Imaging of the Stellar Population of IC10 with Laser Guide Star Adaptive Optics and the Hubble Space Telescope

0.95 MB·English

by  William D. Vacca

ApJ in press; submittedAugust17,2006: accepted January19,2007
PreprinttypesetusingLATEXstyleemulateapjv.10/09/06
IMAGING OF THE STELLAR POPULATION OF IC 10
WITH LASER GUIDE STAR ADAPTIVE OPTICS
AND THE HUBBLE SPACE TELESCOPE
William D. Vacca1, Christopher D. Sheehy2 , and James R. Graham2,3
Dept.ofAstronomy,UniversityofCalifornia,Berkeley,CA94720-3411
Draft versionFebruary 5, 2008
ABSTRACT
We present adaptive optics (AO), near-infrared images of the central starburst region of the Local
Group dwarf irregular IC 10. The Keck 2 telescope laser guide star facility was used to achieve near
7 diffraction limited performance at H and K′ with Strehl ratios of 18% and 32%, respectively. The
0
images are centered on the putative Wolf-Rayet (W-R) object [MAC92] 24. Photometry from AO
0
images can be subject to large systematic errors (≃1 mag.) due to uncertainties in the point spread
2
functions (PSFs) and the associated encircled energy curves caused by Strehl variations. However,

n IC10presentsarichstarfield,andthereforeweareabletousetheFourierpowerspectrummethodof
a
Sheehy,McCrady andGraham(ApJ, 647,1517)to reconstructthe photometric curveofgrowthfrom
J
the datathemselves,andtherebyreduce the magnitude ofthe systematicerrorsinourphotometryto
2 ≤ 0.04 mag. We combine our ground-based images with an F814W image obtained with the Hubble
2 Space Telescope. By comparing the K′ versus [F814W]−K′ color-magnitude diagram for the IC10

field with theoretical isochrones, we find that the stellar population is best represented by at least
1
two bursts of star formation, one ∼ 10 Myr ago and one significantly older (150−500 Myr). The
v
young, blue stars are centered around and concentrated in the vicinity of [MAC92] 24. We suggest
8
that this populationrepresentsthe resolvedcomponents ofan OB associationwith a half-lightradius
2
6 of about 3 pc. We resolve the W-R object [MAC92] 24 itself into at least six blue stars. Four of
1 these components have near-IR colors and luminosities that make them robust WN star candidates.
0 By matching the location of Carbon stars in the color-magnitude diagramwith those in the SMC we
7 deriveadistancemodulusforIC10ofabout24.5magandaforegroundreddeningofE(B−V)=0.95.
0 We find a more precise distance by locating the tip of the giant branch in the F814W, H, and K′
/ luminosity functions for IC 10. We find a weighted meandistance modulus (m−M) =24.48±0.08.
h 0
The systematic error in this measurement, due to a possible difference in the properties of the RGB
p
populations in IC 10 and the SMC, is ±0.16 mag.
-
o Subject headings: galaxies: individual (IC 10) – galaxies: starburst – galaxies: star clusters
r
t
s
a 1. INTRODUCTION ficialornaturalguidestarinthefieldofview. Suchguide
:
v Above the turbulence of the earth’s atmosphere, and stars provide input to the tip-tilt and wavefront sensors
i and,inthecaseofanaturalguidestar,astraightforward
X able to record near-diffraction-limited images, instru-
meansofdeterminingthepoint-spreadfunction(PSF)of
ments aboard the Hubble Space Telescope (HST) have
r the system. Examples of extragalactic studies using an
a provided data that have driven the field of high resolu-
AO system employing natural guide stars can be found
tion imaging of starburst galaxies (see, e.g., Whitmore
in Davidge (2003) and references therein. Nevertheless,
2000, and references therein). While most of the work
care must taken when using a natural guide star in the
with HST has been done at ultravioletand visible wave-
field to avoidsaturation,persistence,and scatteredlight
lengths, which are ideal for tracing hot, young stars,
problems. Furthermore, in general, extragalactic fields
imaging and spectroscopy in the near-infrared(near-IR)
selected for stellar population studies tend not to con-
is invaluable when there is substantial foreground or in-
tain bright isolated stars.
ternalextinctionoralargepopulationofredsupergiants.
With the deployment of laser guide star (LGS) tech-
Because angularresolutionat the diffractionlimit scales
niques, the requirement for a bright natural guide star
linearlywithwavelengthandinverselywithtelescopedi-
within the isoplanatic patch (a severe constraint when
ameter,the 2.5-mapertureof HSTrestrictsits effective-
attempting to observe individual extragalactic sources
ness in the near-IR. With the advent of adaptive optics
such as nearby starburst galaxies) can be removed. The
(AO) facilities at ground-based, 4-m and 8-m class tele-
need for a tip-tilt guide star (R . 17 mag.) within the
scopes,itis nowpossible to acquirenear-IRimageswith
isokinetic angle (≃ 30′′) remains because, by Fermat’s
angular resolution superior to those from HST and the
principle, a monochromatic laser beacon cannot be used
NICMOS instrument.
to sense wavefronttilt. However,this constraintis much
AOobservationsrequirethe presenceofeither anarti-
moreeasilysatisfied,evenwhenobservingsourcesoutof
1SOFIA-USRA, NASA Ames Research Center, MS N211-3, the Galactic plane.
MoffettField,CA94035; wvacca@sofia.usra.edu InthispaperwereportLGSAOnear-IRimagesofthe
2Center for Adaptive Optics, University of California, Santa central region of IC 10 obtained with the NIRC2 cam-
Cruz,CA95064, U.S.A.
