Isolated OB Associations in Stripped HI Gas Clouds

0.8 MB·English

by  J. K. Werk

Draftversion February3,2008
PreprinttypesetusingLATEXstyleemulateapjv.10/09/06
ISOLATED OB ASSOCIATIONS IN STRIPPED H I GAS CLOUDS
J. K. Werk1, M. E. Putman1, G. R. Meurer2, M. S. Oey1, E. V. Ryan-Weber3, R. C. Kennicutt, Jr.3, and K. C.
Freeman4
Draft versionFebruary 3, 2008
ABSTRACT
HST ACS/HRC images in UV (F250W), V (F555W), and I (F814W) resolve three isolated OB associations that
lie up to 30 kpc from the stellar disk of the S0 galaxy NGC 1533. Previous narrow-band Hα imaging and optical
spectroscopy showed these objects as unresolved intergalactic H II regions having Hα luminosities consistent with
8
single early-type O stars. These young stars lie in stripped H I gas with column densities ranging from 1.5 - 2.5
0
×1020 cm−2 and velocity dispersions near 30 km s−1. Using the HST broadband colors and magnitudes along with
0
2 previously-determined Hα luminosities, we place limits on the masses and ages of each association, considering the
importance of stochastic effects for faint (MV > −8) stellar populations. The upper limits to their stellar masses
n
range from 600 M⊙ to 7000 M⊙, and ages range from 2 - 6 Myrs. This analysis includes an updated calculation of
a
the conversionfactorbetweenthe ionizingluminosityandthe totalnumberofmainsequenceOstarscontainedwithin
J
an H II region. The photometric properties and sizes of the isolated associations and other objects in the HRC fields
0
are consistent with those of Galactic stellar associations, open clusters and/or single O and B stars. We interpret
3
the age-size sequence of associations and clustered field objects as an indication that these isolated associations are

most likely rapidly dispersing. Furthermore, we consider the possibility that these isolated associations represent the
]
h first generation of stars in the H I ring surrounding NGC 1533. This work suggests star formation in the unique
p environmentof a galaxy’soutermost gaseousregions proceeds similarly to that within the Galactic disk and that star
- formation in tidal debris may be responsible for building up a younger halo component.
o
r Subjectheadings: galaxies: starclusters–HIIregions–intergalacticmedium–galaxies: halos–stars:
st formation
a
[
1. INTRODUCTION as they lie at similar distances from the host galaxies,
1 andexhibitsimilarlevelsofstar-formation(Thilker et al.
H II regions in gaseous tidal arms, far beyond the ro-
v 2007). Muchspeculationexistsastowhetherintergalac-
tating disks of galaxies,present an opportunity to study
8
tic H II regions represent isolated stars, what role tidal
5 the star formation process in conditions vastly different
interactionsplayintheirformation,whethertheirstellar
7 from those in typical galactic spiral arms. They may
populationsaretypicalofyoungstarclustersinquiescent
4 even contribute to our understanding of satellite galaxy
systems,andwhethertheycontributesignificantlytothe
. formation and evolution, halo stellar populations, and
1 enrichment of the IGM.
the enrichment of the intergalactic medium. Such inter-
0 Five intergalactic H II regions, three of which have
galactic H II regions have been found in galaxy clusters,
8 beenspectroscopicallyconfirmed,lieinthevicinityofthe
compact groups, and in the halos of galaxies (e.g. Bo-
0
galaxy NGC 1533. Ryan-Weber et al. (2004) discovered
: quien et al. 2007; Walter et al. 2006; Ryan-Weber et
v al. 2004; Oosterloo et al. 2004; Mendes de Oliveira these H II regions in the continuum-subtracted narrow-
i band image of NGC 1533 taken by the Survey for Ion-
X et al. 2004; Cortese et al. 2003; Sakai et al. 2002;
ization in Neutral Gas Galaxies (SINGG, Meurer et al.
