y, written [X/Y]. The uncertainty in this
quantity is
σ([X/Y ]) =
?
?
?
?σ(X)2+ σ(Y )2− 2
4
?
k=1
αX
kαY
kσ2
k
(6)
The uncertainties calculated with Equations (5) and (6)
are given in Table 4.
3.3. Photometric Parameters
De-reddened colors and magnitudes of the microlensed
source are estimated using standard microlensing tech-
niques (Yoo et al. 2004). We assume that the redden-
ing to the microlensed source is the same as the redden-
ing to the red clump giants in the bulge. Stanek et al.
(2000), Sumi (2004), and Bennett et al. (2008) report
a de-reddened (V − I)0 for the bulge red clump in the
range 1.028-1.066. Taking (V − I)RC= 1.05, photomet-
ric observations from µFUN SMARTS CTIO 1.3m tele-
scope determined the dereddened Johnson-Cousins color
and magnitude of OGLE-2007-BLG-514S are (V −I)0=
0.70±0.05and I0= 19.2±0.1mag, respectively. The sim-
ilarity between OGLE-2007-BLG-514S’s color and the
Sun’s (V − I)⊙= 0.688 ± 0.014 (Holmberg et al. 2006)
implies that the temperatures of these stars must also
be close. Using the relationship between temperature,
color, and metallicity in Ram´ ırez & Mel´ endez (2005)
(updated from the color-temperature relation in Alonso
et al. 1996), we estimate a photometric temperature of
5760±200 K. This photometric temperature agrees with
the spectroscopic temperature within error.
3.4. Solar Spectrum
We performed an identical analysis using the same set
of lines, line parameters, and model atmosphere grid
as for OGLE-2007-BLG-514S on a high-resolution solar
spectrum also taken by the MIKE spectrograph of re-
flected light from Jupiter’s moon Ganymede (Bensby et
al. 2009a). We fixed the well-known model solar atmo-
sphere parameters at Teff= 5777 K, logg = 4.4 dex, and
[Fe/H] = 0 dex and adjusted ξ until excitation equilib-
rium was achieved at ξ = 0.88 km s−1. Table 4 presents
both the solar abundances from our measurements and
the literature values from Grevesse & Sauval (1998). All
final elemental abundances given in Table 4 were nor-
malized to the measured solar spectrum. This differen-
tial analysis is appropriate for bulge dwarfs since they
are similar type stars to the Sun and the same lines are
visible in both stars.
For the solar spectrum, we find logǫ(Ba)⊙= 2.3 dex
from the Ba II line at 6496.898˚ A, which is the only
Ba line we measured in the OGLE-2007-BLG-514S spec-
trum. Our measurement is higher than both the pho-
tospheric abundance of 2.13 ± 0.05 dex and meteoritic
abundance of 2.22 ± 0.02 dex from Grevesse & Sauval
(1998)12. The measurement error for Ba in our very high
S/N solar spectrum is negligible compared to possible
systematic effects. The disagreement in literature me-
teoritic and photospheric Ba abundances indicates that
resolving ∼ 0.09 dex differences is an ongoing issue for Ba
measurements. The photospheric absolute Ba abundance
depends sensitively on the specifics of the solar model
atmosphere and the line parameters. In particular, we
find that changing the model atmosphere parameter ξ by
+0.3 km s−1results in a +0.2 dex change in the Ba abun-
dance. The absolute Ba abundance is also influenced by
the van der Waals damping coefficient (Mashonkina &
Zhao 2006); although we used the most up-to-date val-
ues, they may not be absolutely correct. Fortunately,
uncertainties in the absolute abundance due to errors in
the model atmosphere and line parameters cancel to first
order when considering the stellar abundance relative to
the solar values found using the same technique. For this
reason, we use the metallicity normalized to a solar spec-
trum consistent with the stellar analysis if possible when
comparing OGLE-2007-BLG-514S with other stars.
4. BULGE MEMBERSHIP
From OGLE-2007-BLG-514S’s high heliocentric radial
velocity (158.8±0.4 km s−1), we infer that the source is
in the bulge rather than the disk. Disk stars at the same
galactic longitude have radial velocities
RVdisk∼220 km s−1R0sinl
η
∼ 80 km s−1
?1 kpc
η
?
where R0= 8 kpc and η is the distance from the Galactic
Center to OGLE-2007-BLG-514S. We limit η > 1 kpc to
exclude the bulge. Given the disk’s radial velocity dis-
persion of σ ∼ 35 km s−1(Dehnen & Binney 1998), the
probability is low that a disk star would achieve a radial
12We compare to Grevesse & Sauval (1998) since they also use
a 1-dimensional model atmosphere.
higher photospheric Ba abundance of 2.18 ± 0.09 dex (Asplund et
al. 2009).
Recent 3-D models yield a