[email protected] era at the Keck Observatory. IC 10 is a nearby irregular
2 Vacca et al.
starburst galaxy, often characterized as a blue compact 2. OBSERVATIONSANDDATAREDUCTION
dwarf(Richer et al.2001),locatedontheoutskirtsofthe 2.1. Near-Infrared Data
Local Group. The physical properties of IC 10 are sum-
marizedinthereviewbyMateo(1998). IC 10liesatlow Near-infrared H and K′ images of IC 10, centered
Galacticlatitude(l=119.0◦,b=−3.3◦)andreliableval- on [MAC92] 24 and approximately 10 arc sec E and
ues of the foreground reddening and distance have been 18 arc sec S of [HL90] 111c, were acquired with the
elusive,with estimates rangingfrom 0.47to 2.0 mag.for NIRC2 camera on the Keck 2 telescope on 2005 Novem-
E(B−V) andfrom22tomorethan27mag.forthe dis- ber10. Theseobservationswereobtainedusingthelaser-
tance modulus. A summary of distance and reddening guide star adaptive optics facility (Wizinowich et al.
estimates for IC 10 can be found in Sakai et al. (1999) 2006; van Dam et al. 2006). The NIRC2 pupil stop was
and Demers et al. (2004). in the LARGEHEX position and the camera was operated
IC 10 is remarkable for its high star formation rate, inthenarrowfieldmode,whichyieldsascaleof0′.′01per
as evidenced by its large number of H II regions pixel and a field of view of 10′.′24×10′.′24. The Nyquist
(Hodge & Lee 1990), Hα luminosity (Mateo 1998), and frequencies with this pupil stop on Keck 2 at H and K′
far-IR luminosity (Melisse & Israel 1994). Normalized are 65 and and 50 arc sec−1, respectively, and therefore
to its H surface density, IC 10 has much higher rate the images are well sampled.
2
of star formation than most spirals or dwarfs (Mateo Observing conditions were photometric according to
1998; Leroy et al. 2006). IC 10 has a large popula- data from the Mauna Kea all-sky monitor CONCAM4,
tion of Wolf-Rayet (W-R) stars (Massey et al. 1992; andthenaturalseeingwasmeasuredtobe0′.′3full-width
Massey & Armandroff 1995), and the highest surface at half maximum (FWHM) at H. Analysis of the AO
density of W-R stars among all Local Group galaxies wavefront sensor telemetry indicated that the Fried pa-
(Massey & Armandroff 1995). Furthermore, the ratio of rameterwasr0 ≃28cmandtheouterscaleL0 was∼57
carbon-type W-R’s (WC stars) to nitrogen-type W-R’s m at 500 nm. The airmass during the observations was
(WN stars) in IC 10 is unusually high for its metallic- about 1.3. The Na I D2 589.0 nm laser guide star, with
ity of approximately 0.2-0.3 solar (Skillman et al. 1989; an equivalent brightness of R ≃ 11 mag., was projected
Garnett 1990). These characteristics suggest that IC 10 atthe centerofthe sciencecamerafieldofview;a bright
has experienced a brief but intense galaxy-wide burst of star5 located due west ∼ 8′′ from the center of the field
star formation within the last ∼10 Myr. was used to perform tip/tilt corrections. The total ex-
Oneofthemostconspicuoussitesofrecentstarforma- posure times were 1200 s in H (1.63 µm) and 1500 s in
tioninIC10istheHIIregion[HL90]111c(Hodge & Lee K′ (2.12µm), brokenintofour andfive ditheredimages,
1990). This region also harbors the brightest putative respectively. Each image in the dithering sequence was
W-R star in the galaxy, [MAC92] 24, whose precise separated from the previous one by 1′.′0.
spectral type is uncertain. Richer et al. (2001) identi- Therawimageswerereducedusingourpipelineproce-
fied [MAC92] 24 as a WN star from the broad He II durefor Lick andKecknear-IRAO imagingdata,which
λ 4686 emission (and the lack of C IV λ4650 emis- is currently maintained at Berkeley by M. D. Perrin.
sion) in its optical spectrum. Based on its He II λ Thepipelineperformsdarksubtraction,flatfielding,bad
4686 equivalent width, Massey & Holmes (2002) classi- pixel repair, and stacking of images to create the final
fied this [MAC92] 24 as a WN star blended with an- mosaics. Flat-fields were generated from exposures of
other object. However, [MAC92] 24 was not recovered the interior of the dome obtained at the beginning of
by Royer et al. (2001) in their emission line survey of the night. Dome flats are preferable to twilight sky flats
this galaxy, possibly because their narrow band filters because they can be acquired with high signal-to-noise.
dedicated to W-R detection were contaminated by [O ThepupililluminationintheKeck2/AO/NIRC2system
III] λ 5007 from [HL90] 111c. Crowther et al. (2003) is well controlled and comparison of dome and twilight
subsequently noted that there were three closely-spaced sky flats shows only very small (negligible) differences,
sources, [MAC92] 24-A, B, and C, all located within presumably due to color and polarizationdifferences be-
1−2′′ of each other, that might be contributing to the tweenthetwosources. Asthedataarewellsampled,bad
spectrum recorded by Massey & Holmes (2002). They pixels in each 300-s exposure were repaired by replacing
also suggested that this object might be an O3If WN them with interpolated values derived from their neigh-
star, a WN+OB binary, or a stellar cluster containing bors. The dark-subtracted, flat-fielded images were cor-
a WN star. The high stellar density in the vicinity of rected for the known and previously mapped geometric
[MAC92] 24 is apparent from the HST/WFPC2 images distortion introduced by the NIRC2 camera, registered
reported by Hunter (2001) and Crowther et al. (2003): to a common coordinate frame, and combined. The re-
[MAC92]24liesat,ornearthecenteroftheyoungstellar sultingmosaicineachfilterhasasizeof12′.′2×12′.′2,and
cluster [H2001] 4-1 visually identified by Hunter (2001). the image size is 0′.′048 FWHM at H and 0′.′051 at K′.
The richness of this field, combined with the large fore- The final images are displayed in Figure 1.
groundreddening,makesIC10ahighlyattractivetarget Extracting photometry from the near-IRLGS AO im-
for high resolution study with near-IR laser-guide-star ages requires two distinct steps: 1) PSF-fitting, which
adaptive optics imaging techniques. establishes the relative brightness of stars; 2) analysis of
Ourobservationsanddata reductions arepresentedin the PSF to determine the photometric curve of growth.