Gerhard et al. H II 2002; Arnaboldi et al. 2002; Gal-
r 2006). The peculiar ring-like distribution of H I around
a lagher et al. 2001). In each case, they represent young
NGC 1533 as seen in Figure 1 (and figure 2 of Ryan-
stellarassociationsformingostensiblywherenostarsex-
Weber et al. 2004 with higher H I cut-off levels) may be
isted previously. Intergalactic H II regions are distinct
theresultofaninteractionwiththenearbydwarfgalaxy
from the H II regions in the outer arms of spiral galax-
seenin the NW corner,IC2038(with the SINGG identi-
ies (e.g., Ferguson et al. 1998) in that they are further
ficationJ0409-56:S2). No obviousoptical counterpartto
fromthehostgalaxy,theydonotshowsignificantcontin-
thedisturbedHIappearsineithertheDSSimageorthe
uumemission,andthushavemuchhigherHαequivalent
SINGG image. The velocities of the three spectroscop-
widths (Ryan-Weber et al. 2004; Gerhard et al. 2002;
ically confirmed regions coincide well with the H I gas,
Mendes de Oliveira et al. 2004). These regions may rep-
which is bound to NGC 1533 (Ryan-Weber et al. 2004).
resent Hα-emitting counterparts to the extended UV-
The projected radius from the center of NGC 1533 to
disk population of star clusters revealed by GALEX,
the isolated H II regions ranges from 19 kpc (region 5)
1DepartmentofAstronomy,UniversityofMichigan,500Church to31 kpc (regions1and2), upto fourtimes the R-band
St.,AnnArbor,MI48109, [email protected] 25th magnitude isophotal radius of NGC 1533. Table 1
2Department of Physics and Astronomy, The Johns Hopkins summarizes the properties of the three H II regions pre-
University,Baltimore,MD21218-2686
3Institute of Astronomy, University of Cambridge, Madingley sented in Ryan-Weber et al. (2004) and imaged here. It
Road,Cambridge,CB30HA,UK gives the regions’ positions, projected radii from NGC
4ResearchSchoolofAstronomyandAstrophysics(RSAA),Aus- 1533,velocitiesdeterminedfromtheirHαemissionlines,
tralian National University, Cotter Road, Weston Creek, ACT Hα luminosities, and H I columns at each region’s posi-
2611,Australia
tion. NotethatthevaluesinourTable1differfromthose
2
Fig. 1.—TheHαSINGGimageofNGC1533overlaidwithAustraliaTelescopeCompactArray(ATCA)HIcontoursandmarkedwith
the locations of the individual spectroscopically confirmed H II regions studied here. The detection images are shown to the right of the
SINGG image. The ATCA HI contours are1.0(inbold; 3σ), 1.5, 2.0, 2.5, and 3.0×1020 cm−2 and have aresolutionof ∼1arcminute.
Thereisnodetectable HIemissionontheopticalcoreofNGC1533.
presentedinRyan-Weber et al.(2004)becauseofthe re- and the potential growth of the stellar associations into
centrevisioninthemeasurementofthedistancetoNGC a tidal dwarf galaxy.
1533 using images taken in parallel to those discussed Independent of the gaseous properties of cluster envi-
here (see Barber DeGraaff et al. 2007). In this paper, ronments, star-cluster “infant mortality” suggests that
wepresentHubbleSpace TelescopeAdvancedCamerafor the isolated, young star clusters presented here may be
Surveys/ High Resolution Channel (ACS/HRC) images short-lived (Rafelski & Zaritsky 2005; Fall et al. 2005;
andphotometryofthethreespectroscopically-confirmed Chandar et al. 2006). Will supernova and mass loss in
isolated H II regions associated with NGC 1533. These these clustersremoveenoughinterstellarmatter to leave
observations enable us to resolve the star-forming re- thestarsfreelyexpanding,resultingintheclusters’rapid
gions, identify their ionizing sources, and carry out an dissolution? StudiesoftheAntennaegalaxies(Fall et al.
in-depth study of their stellar populations. 2005; Chandar et al. 2006) and the Small Magellanic
In combination with the HST data, we make use Cloud (Rafelski & Zaritsky 2005) purport that over half
of an ATCA H I synthesis map of NGC 1533 of clusters formed dissolve less than 10 Myr after their
(Ryan-Weber et al.2003b)to helpconstrainsomeofthe formation independent of their initial masses. How-
detailsoftheformationandevolutionoftheisolatedHII ever, a recent study by Gieles et al. (2007) concludes
regions. Intergalactic H II regions are often found form- that there is little or no “infant mortality” for ages be-
ing in relatively low column density H I gas. There is yond 10-20 Myrs in the Small Magellanic Cloud, and
no H I detected to a limit of 2 ×1019 cm−2 at the lo- de Grijs & Goodwin (2007) find less than 30% of young
cation of the H II region near NGC 4388 (Oosterloo & (< 20 Myrs) SMC clusters undergo “infant mortality.”
van Gorkom, 2005), and the H I surface density in the Generally, studies of this kind require a statistical sam-
vicinity of the three confirmed systems, regions num- ple of clusters. Instead, resolving a few star clusters in
bered 1, 2, and 5 in Figure 1 is 1.5−2.5×1020 cm−2 an isolated environment in a period during which they
(Ryan-Weber et al. 2004). Furthermore, the H I peaks may be actively dissolving offers a different approach to
intheneutralgasringsurroundingNGC1533andinthe examining their fates.
