**FULL TITLE**

ASP Conference Series, Vol. **VOLUME**, **YEAR OF PUBLICATION**

**NAMES OF EDITORS**

Planets in the Galactic Bulge: Results from the SWEEPS

Project

Kailash C. Sahu,1 Stefano Casertano,1 Jeff Valenti,1 Howard E. Bond,1

arXiv:0711.4059v1 [astro-ph] 26 Nov 2007

Thomas M. Brown,1 T. Ed Smith,1 Will Clarkson, 1 Dante Minniti,2

Manuela Zoccali,2 Mario Livio,1 Alvio Renzini,3 R. M. Rich,4 Nino
Panagia,1 Stephen Lubow,1 Timothy Brown,5 Nikolai Piskunov6

Abstract. The exoplanets discovered so far have been mostly around rela-
tively nearby and bright stars. As a result, the host stars are mostly (i) in the
Galactic disk, (ii) relatively massive, and (iii) relatively metal rich. The aim
of the SWEEPS project is to extend our knowledge to stars which (i) are in a
different part of the Galaxy, (ii) have lower masses, and (iii) have a large range
of metallicities. To achieve this goal, we used the Hubble Space Telescope to
monitor 180,000 F, G, K, and M dwarfs in the Galactic bulge continuously for 7
days in order to search for transiting planets. We discovered 16 candidate tran-
siting extrasolar planets with periods of 0.6 to 4.2 days, including a new class of
ultra-short period planets (USPPs) with P < 1.2 days. Radial-velocity obser-
vations of the two brightest candidates support their planetary nature. These
results suggest that planets are as abundant in the Galactic bulge as they are in
the solar neighborhood, and they are equally abundant around lower-mass stars
(within a factor ∼2). The results also suggest that planet frequency increases
with metallicity even for the stars in the Galactic bulge. All the USPP hosts are
low-mass stars, suggesting that either close-in planets around higher-mass stars
are irradiatively evaporated,or that planets are able to migrate to and survive
in close-in orbits only around such old and low-mass stars.

1. Introduction

More than 250 extrasolar planets have been discovered within the past few years,
most of them through the radial velocity (RV) measurements, and some through
transits and microlensing (see J. Schneider, Extrasolar Planet Encyclopaedia for
an up-to-date listing). These discoveries have led to tremendous advancements
in our knowledge of exoplanets. However, the exoplanet discoveries have so far
been mostly around relatively nearby and brighter stars: all of the RV detections
and a large number of transit detections are confined to host stars within about

1
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD. 21218, USA
2
Universidad Catolica de Chile, Av. Vicua Mackenna 4860, Santiago, Chile
3
INAF - Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5, 35122 Padova, Italy
4
University of California at Los Angeles, Los Angeles, CA 90095-1562, USA
5
Las Cumbres Observatory Global Telescope, Goleta, CA.
6
Department of Astronomy, Uppsala University, Box 515, 75120 Uppsala, Sweden

1
2

200 pc, a few of the transit detections have host stars as far away as 2 pc, and the
small number of the microlensing detections have host stars as far away as 6 kpc.
In addition, the RV detections have been mostly confined to relatively higher-
mass stars, although RV studies are now being extended to M dwarfs (Marcy,
2005; Butler et al. 2004; Bonfils et al. 2004). In contrast with the RV results,
an intensive transit search in the globular cluster 47 Tuc (Gilliland et al. 2000)
found no hot Jupiters around ∼34,000 cluster members, compared to the ∼17
expected from the frequency in the solar neighborhood. This discrepancy was
tentatively attributed to either environment or metallicity effects, since 47 Tuc
stars lie in a very dense stellar environment and are significantly metal-poor
compared to those in the solar neighborhood. Indeed, Fischer & Valenti (2003)
find that the frequency of planets in the RV sample rises rapidly with metallicity.
So, some of the key questions in the study of extrasolar planets, at present, are
the following: (i) Are planets equally abundant in other parts the Galaxy? (ii)
Are planets equally numerous around lower mass stars? (iii) Are hot Jupiters
common around a very different population? (iv) Does heavy element abundance
favor planet formation at other parts of the galaxy?
Our SWEEPS (Sagittarius Window Eclipsing Extrasolar Planet Search)
project was designed to provide answers to these key questions. At a distance
of ∼ 8.5 kpc, the Galactic bulge has a large concentration of stars whose metal-
licities range over −1.5 < [F e/H] < +0.5 (Rich and Origlia, 2005; Zoccali et al.
2003; Fulbright et al. 2005), and hence is an ideal choice for this study. We used
the HST and the Wide Field Camera of the Advanced Camera for Surveys to
monitor ∼180,000 F, G, K and M dwarfs with 18.5 < V < 26 in a dense stellar
field (3.3 × 3.3 arcmin) in the Galactic bulge for transits by orbiting Jovian-sized
planets.