§2. The results are presented in §3 and discussed in §4. While conventionalmethods aresatisfactorytoestablish
A summary is given in §5. relative photometry, measuring the photometric curve
4 \protecthttp://nightskylive.net/mk/
5 USNOB1.01492-0009059, Iphoto=13.59mag
IC 10 with LGSAO 3
Fig. 1.—Adaptive optics imagesofIC 10obtained withtheKeck2laserguidestarsystem andtheNIRC2facilitycamera. Left: H
band andRight: K′ band mosaics about 20 arcsecSE of the center ofthe HIIregion [HL90] 111c (Hodge&Lee 1990)and includethe
W-R candidate [MAC]24 (Masseyetal. 1992). Theimages arenear-diffractionlimitedwithStrehl ratios of 18% and32% at H andK′,
respectively. TheoriginofthecoordinatesystemcorrespondstothelocationofthenortherncomponentoftheW-Robject[MAC92]24-A
atRA(J2000)00h 20m 27s.36,DEC(J2000)+59◦ 17′ 37′.′33. Theorientationoftheimagesisconventional, withnorthupandeasttothe
left. Thestellarpopulationiswellresolvedandfaintnebularemissionarcsacrossthecenter oftheimages.
of growth needed to place AO measurements on an ab- 2 and HST/NICMOS data, Sheehy et al. (2006) estab-
solute scale requires special techniques. In a crowded lished that this procedure reduces systematic errors in
field, relative photometry is best done using the narrow photometric measurements to the few percent level.
diffraction-limited core. However, this core may contain We have appliedthe OTF-fitting algorithmsdescribed
as little as 5–10% of the total light. The degree of AO in Sheehy et al. (2006) to the current data set with no
correctionissensitivetotheobservingconditions(seeing, modification. ThefitstotheH andK′-bandpowerspec-
wind speed, brightness of the wavefront reference, etc.), traareshowninFigure2. Notethatthelowestwavenum-
andvariationsintheseconditionscausesvariationsinthe bers are excluded from each fit in order to mitigate the
amount of energy in the core of the PSF relative to the effects of possible contamination of the power spectra
uncorrectedseeing halo. This ratio,known as the Strehl by flat-fielding andsky subtractionerrors,and extended
ratio (SR), is exponentially sensitive to the variance of (nebular) emission. Based on the uncertainties in the
wavefront errors, and Strehl variations can be large; for reconstruction of the photometric curves of growth (en-
moderatecorrectionσ /SR≃1(Fitzgerald & Graham circled energy curves), which are shown in Figure 3, we
SR
2006). As the fractional photometric error is approxi- estimate that the systematic errors in the photometry
matelyequaltoσ /SR,failuretomeasureandaccount are ≤ 0.04 mag. in both bands. The reconstructed PSF
SR
forStrehlvariationsandthecorrespondingchangeinthe impliesthattheStrehlratiosofthefinalmosaicsare18%
curve of growth can introduce order of magnitude sys- in H and 32% in K′.
tematicerrorsinphotometry. Therefore,itisdesirableto WeinvestigatedPSFvariationsacrossthefieldofview
estimate the PSF from the data themselves rather than by segmenting the final mosaics into a number of sepa-
fromobservationsofaPSFstaratanothertime. (Thisis ratesub-regions,performingtheanalysisdescribedabove
a much more straightforwardtask in the case of natural on each sub-region separately, and constructing the en-
guide star AO observations, when the guide star itself circled energy curves for each region. Comparison of
canbe usedto determine the PSFandmeasurethe pho- the results show variationsofthe aperture correctionsof
tometriccurveofgrowth. See,e.g.,Davidge & Courteau less than 5%. Furthermore, we find no systematic varia-
(1999); Davidge (2003).) tion ofthe aperture correctionwith position in the field.
Sheehy et al. (2006) have shown that when the scene These results agree with both visual inspection of the
consists of a rich star field the photometric curve of images, which reveals no significant astigmatism at the
growth for adaptively-corrected data can be extracted edges of the field, and other measurements of anisopla-
bymodelingthepowerspectrumoftheimage. Modeling natism in AO images obtained under typical seeing con-
is performed in Fourier space and employs physical de- ditions at Mauna Kea (e.g., Davidge & Courteau 1999;
scriptions of the optical transfer functions (OTF) of the Davidge 2003). Hence, we conclude that the isoplanatic
atmosphere, telescope, AO system, and science camera. angle was large for these observations and we neglected
Although the Fourier power spectrum of the image of anyPSFvariationsoverthe field ofview inouranalysis.
a single star is proportional to |OTF|2, the power spec- We established the photometric zero point using ob-
trum of a star field is given by the product of |OTF|2 servations of the UKIRT near-IR standard star FS6
and the power spectrum of the spatial source distribu- (Hawarden et al.2001), obtainedonthe same night. We
tion. Therefore, both the imaging system and the as- didnotobserveenoughstandardsstarsto determine the
tronomical scene must be modeled. By comparing Keck atmospheric extinction coefficient from our data. How-
4 Vacca et al.
0.20
0.15 σσHH SRteacrofivnedreerd
σH 0.10
0.05
0.00
106.20 18 20 22
0.15 σσKK′′ SRteacrofivnedreerd
0.15
′K 0.10
σ σ′K0.10
0.05
0.05
0.00
0.00
1166 1188 2200 2222
mm
Fig. 2.—MeasuredandbestfitpowerspectrafortheH(top)and
K′ (bottom) band mosaics in Figure1. Fitting the optical trans- Fig. 4.— Two estimates of the internal errors in the H and
fer function (OTF) of the atmosphere, telescope, AO system and K′bandphotometry. Greypointsarephotometricerrorsreturned
science camera yields the Strehl ratio and allows the photometric byStarFinder forindividualstars. Thesolidlinesarethe photo-
curve of growth to be evaluated (Sheehy etal. 2006). Open dia- metricerrors derived froman artificial star test. The StarFinder
monds show the one-dimensional power spectrum extracted from errors vary with field position because the exposure time is non-
the images. Solidlines arethe best fit model power spectra com- uniformacross the mosaic. The solidlines from the artificial star
putedfrommodelsystemOTFmultipliedbythepowerspectrum experimentrepresentanaverageoverthefieldofview.
of the source distribution. Spatial frequency is measured in nor-
malizedunitsrelativetotheNyquistfrequencyateachwavelength.