stripped material around NGC 4388 do not exhibit any The photometric properties of the underlying stars
detectable level of star formation. These observations in these intergalactic H II regions can help to answer
offer further support for the lack of correlation between some of the questions surrounding their formation and
starformationratedensitiesandHIsurfacedensitieson evolution. Previous studies have used stellar popula-
localscalesfoundbyKennicutt et al.(2007),unlikewhat tion synthesis models and evolutionary tracks to infer
is seen on global galactic scales. The overall kinematics cluster ages and masses. For example, Whitmore et al.
of the gas in the region of the intergalactic H II regions (1999) identify 3 distinct populations of star clusters in
may provide clues as to the trigger of the star formation and around NGC 4038/4039 (Antennae Galaxy) using
Isolated OB Associations 3
TABLE 1
Propertiesof theConfirmedIsolated HII Regionsin NGC1533a
HII RA Dec Separation Velocity LogLHα b NH I
Region kpc kms−1 ergss−1 1020cm−2
1 041013.7 -561137.5 31 846±50 37.67±0.26 2.4
2 041014.5 -561135.9 31 831±50 37.50±0.26 2.4
5 041015.7 -560616.2 19 901±50 37.35±0.26 1.5
aSummaryofpropertiesoriginallypresentedinRyan-Weberetal. (2004): Positionsaregiven
in J2000 coordinates, with RA in hours, minutes, and seconds, and Declination in degrees,
arcminutes,andarcseconds. Separationistheprojectedseparationinkiloparsecstotheoptical
centerofNGC1533. TheHIcolumndensitiesaremeasuredfromATCAmaps,attheposition
ofeachHIIregion. NGC1533hasaheliocentricvelocitymeasuredbyHIPASStobe785km
s−1. ThenumbersforprojectedseparationandHαluminositydifferfromthosepresentedin
Ryan-Weberetal.(2004)becauseweusethemostrecentmeasurementofthedistancetoNGC
1533,19.4Mpc(Blakeslee,privatecommunication).bAsmeasuredintheSINGGHαimages
their integrated photometric properties. Gerhard et al. each pointing at the optimal pixel scale of ACS/HRC,
(2002) determine the mass and age of an isolated H II 0′.′025 pixel−1. We create a detection image by sum-
region in the Virgo Cluster from its measured Hα and ming the images in the three filters. Due to the paucity
V band luminosities. Thus far, studies of these sorts of known point sources in our HRC images, we mea-
of objects have focused on what the star clusters’ inte- sured the PSF from HRC images taken in the same
grated light reveals about their formation histories, and cycle and in the same filters of the center of 47 Tuc,
have been unable to resolve the clusters into individual using at least 50 stars per image. The point sources
stars or stellar components. With the discovery of more in each image have an average Gaussian FWHM of 2.6
and more young, faint clusters in tidal tails of merging pixels, independent of the 3 filters investigated. At the
galaxies,intraclusterspaceandtheintergalacticmedium, adopteddistance to NGC 1533,assumedto be 19.4Mpc
there arises a need for information about the nature of (Barber DeGraaff et al. 2007), this FWHM corresponds
the forming stars. Yet, there is an unavoidable, inher- to a physical size of 6.1 pc.
ent uncertainty in the properties of lower mass clusters Figure 1 shows the ACS/HRC detection images of the
derived from models that assume a fully populated IMF OBassociationspoweringthe isolatedHIIregionsalong
(Cervin˜o & Luridiana 2004). Resolving the star clusters with their locations in the R-band continuum SINGG
presents an opportunity to examine their populations in image of NGC 1533 (Meurer et al. 2006); we show the
greater detail. H I distribution (Ryan-Weber et al. 2003b) with con-
Weorganizethispaperasfollows: Section2containsa tours. The half-light radii of associations 1, 2, and 5
description of our observations and measurements; Sec- are7.9,9.0,and6.2pixels respectively,correspondingto
tion 3 presents the results and model comparisons; Sec- physicalsizesof18.5,21.1,and14.5pc. Ata distance of
tion4.1explorescomparisonsbetweentheseintergalactic 19.4 Mpc, 1 arcsecond corresponds to 94 parsecs. Three
H II regions and other extragalactic and Galactic young tonineclumps,mostcloseinsizetotheimagePSF,com-
stellar populations; in Sections 4.2 and 4.3, we discuss prise each association. While multiple sources bright in
thegaspropertiesinthevicinityoftheintergalacticHII UVandVsurroundassociations1and2,nosuchobjects
regionsand examine the likely fates of the stellar associ- appear within the same proximity to association 5.