2. Observations

The SWEEPS field lies in the Sagittarius-I Window of the Galactic bulge. We
monitored this field for planetary transits over a continuous 7-day interval during
February 22-29, 2004. At the distance of the Galactic bulge, an M0 dwarf
of 0.5 M⊙ has an apparent visual magnitude of ∼ 25.5, for which the HST
photometry is capable of detecting planetary transits. The observations include
254 exposures in F606W (wide V) and 265 exposures in F814W (I) for the
primary time series, all with an exposure time of 339 sec.

3. Analysis

The analysis technique employed is Difference Image Analysis (DIA; e.g., Alard
1999), similar to the procedure adapted by Gilliland et al. (1999, 2000) for the
analysis of 47 Tuc data. Combining together all the exposures taken in each
filter using the above procedure produces extremely deep, twice-oversampled V
(F606W) and I (F814W) images. Figure 1 shows the combined image of the
SWEEPS field in F606W and F814W filters.
The absolute photometry (Vegamag system) of the stars in the SWEEPS
field was determined from twice-oversampled co-added images of the entire
dataset in V and I. The DAOPHOT II PSF-fitting photometry package was
 3

Figure 1. V (F606W) and I (F814W) composite image of the SWEEPS

field, which has a size of 202 x 202 arcsec. There are 245,000 stars down to
V ∼ 30, out of which there are 180,000 stars brighter than V ∼ 26 around
which the observations are sensitive to detecting Jovian planets.

used for this purpose, with the photometric zero-points taken from the calibra-
tion work at STScI (Sirianni et al. 2005).
About 245,000 stars are detected in this combined image down to V ∼ 30,
of which 180,000 stars are brighter than V ∼ 26 around which our program
is sensitive to detecting Jovian planets. The color-magnitude diagram (CMD),
presented in Figure 2, shows two stellar components: a dominant population
of old stars with a main-sequence turnoff near V = 19.6 and well-populated
sub-giant and giant branches, and a less numerous, closer, younger and brighter
main sequence. We associate the old population with the Galactic bulge, and
the younger objects with the foreground Galactic disk (Kuijken & Rich 2002,
Zoccali et al. 2000). A modified version of the code developed by Kovacs et al.
(2003) was used for transit search.
4

Figure 2. The color-magnitude diagram (CMD) of the SWEEPS field as

derived from the deep, combined ACS images, with total integration times of
86,106 and 89,835 s in the V and I filters, respectively. The red (solid) line
shows a 10-Gyr old solar-metallicity isochrone which its the dominant bulge
population. The dashed blue (upper) line shows an unevolved main sequence,
representative of the foreground young disk population. An higher-metallicity
isochrone with [Fe/H]=0.5 is shown by the dashed magenta (lower) curve.
Large circles represent the 16 host stars with transiting planet candidates.

4. Results and Screening for false positives

A series of criteria as described by Sahu et al. (2006) was employed to eliminate

false positives, which include eliminating candidates with (i) a transit depth
implying a companion radius > 1.4RJ (ii) ellipsoidal light variations, (iii) sec-
ondary eclipses, (iv) different transit depths in V and I. We also eliminated
objects in which the photo-center of the transit signal is offset with respect to
that of the uneclipsed star. As an additional check, we doubled the period and
re-calculated the transit depths, and eliminated candidates with varying primary
 5

Figure 3. Five examples of observed transit light curves. The left panels
show the entire light curve, phased at the derived orbital period, and the right
panels show magnified views of the transit with 2σ error bars. The light curves
have been binned in phase to a bin width of 1/6th of the transit duration.
(Blue) squares are the V-band observations, and (red) circles are the I-band
observations. The black solid curves are the best-fitting model transit light
curves.

and secondary depths. This process led to the detection of 16 candidate planets.
The magnitudes of their host stars range from V=18.8 to 26.2, corresponding to
stellar masses of 1.24 to 0.44 M⊙ . Figure 3 shows a few typical examples of the
observed transit light curves.
In addition to the 16 exoplanet candidates, we have also detected 165 low-
mass eclipsing binaries, which we used to statistically estimate the possible con-
tribution from grazing eclipses and low-mass stars. Unlike most other ground-
based experiments, the HST experiment has (i) near-continuous time coverage,
(ii) observations in 2 different bands, (iii) same exposure times for all observa-
tions, and (iv) same psf-characteristics in all the images. Such a consistent set
6

2
SWEEPS−04
MP < 3.8 MJ
1

0
Heliocentric velocity shift (km s−1)

−1

Brown dwarf limit (M2 = 13 MJ)

−2
3
SWEEPS−11
2 MP = 9.7 MJ

−1

−2
Brown dwarf limit (M2 = 13 MJ)
−3
0 1 2 3 4
Julian date − 2453179 (days)