Comparisonofthe power atνn ≃0.6atH andK′ show that the etrypackageStarFinder(Diolaiti et al.2000)toperform
correctionatthelongerwavelengthissuperior. Thecorresponding source detection and PSF fitting. Photometry result-
Strehl ratios at H and K′ are 18% and 32%, respectively. The ing from PSF fitting returns the relative brightness and
firstfewwavenumbersareexcludedfromthefits,asthesearecon-
locations of the stars in the field of view, and therefore
taminated by flat fielding and sky subtraction errors, as well as
extended emission. providestheinformationnecessarytoestimatetheshape
of the power spectrum of the source distribution.
We identified 661 stellar sources in the H image and
1.0 585sourcesinthe K′ image. Forthe brightestfew stars,
H
K’ speckles in the PSF halo that lie outside the 0′.′19 em-
0.8 piricalPSFradiuswereerroneouslyidentifiedassources.
As the locationofspeckles varies with wavelength,these
y
erg 0.6 spurious detections were easily identified and eliminated
n
d E by hand. A variance map for each mosaic was con-
e
cl structed using the formulae given by Vacca et al. (2004)
ncir 0.4 andsuppliedtoStarFindertocalculatetheinternalpho-
E
tometric errors. The photometric errors returned by
0.2 StarFinderareshowninFigure4. Severaldiscretetracks
areevidentinthisfigurebecausethetotalexposuretime,
0.0 andhencethesignal-to-noiseratio,isnon-uniformacross
0 20 40 60 80 100
pixels themosaics. Forexample,thetotalexposuretimeforthe
centralareaofthefive-pointditherpatternwasfivetimes
Fig. 3.— Reconstructed curves of growth (encircled energy) as
afunctionofradiusfortheH andK′ bands. larger than that at the edges.
Photometric completeness limits were determined by
inserting artificial stars of known magnitude into each
ever, the H and K′ filters in NIRC2 are similar to those
mosaic and attempting to recover them. We performed
in use at the IRTF, as they were produced as part of
this procedure for magnitudes between 16.0 and 25.0 in
the Mauna Kea filter consortium (see Tokunaga et al.
steps of 0.5 mag. To accumulate good statistics without
2002). Therefore,wehaveusedthenominalatmospheric
substantially increasing the crowding in the images, we
extinctioncoefficientsof0.059mag.airmass−1 and0.088
restrictedthenumberofstarsinsertedto50andrepeated
mag. airmass−1 for H and K′ respectively given on the
the procedure five times for each magnitude bin. Figure
IRTF web site6 to make the small corrections necessary
5 presents the resulting completeness as a function of
to account for the small difference in airmass between
magnitude. The limiting magnitudes are 22.3 and 21.5
our science target and the standard star. inH andK′,respectively,foracompletenessthresholdof
We performed relative photometry by constructing an
50%recovery. Inordertotestthephotometricerrorspre-
empirical PSF with a radius of 0′.′19 from seven bright
sented in Figure 4, we measured the difference between
andrelativelyisolatedstarsinthefieldusingDAOPHOT
the input magnitudes and the recovered magnitudes for
(Stetson1987). Wethenusedthe crowdedfieldphotom-
eachartificialstar. Thewidthofthebest-fitGaussianto
theseresidualsasafunctionofinputmagnitudeisshown
6 http://irtfweb.ifa.hawaii.edu/$\sim$nsfcam/hist/backgrounds.html;
asthe solidline in Figure4. The agreementbetweenthe
seealso
http://www.jach.hawaii.edu/UKIRT/astronomy/calib/phot cal/cam zpt.whtomlerror estimates is satisfactory.
IC 10 with LGSAO 5
K′
1.0 H
0.8
s
s
e
n
e 0.6
et
pl
m
o 0.4
C
0.2
0.0
16 18 20 22 24 26
magnitude
Fig. 5.— Photometric completeness of the H and K′ band Fig. 6.— Internal errors in the F814W photometry were es-
photometry as afunction oflimitingmagnitude, derivedfromthe timated by comparing photometry from the three individual flt
artificialstartests. images. Thisplotshowstheresidualsbetweenthe fltmagnitudes
andthemeanmagnitude. Thesolidlineshowsthestandarddevi-
ationoftheresidualsin1mag.widebins. For[F814W]<23mag.
We note that our near-IR photometry is substan-
thermsinternalerrorisapproximately0.04mag.
tially deeper and more accurate than that presented by
Borissova et al. (2000). Our completeness limits are ∼5 well within the RMS error found by examining the scat-
and ∼ 3 mags deeper in H and K′, respectively, and at ter of the photometry derived from the three individual
any given magnitude, our internal errors are a factor of flt images about the mean (Fig. 6). This comparison
∼ 5 smaller. Of course, our field of view is considerably revealsthattheuncertaintiesare<0.04mag.forobjects
smaller (0.04 arcmin2 vs. 13 arcmin2). brighter than [F814W]=23.0 mag.
Instrumental magnitudes were converted to a Vega-
2.2. HST Data basedsystembyadoptingazero-pointof25.501mag. We
shallrefertotheseas[F814W]magnitudes. Wehavenot
Optical imaging was obtained with the Wide Field
transformedthesevaluestoaground-basedsystem(e.g.,
Camera (WFC) of the Advanced Camera for Surveys
Bessell & Brett 1988). Restricting the analysis to stars
(ACS) (Ford et al. 2003). Three F814W-band images
within the NIRC2 field, we find ∼ 870 sources brighter
were acquired as part of program GI-9683 (PI: Bauer)
and were retrieved from the HST archives7. The to- than [F814W] = 26.5 mag. Our estimated 50% com-
pletenesslimit is between25and25.5mag,for whichwe
tal exposure time was 1080 s. We used the flt (cal-
find ∼690 sources.
ibrated individual exposure) and crj (calibrated, cos-
Stars in the NIRC2 field were matched with stars in
mic ray-split-combined image) files processed with the
theACSfieldusinganearest-neighboralgorithm. Detec-
standardHST/ACS pipeline, whichincluded corrections
tor coordinates of sources in the geometrically-distorted
for the bias level, dark current, flat fielding, flux cali-
ACS frame were transformed into sky coordinates using
bration and (for the crj files) cosmic ray rejection and
the PyRAF8 utility tran, and back into NIRC2 detec-
image combining. The mean ACS/WFC pixel scale is
approximately0′.′05. Since the Nyquistfrequency atthis tor coordinates. As the pixel scale of the ACS/WFC is
wavelength on HST is 30 arc sec−1, these data are un- 5 times coarser than that of the NIRC2 narrow camera
and pixel mismatches may be large, we selected sources
dersampled.