ations;andSection5brieflysummarizesourresults. The
Appendix containsthe new calculationofthe conversion 2.2. Source Detection and Definitions
factorbetweentheionizingluminosityandthetotalnum-
ber of main sequence O stars contained within an H II We catalogued the sources in our images by applying
region following the methodology of Vacca (1994). the SExtractor code (Bertin & Arnouts 1996), as imple-
mented in the Apsis pipeline, to the relevant detection
2. OBSERVATIONSANDMEASUREMENTS images. Weeliminatedspurioussources(e.g. singlepixel
detections, image edge artifacts) and objects detected
2.1. HST Observations
in only one band by inspecting the SExtractor detec-
HST ACS/HRC observations were carried out in Oc- tions by-eye. Because the bulk of the following analysis
tober 2004. The targets were observed in three filters: requires color measurements, single-band detections (4
F250W (5808 s); F555W (2860 s); and F814W (2892 s), sources) are not of particular usefulness to us, though
covering a wide wavelength range chosen to sample the they may indeed be real. Of the initial 30 sources iden-
shapeofthe ultravioletto opticalSED.Throughoutthis tifiedbySExtractorintheimagecontainingassociations
paper,wewillrefertotheF250WfilterasUV,F555Was 1 and 2, we include 19 in this analysis. Of the initial
V,andF814WasIintheinterestofbrevityandsimplic- 36sourcesidentifiedbySExtractorintheimagecontain-
ity. Eachorbitwassplitintotwoexposuresseparatedby ing association 5, 11 appeared to be genuine detections,
asmalldither,resultingin5σlimitingABmagnitudesof including one single-band detection (F814W) of an ob-
28.0inUV,28.9inV,and29.0inI.Theimageswerepro- vious background galaxy. Figure 1 shows the HRC de-
cessedwiththeCALACScalibrationpipeline inorderto tection images containing the labelled isolated associa-
remove the instrumental signature, and then combined tions. SExtractor split association 1 into three separate
using the Apsis pipeline (Blakeslee et al. 2003) resulting objects, and detected associations 2 and 5 as single ob-
in cleaned and combined single images in each filter at jects. Further examination revealed each association to
4
Fig. 2.— Cut-out images in each HRC filter of the three isolated associations and their components (which we refer to as “clumps”),
circledandlabeled. Theimagescaleisshowninthetopleftcorner. F250WisaUVbandpass,whileF555WandF814Wcloselycorrespond
toVandIbands,respectively. WepresentadetaileddescriptionofsourcedetectionanddefinitioninSection2.2.
be composed of multiple clumps insufficiently deblended exclude them from most of this analysis. We address
by SExtractor. Within the associations, we separated backgroundgalaxy contaminationof the rest of the field
the components using IRAF contour plots with levels of objectsinSection2.4. Theimagesoftheobjectsincluded
2σ, 4σ, 6σ and 8σ, defining an individual component as inthisstudyarepresentedinFigures1-4. Figure1gives
having at least two contours, corresponding to at least an idea of the environment containing the intergalactic
4σ above the background. These criteria resulted in six associations. All three lie in a low column density H I
components in association1, nine components in associ- ring,withprojectedradii19-31kpcfromthenucleusof
ation 2, and three components in association 5 (see Fig- NGC 1533. Figure 2 displays close-ups of the clumps in
ure2). Finally,usingtheIRAFtaskIMEXAM,wefitted the three bands, UV (left column), V (center column),
Gaussian distributions to all field objects and clumps in and I (right column), circled and labeled. The letters in
order to determine their FWHMs. theimagescorrespondtothe namesinTable2. Figure3
Therearethree maintypes ofobjectsto whichwe will showsthetwoHRCfields(inthethreecombinedfilters),
referthroughoutthispaper: thestellarassociationspow- with labels next to each object. Finally, Figure 4 shows
ering the total H II regions (≡associations), the compo- three-color image close-ups of the associations, clumps,
nentsoftheassociations(≡clumps),andthefieldsources andfieldobjectsinthetwosetsofHRCimages. Inthese
in the vicinity of the associations (≡ field objects). Fur- color close-ups, blue, green, and red represent the UV,
thermore, we assume that all of the extended and red V, and I bands, respectively.
field objects in the vicinity of the associations are back-
ground galaxies (see Section 2.4 and Table 4), and thus 2.3. Aperture Photometry
Isolated OB Associations 5
Fig. 3.—HRCdetectionimagesshowthefullfieldscontainingtheisolatedstellarassociations1,2,and5. The30detectedfieldobjects
arelabeledaccordingtotheirnamesinTables2,3and4.