Figure 4. Radial-velocity measurements of SWEEPS-04 and SWEEPS-11

from VLT spectra. The measured radial velocities and their associated er-
rors are shown as black points. The red (short-dashed) curves show the RV
variation expected for a minimum-mass brown dwarf companion of 13 MJ .
For SWEEPS-11, there is a clear detection of RV variations, which imply a
planetary mass of 9.7 ±4.5 MJ . For SWEEPS-4, there is no detection, and at
the 95% and 99.9% confidence levels, we rule out companions more massive
than 3.8 MJ and 5.3 MJ . The zero-point uncertainty in phase due to the
extrapolation from the 2004 February HST transit observations to the 2004
June date of the RV observations is 0.45 days.

of observations makes it possible statistically estimate the contributions from

astrophysical false positives (such as grazing eclipses, low-mass stars, etc.) Fur-
thermore, the HST observations do not suffer from blending problems or “red
noise”, which makes the detections more robust. Taking into account all possi-
ble contributions from other sources of false positives, we estimate that > ∼ 45%
of the candidates are genuine planets (See Sahu et al, 2006 and 2008 for more
details).
 7

Figure 5. Orbital periods and host-star masses for extrasolar planets with
periods up to ∼12 days. Solid (red) circles are the 16 SWEEPS candidates,
(green) triangles are transiting planets around brighter stars as derived from
ground-based observations, and (red) crosses are for planets detected through
RV variability. The SWEEPS candidates extend the range of planetary orbital
periods down to 0.42 days. Very few planets have irradiances above 2 ×
106 W m−2 which corresponds to an equilibrium temperature of 2000 K. None
in the SWEEPS sample have equilibrium temperatures larger than 2000 K.
The absence of ultra-short-period planets around stars > 0.9M⊙ may be due
to irradiative evaporation.

5. Radial Velocity Followup Observations

Most of the host stars are too faint for radial velocity followup observations, but
SEEPS-4 and SWEEPS-11 were bright enough and lie in a relative uncrowded
region so that we could obtain radial velocity observations of them, using the
ESO 8m VLT and the FLAMES/UVES spectrograph. For SWEEPS-11, we
clearly detected RV variations, which indicate the mass to be 9.7 MJ . For
8

SWEEPS-4, for which the transit detection has a high S/N, the RV variations
were below the detection limit suggesting an upper limit to its mass of 3.8 MJ .
If only 50% of our candidates are genuine planets, the probability that both
selected objects would be planets is 25%. If 30% of the candidates are genuine
planets, this probability is only 10%. This gives us extra confidence that a
large fraction must be planets, and supports our estimate that > ∼ 45% of the
candidates are genuine planets. .

6. Results

After correcting for geometric transit probability and our detection efficiency,
our detections suggest that the frequency of planets in the SWEEPS field is
similar to that in the local neighborhood.
The frequency of planets around low-mass stars is also similar to the fre-
quency of planets around higher-mass stars, but given the small number statis-
tics, the uncertainty is large which can easily be a factor of 2 or 3.
The host stars of the detected planets preferentially lie towards higher-
metallicity isochrones. This is consistent with the fact that metallicity favors
planet frequency in the Galactic bulge, similar to the findings in the solar neigh-
borhood.
The USPPs with orbital periods shorter than 1 day occur only around stars
less massive than 0.88 M⊙ , and which have preferentially higher-metallicity. This
suggests that planets orbiting very close to more massive stars might be evap-
oratively destroyed, or that planets can migrate to close-in orbits and survive
there only around such old and low-mass stars.
Acknowledgments. We would like to thank Ron Gilliland for his help on
several stages of the project.

References

Alard, C. 1999, A& A, 343, 10

Bonfils, X. et.al. 2005, A& A., 443, L15
Butler, R.P. et.al. 2004; ApJ. 617, 580
Fischer, D.A. & Valenti, J.A. 2005, ApJ, 622, 1102
Fulbright, J.P., McWilliam, A., & Rich, R.M. 2006, 636, 821
Gilliland, R. L. et al. 2000, ApJ, 545, L47
Gilliland, R. L., Nugent, P. E., Phillips, M. M. 1999, ApJ, 521, 30
Gilliland, R. L. 2004, ACS Instrument Science Report, 2004-01 (Baltimore: STScI)
Kovacs, G., Zucker, S., & Mazeh, T., 2002, 391, 369
Kuijken, K. & Rich, R. M. 2002, Astrophys. J. 124, 2054
Marcy, G. 2005, Prog. in Th. Phys. Suppl. 158, 24
Rich, R.M., & Origlia, L. ApJ. 2005, 634, 1293
Sahu, K.C. et al. 2006, Nature, 443, 1038
Sahu, K.C. et al. 2008, in prep.
Schneider, J., Exoplanet Encyclopaedia (http://www.obspm.fr/encycl/encycl.html)
Sirianni, M. et al. 2005, PASP, 117, 1049
Zoccali, M. et al. 2000, ApJ, 530, 418
Zoccali M. et al. 2003, Astron. Astrophys. 399, 931