that had common centroids within a 7 NIRC2 pixel ra-
Photometry was performed using the software de-
dius (70 mas). To further prevent false associations in
scribedbyAnderson & King(2006). Thiscodeidentifies
the automatically generated catalog, we restricted our
sourcesontheimagesaboveaskythreshold,generatesa
matches to stars separated from one another by at least
spatiallyvariablePSFforthespecific filter,andfits itto
thisdistanceinbothindividualcatalogs. Inthecrowded
each object detected on the image to yield a distortion-
region around [MAC92] 24 the threshold for matching
corrected instrumental magnitude and position. The
was lowered and each potential match was inspected by
PSFisnormalizedtohaveafluxofunitywithinaradius
of10pixels,or0′.′5. Althoughthiscodewasdevelopedfor hand before being added to the catalog. This procedure
resulted in [F814W]−K′ colors for 380 stars.
useonthe fltimages,wefoundonlyasmallsystematic
difference (0.025 mag.) between the instrumental mag-
3. RESULTS
nitudes derived from the flt images and those derived
from the crj image. The latter has the advantage of 3.1. Three-Color Image
eliminating many spurious sources due to detector ar- A three-color image of the IC 10 field, combining the
tifacts and cosmic rays. Furthermore, this difference is F814W, H, and K′ data and produced using the algo-
rithmdescribedbyLupton et al.(2004),isshowninFig-
7 ACS/WFC F435W and F606W images are also available in ure 7. The Lupton et al. (2004) algorithm ensures that
the archives as part of this program. We chose to use only the
anobjectwithaspecifiedastronomicalcolorhasaunique
F814W data because the shorter wavelength images suffer from
greater undersampling and greater sensitivity to foreground and
internalextinction. 8 http://www.stsci.edu/resources/software hardware/pyraf
6 Vacca et al.
colorintheRGBimage. Thecolorimagedrawsattention tion of H −K color9. We used the relations given by
to the spatial distribution of the red and blue sequences Sirianni et al. (2005) to convert Cousins I band magni-
that are evident in the color-magnitude diagram(CMD; tudesgivenbyZaritsky et al.(2002)toHST/ACS/WFC
see Fig. 10) and serves as a useful aid in the identifi- [F814W]. Cioni et al. (2006) identify C-rich and O-rich
cation and discussion of individual sources. In Figure 8 asymptotic giant branch (AGB) stars in the SMC based
we show an annotated version of the K′ image of this upon their J and K photometry in the DENIS and
s
field. We have labeled the brightest, bluest and reddest 2MASScatalogs. Usingtheircriteriaforpartitioningthe
objects in the image, as well as other sources whose lo- K versus J −K CMD into regions of distinct stellar
s s
cations in the color-magnitude diagram are noteworthy. types, we identified AGB stars in the K versus J −K
s s
Photometry for these objects is listed in Table 1. CMD of Zaritsky et al. (2002) and carried the divisions
over to the K′ versus [F814W]−K′ CMD. The O-rich
and C-rich AGB stars in the SMC sample are plotted in
Figure 11 as magenta and red points, respectively. The
3.2. The Color-Magnitude Diagram
restoftheSMCstars,includingtheredgiantbranch,are
From our photometric catalog we generated the plotted as green points. We assumed a distance modu-
[F814W], H, and K′ luminosity functions as well as the lus to the SMC of 18.95± 0.05 mag. and a reddening
K′ versus [F814W]−K′ CMD. The former are shown in of E(B − V) = 0.15±0.06 mag. for consistency with
Figure 9 while the latter appears in Figure 10. Two Cioni et al. (2000).
distinct sequences are populated in the CMD: a vertical To compare the SMC and IC 10 CMDs directly (in
band of blue stars with [F814W]−K′ ≃0.5 mag., which the same figure), it is necessary to adopt a foreground
are most likely reddened main sequence objects, and a reddening and a distance modulus for IC 10. After ex-
well-populated plume of red stars with [F814W]−K′ in perimenting with a range of values found in the liter-
the range 2–3.5 mag. In addition there are several very ature (see Demers et al. 2004), we found that a good
red objects with [F814W]−K′ >4 mag. match between the SMC and the IC 10 distributions
We estimated the contamination of foreground Galac- couldbe achievedwithaforegroundreddeningofE(B−
tic stars in our CMDs using the Galactic models of V) = 0.95±0.15 mag. and a distance modulus of 24.5
Cohen (1994). Although IC 10 is at low Galactic lati- mag. The reddening is determined primarily by match-
tude, foreground contamination is not severe in such a ing the [F814W]−K′ color of the red giant branch (cf.
small field. In H band, we find that we expect about Sakai et al. 1999). If we focus on the less abundant blue
0.01 stars arcsec−2 brighter than 18 mag., 0.025 stars stars, and attempt to match their location in the CMD
arcsec−2 brighterthan20mag.,and0.048starsarcsec−2 with that of the very top of the main sequence in the
brighterthan22mag. Thesestellarsurfacedensitiescor- SMC distribution, we find that a somewhat smaller red-
respond to 1, 4, and 7 stars, respectively, in our NIRC2 dening valueofE(B−V)∼0.65mag.is preferred. This
field. Similar values are found for the K′ band. Thus, estimatehasalargeuncertainty,however,giventhesmall
above our 50% completeness limit we suffer about 1% numberofblue main-sequencestarsabovethe complete-
contamination. The contaminating stars are expected nesslimitsinboththeIC10andSMCsamples. Wenote
to be main sequence dwarfs, and therefore should have that IC 10 has abundant molecular clouds (Leroy et al.
moderately red colors with [F814W]−K′ in the range 2006) and there is ample evidence for substantial inter-
0.8–2.2 mag. We do not expect that any of the blue nal reddening. The differential extinction between the
stars ([F814W] − K′ ≃ 0.5 mag.) or any of the ex- blueandredstarsmayindicatethatthe bluestarslieon
tremely red stars ([F814W]−K′ > 4 mag.) are fore- the front face of IC 10,in a regionthat has been cleared
ground objects. Two bright (K′ ≃ 17 mag.) stars, la- of gas and dust by the O-star winds and supernovae as-
beled9and10inFigure10,havethecolorsofforeground sociated with the star formation activity evident in the
dwarfs. However, the close spatial association of star 9 vicinity of [HL90] 111c.
with [MAC92] 24 suggests that it is a reddened super- The comparison with the SMC indicates that the red
giant in IC 10, with an initial main sequence mass of population in IC 10 comprises both red giants and AGB
≃25M⊙. The status of star 10 is uncertain. stars. There are two branches of very red ([F814W]−
We also investigated possible contamination by back- K′ >4mag.) starsfoundinourIC10field. C-richAGB
ground galaxies. Based on the galaxy counts given by stars comprise the bright branch. Highly evolved AGB
Kochanek et al. (2001) and Drory et al. (2001) and the stars with optically thick circumstellar dust envelopes
size of our fields, we would expect to find less than one populate the dim branch. Additional red stars between
background galaxy in our K′ band image down to our K′ = 20 and 22 mag. are undoubtedly present, but the
50% completeness limit of 21.5 mag. This agrees with lower right corner of the CMD is unpopulated due to
our finding that most of the objects in our image are incompleteness in the F814W measurements.
point-like (see below). We have identified red supergiants (RSGs) in the
The metallicity of the SMC is close to that of IC Zaritsky et al. (2002) catalog of SMC stars and plot-
10, and therefore this nearby system forms a natu- ted them on Figure 11. We began with the list
ral template against which IC 10 can be compared. of spectroscopically-confirmed Magellanic Clouds RSGs
We used the 2MASS and DENIS photometric cata- from Massey & Olsen (2003). Cross-correlation of the
logs for the SMC provided by Zaritsky et al. (2002) for coordinatesandBandV magnitudesoftheseRSGswith
this task. We transformed the 2MASS magnitudes to
the NIRC2/MKO system using the relations given in 9 (K′−K) ≃ 0.11(H −K). This relation was derived using
the method of Wainscoat & Cowie (1992), based on the central
Carpenter (2001) along with an estimate of the con-
wavelengths ofourfilters
version between K and K′ filter magnitudes as a func-
IC 10 with LGSAO 7
Fig. 7.—Three-colorIHK′ compositeofthe[HL90]111cHIIregionofIC10. ThecoordinatesystemiscenteredontheW-Rcandidate
[MAC92]24-A/N.Northisupandeastistotheleft. Thenear-IRH andK′imagesarelaserguidestaradaptiveopticsdatafromtheKeck
2telescope obtained withNIRC2, andthe I-banddata isF814W fromtheAdvanced CameraforSurveys ontheHubble Space Telescope.
Thisimagewasconstructed usingthealgorithm ofLuptonetal.(2004), whichensures that anobjectwithaspecifiedastronomical color
hasauniquecolorintheRGBimage. Distinctpopulationsofredandbluestarsareevident. Thebluestarscomprisemainsequencestars
andbluesupergiants;theredstarsarepredominantlyredgiants. ThebrightestredstarsareAGBstars. Thereddeststarsarecarbonstars
andevolved AGB starswiththickcircumstellarenvelopes. Consultthe color-magnitudediagram (Fig. 10)andthe finding chart(Fig. 8)
toidentifyspecificobjects. Diffuselighttracesnebularemissionfromthe111cHIIregion.
objectsintheZaritsky et al.(2002)catalogyieldedtheir location of red supergiants. These tracks also better
location in the [F814W] versus [F814W]-K′ CMD. This match the observed integrated colors of star clusters in
exercise indicates that there are no unambiguous RSGs IC 10 (Hunter 2001). We convertedthe WFPC2 F814W
withinourfieldofview,exceptforperhapsthebrightstar magnitudes given by Lejeune & Schaerer (2001) to the
located on the southwest edge of the frame for which we ACS/WFCF814Wbandusingthe prescriptiongivenby
did not attempt to perform photometry. Sirianni et al.(2005). TheBessell & Brett(1988)H and
In Figure 11 we have overlaid theoretical stellar K bandmagnitudesgivenby Lejeune & Schaerer(2001)
isochronesforZ =0.008fromLejeune & Schaerer(2001) were converted to the NIRC2/MKO system using the
on our observed CMD. The oxygen abundance in IC transformations given by Carpenter (2001), along with
10 in HII regions corresponds to Z ≃ 0.005 (0.25Z⊙) the aforementioned correction between the K an K′ fil-
(Skillman et al. 1989; Garnett 1990). Thus, one would ter magnitudesas a functionof H−K color. We shifted
expect that Z = 0.004 tracks of Lejeune & Schaerer the isochrones to correspond to a distance modulus of
(2001) would be most suitable for such a comparison. 24.5. We alsoreddenedthe isochrones,withthe adopted
However,asfoundbyHunter(2001),themoremetal-rich reddening value dependent on the corresponding age of
Z =0.008 isochrones extend farther to the red than the each isochrone. This was done in order to account for
Z =0.004 isochrones do and so encompass the observed the differing reddening values found above for the red
8 Vacca et al.

14
16
9 10 18 20
16 19
′K 18 3(244(-2A4/-SA)/N) 11 1123141517 23
7
2122
20
5(24-B/E)
2(24-B/W)8(24-C)
6(24-A/W)
22
1
0 2 4 6
[F814W] - K′
Fig. 10.—K′vsI−K′color-magnitudediagramfortheNIRC2
field. Consult Fig. 8 to identify the location of the numbered
objects. Thedashedlineisthe50%completeness limit.
Fig. 8.— K′ finding chart with scale and orientation matching the respective reddenings of the blue and red stars de-
Fig. 7. Thecomponentsof[MAC92]24-AandBarelabeledWR24-
terminedfromcomparisonoftheIC10andSMCCMDs.
A and WR24-B, respectively. The object number can be used to
find the stars’ location on the K′/ [F814W]−K′ color-magnitude For isochrones corresponding to ages older than 20 Myr
diagram (Fig. 10). Note that the object numbering scheme runs we applied a reddening of E(B − V) = 0.95 mag., as
fromleftto rightinFig. 10. Thus, object number 1is the bluest determined from our comparison of the locations of the
objectandobjectnumber22isthereddest.
red stars in the SMC and IC 10 CMDs.
The comparison between our data and the isochrones
indicates thatthe [MAC92]24regionofIC 10comprises
20
15 (a) (b) (c) twoseparatestellarpopulations: onegroupofbluemain-
N(m)10 IC10 sequence stars and a few blue supergiants,younger than
5
0 about 10–20 Myr, and a second group of red giants and
2m)/dm 000...012 (d) (e) (f) AchGotBomstyarssuwggitehstsagthesatbtehtweereence1n5t0staanrdbu5r0s0ttMhaytr.prTohdiuscdeid-
2dN(--00..21 the W-R stars in [MAC92] 24 sits atop an older field
18 19K′20 21 18 19 H 20 21 21 2[2F814W2]3 24 population. The comparison with the isochrones draws
attentiontoseveralstarsthataremuchredderthanpre-
5000
4000 (g) (h) (i) dicted by the theoretical models. As noted above, the
m)3000 SMC
N(2000 bright objects are C-stars, while the fainter objects are
2m101200000 (j) (k) (l ) pmrionboasbitlyy,hbiugthleyxetivnoglvueisdhAedGbBysotaprtsiccallolsyetthoictkh,eidrufisntyalcliur--
m)/d 0 cumstellar envelopes.
2dN(-10
-20
11 12 13 14 15 11 12 13 14 15 13 14 15 16 17 3.3. Tip of the Red Giant Branch Distance
K′ H [F814W]
The qualitative comparison between the CMDs for
Fig. 9.— Top six (a–f) panels show luminosity functions in the SMC and IC 10 indicates a distance modulus of
K′, H, and [F814W] (a–c, respectively) their corresponding sec-
about 24.5 mag (see §3.2). A quantitative approach
ondderivatives(d–f)forIC10. Thebottom sixpanels(g–l)show
the same plots for the SMC. The second derivative (lower row of to matching the two CMDs using the two-dimensional
each set), which is used to locate the tip of the red giant branch, Kolmogorov-Smirnovtest (Fasano & Franceschini 1987)
is computed using according to the recipe of Cionietal. (2000) proved unsatisfactory because the results were strongly
includingsmoothing withaSavitzky-Golay filterandacorrection
biasedbytheincompletenessofthetwophotometriccat-
that accounts for the systematic offset between peak in the sec-
ond derivative and the location of the TRGB. The displacement alogs. However, we can refine the distance estimate by
between the vertical linesinthe upper andlower panels indicates using the tip of the red giant branch (TRGB) method
themagnitudeofthiscorrection. Thelightgreylineisthebestfit
(Lee et al. 1993). The TRGB marks a strong break
Gaussiantothesecondderivativepeak.
in the luminosity function of the Magellanic Clouds at
and blue stars. For ages of ≤ 20 Myr, we found that a K ≃ −6.6 mag. and provides a readily identifiable fidu-
reddening value of E(B−V)=0.60 mag. yields a good cial point for distance measurements.
match between the theoretical main sequence given by We adopted an empirical approach, comparing the
the models and the location of the blue stars in the IC TRGB brightness levels in the SMC and IC 10 data
10 CMD. This reddening value is somewhat lower than sets to derive a relative distance modulus between the
the foregroundextinction of E(B−V)=0.8±0.1 mag. two systems. We followed the method developed by
recommended by Massey et al. (1992) and Richer et al. Cioni et al.(2000)toestimatethelocationoftheTRGB
(2001), but is perhaps preferable given the large wave- from the [F814W], H, and K′ luminosity functions in
lengthbaselineaffordedbythecurrentmeasurements. In these two systems. The luminosity functions and their
addition, it agrees well with the results found above for secondderivatives,fromwhichthelocationoftheTRGB
IC 10 with LGSAO 9
Fig. 11.— K′ versus [F814W]−K′ color-magnitude diagram for IC 10. The solid lines denote the theoretical Z = 0.008 isochrones
fromLejeune&Schaerer(2001). Theinitialmassfunctionterminatesatanuppermassof120M⊙. ColoreddotsarestarsfromtheSMC
photometric catalog of Zaritskyetal. (2002), including O-rich (magenta) and C-rich (red) AGB stars. The rest of the SMC population,
including the red giants and very red, dust enshrouded objects are plotted as green. Both the isochrones and the SMC data have been
reddenedandshiftedinmagnitude,usingthevaluesgivenintheupperleftlegendtomatchtheIC10dataTheisochronescorrespondingto
ages<20MyrhavebeenreddenedbyE(B−V)=0.60mag.,whiletheisochronesforolderageshavebeenreddenedbyE(B−V)=0.95
mag. TheSMCdatahavebeenreddenedbyE(B−V)=0.95mag. Opendiamondsarespectroscopically-confirmedMagellanicCloudred
supergiantsfromZaritskyetal.(2002). AccordingtoartificialstarexperimentstheIC10photometryis50%completeatthedashedblack
line.
is determined, are shown in Figure 9. Considered indi- errors resulting from any differences in the properties of
vidually, the second derivatives of the F814W], H, and the red stellar populations between IC 10 and the SMC.
K′ luminosity functions for IC 10 seem noisy, and the Inordertoestimatethissystematicerror,weusedthere-
choice of which peak corresponds to the TRGB ambigu- sults givenby Montegriffo et al.(1998) for population II
ous. However, simultaneous consideration of all three giants to determine the possible systematic uncertainty
wavelengths reveals that there is a common feature rel- intheK′-bandTRGBmagarisingfromthewidthofthe
ative to the SMC TRGB (bottom panel of Fig. 9) cor- [F814W]−K′ color of the red giant branch in our CMD
responding to a unique relative distance modulus. Com- for IC 10. Assuming that the bolometric luminosity at
biningthedifferentialdistancemodulifromthe[F814W], theTRGBisconstantandadoptingthem −K calibra-
bol
H, andK′ luminosity functions, using a weightedmean, tionforpopulationII giantsofMontegriffo et al.(1998),
andcorrectingforthefactthattheuncertaintyintheex- wefindthatanydifferencesinthethe[F814W]−K′color
tinction correction introduces covariance between these of the red giant branches between the SMC and IC 10
measurements, we find (m−M) = 24.48±0.08 mag., should produce a systematic error in the true distance
0
where we have used the distance and reddening to the modulusofnomorethan±0.16mag. Thisvalueincludes
SMC adopted in §3.2. Because we adopted a large un- the uncertainty in the reddening, E([F814W]−K′).
certainty in the reddening towards IC 10 (E(B −V) = Wenotethatourreddeninganddistanceestimatesare
0.95 ± 0.15 mag.), the K′-band TRGB brightness is in good agreementwith those determined by Saha et al.
strongly weighted in this result. Since the K′-band (1996) and Wilson et al. (1996) from an analysis of
TRGB brightness is sensitive to variations in age and Cepheid data. Our reddening estimate is about 20%
metallicity,ourdistanceestimateissubjecttosystematic smaller and our distance is 20% larger than the values
10 Vacca et al.
Fig. 13.— Spatial distribution of young (< 10 Myr) blue stars
(left)andolderredstars(right). Theyoungstarsareconcentrated
in the vicinity of [MAC92] 24. The three circles (dotted, dashed,
Fig. 12.— Anenlargementoftheregionsurrounding[MAC92]24 dot-dash) indicate the centroid and half-light radii of the young
fromFig.7,showingthat the W-R candidate [MAC92]24-Acon- stars in [F814W], H, and K′, respectively. The half-light radius
sists of at least three components, two of which (24-A/N and isabout 0′.′8(3 pc) inall three bands. The coordinate gridis the
24-A/S) have nearly equal brightness and blue colors. A third same as in Fig. 7 with N up and E to the left, and centered on
component 24-A/W(0′.′3Wand0′.′1Sof24-A/N)issignificantly the W-R star [MAC92] 24-A/N. The cross marks the location of
fainter,butisalsoblue. Thefieldisdirectlycomparablewiththe cluster 4-1 determined by Hunter (2001). The largest diamonds
WFPC2/F555W imageofCrowtheretal.(2003). have K′ = 16.5−17.1 mag., the smallest have K′ = 21.9−22.5
mag.
estimatedbySakai et al.(1999),whoalsousedCepheids
for their determinations.
strong function of time; for an instantaneous burst of
3.4. Wolf-Rayet Stars and Young Clusters starformationandaSalpeterIMFextendingto100M⊙,
the W-R/O ratio is ≥ 0.1 only between 3.8–4.7 Myr
An enlargement (5′′ × 5′′) of the region surrounding (Leitherer et al. 1999). Our W-R candidates are drawn
[MAC92] 24 is provided in Fig. 12. This figure can froma list of29 blue stars brighter than the K′ mag ex-
be compared directly with the WFPC2/F555W image pected for a B0V star at the distance and reddening of
shown by Crowther et al. (2003). Our observations re- IC 10. Therefore, it would not be surprising if our four
solve the W-R source [MAC92] 24-A into at least three robustW-Rcandidateswereconfirmedspectroscopically
components, two of which (24-A/N and 24-A/S) have to be WN stars. The corollaryis that we can state with
nearlyequalbrightnessandbluecolors([F814W]−K′ < confidencethatmassivestarformationinthe[HL90]111c
0); a thirdcomponent, 24-A/W(0′.′3 W and0′.′1 S of24- region was active as recently as four Myr ago.
A/N),is significantlyfainter butis alsoblue ([F814W]− IfourproposedW-Rcandidatesareconfirmedspectro-
K′ ≃ 0.1). The sources [MAC92] 24-A/N and S are scopically, the WC/WN ratio in IC 10 would drop from
the brightest blue objects in the K′ versus [F814W]- 1.2–1.3 (Massey & Holmes 2002; Crowther et al. 2003)
K′ CMD (Fig. 11). Their observed colors and mag- to ∼1. While this value is still anomalously high, given
nitudes placethemaboveandto the leftofthe reddened thelowmetallicityofIC10,ourresultssuggestthatthere
main sequence; their extinction-corrected near-IR col- maybeanumberofWNstarsyettobediscoveredinthe
ors (([F814W]−K′) ≃ −0.6 mag.; (H −K′) ≤ 0.1 richstellarfieldsassociatedwiththeregionsofrecentstar
0 0
mag.) and K-band luminosities (MK′,0 ≃ −6.2 mag.) formation in this galaxy.
are consistent with their being late-type WN (WN7–9) In addition to the stars we have discussed above,
stars (e.g., Crowther et al. 2006). If [MAC92] 24-A/N there is also a handful of fainter blue objects clustered
and S are indeed late-type WN stars, they should have within 0′.′2 around [MAC92] 24-A/N. The concentra-
relatively weak lines, which may explain both their ab- tion of luminous blue stars in the region suggests that
sencefromtheemission-linesurveyofRoyer et al.(2001) the [MAC92] 24 complex could be considered a resolved
and the weakness of the He II emission detected by stellar cluster, as originally suggested by Hunter (2001)
Massey & Holmes(2002)inthe spectrumofthe blended basedoninspectionof WFPC2data. Comparisonof the
object observed under seeing-limited conditions. The relative spatialdistributions of blue and red stars,as re-
W-R candidate [MAC92] 24-B is also resolved into two vealedbyourF814Wandnear-IRdataandshowninFig.
blue sources, oriented east-west. The individual compo- 13,considerablystrengthensthis proposal. Inthis figure
nents of [MAC92] 24-B have colors similar to those of we have divided the stellar population into “red” and
[MAC92] 24-A/N and S, but are about 2 mag. dimmer, “blue” samples based on their locations relative to the
whichmakesthemcandidateearly-typeWNstars. Based 10 Myr isochrone in the CMD (Figure 11). To account
on its color and magnitudes, [MAC92] 24-C appears to forphotometricerrorswealsoincludeinthebluesample
be a late O V star. High spatial resolution spectroscopy those stars that lie redward of this isochrone within a
is needed to confirm our proposed spectral types for the swathofwidthequaltothatofthe1σ uncertaintyinthe
various components of [MAC92] 24. [F814W]−K′ color.
The abundance of W-R stars relative to O stars is a Withthisdefinition,theredstarsappeartobefarmore