6
Fig. 4.— Three-color images of associations and field objects where blue, green and red represent UV (F250W), V (F555W), and I
(F814W), respectively. Theobjects arelabeledwithnamescorrespondingtothosegiveninTables2,3,and4.
Isolated OB Associations 7
TABLE 2
Photometry Results and Sizes for Isolated Associations and Their
Components
Assn Fλ (F250W) V-Ia UV-V mV MV FWHM
10−18 cgsb Vega AB pc
1 Total 8.13±0.78 -0.12+.10 -0.21+.11 23.38+.04 -8.06 61
−.10 −.12 −.04
a 0.24±0.04 -0.69+.45 -0.86+.25 27.85+.17 -3.59 3.39
−.52 −.30 −.20
b 0.52±0.07 -0.22+.18 -0.48+.16 26.64+.09 -4.80 4.91
−.21 −.18 −.10
c 0.18±0.04 -0.33+.27 -0.18+.27 27.51+.13 -3.93 5.48
−.35 −.34 −.15
d 2.17±0.14 -0.46+.05 0.33+.07 24.27+.02 -7.17 9.40
−.05 −.07 −.02
e 0.21±0.04 -0.37+.20 0.07+.22 27.06+.09 -4.38 4.00
−.23 −.27 −.10
f 1.95±0.09 -0.45+.10 -0.66+.06 25.38+.04 -6.06 4.98
−.10 −.06 −.04
DLC 2.63±0.78 -0.25+.19 0.33+.29 24.04+.09 -7.40 –
−.22 −.39 −.10
2 Total 11.2±0.77 0.08+.09 -0.67+.08 23.51+.04 -7.99 67
−.10 −.09 −.05
a 0.26±0.04 -0.02+.22 -0.54+.21 27.47+.12 -3.97 4.44
−.27 −.24 −.14
b 1.18±0.07 -0.31+.09 -0.47+.07 25.73+.04 -5.71 9.66
−.10 −.08 −.04
c 1.40±0.07 -0.16+.08 -0.67+.06 25.75+.04 -5.69 8.67
−.09 −.07 −.04
d 0.62±0.05 -0.86+.23 -0.83+.11 26.80+.07 -4.64 5.31
−.28 −.12 −.08
e 1.25±0.07 -0.25+.09 -0.62+.07 25.82+.04 -5.62 9.60
−.10 −.08 −.05
f 0.93±0.05 -0.45+.11 -0.74+.08 26.27+.05 -5.17 <4.70
−.12 −.08 −.05
g 0.85±0.07 -0.17+.13 -0.64+.10 26.26+.06 -5.18 4.83
−.15 −.11 −.07
h 0.17±0.04 -0.09+.23 -0.11+.27 27.47+.12 -3.97 <4.70
−.28 −.34 −.13
i 0.58±0.07 0.19+.20 -1.03+.17 27.06+.12 -4.38 5.03
−.23 −.18 −.13
DLC 3.96±0.77 0.17+.17 -0.26+.20 24.19+.10 -7.25 –
−.19 −.24 −.11
5 Total 3.51±0.66 -0.22+.15 0.02+.20 24.06+.06 -7.38 40
−.17 −.23 −.06
a 0.15±0.05 0.46+.29 -0.55+.36 28.05+.20 -3.39 4.46
−.36 −.48 −.25
b 0.61±0.07 -0.45+.12 0.02+.13 25.96+.05 -5.48 <4.70
−.13 −.15 −.06
c 1.57±0.10 -0.54+.10 -0.33+.08 25.28+.04 -6.16 9.54
−.11 −.08 −.04
DLC 1.33±0.66 -0.54+.25 0.74+.44 24.37+.10 -7.07 –
−.32 −.75 −.11
aV-IcolorshavebeentransformedtoVegaMagnitudesinthestandardJohnson-Cousins
filter system using the prescription of Sirianni et al. (2005). V band magnitudes are
technicallyin units of Vegamagnitudes, although magnitude systems converge in the V
band,makingthegivenvaluesroughlycorrectfortheABmagnitudesystemaswell. The
UV - V colors are left in AB magnitudes, as there is no calibration for the conversion
of the F250W filter to Vega magnitudes. Absolute magnitudes were calculated using a
distance modulus of 31.44 (Blakeslee, private communication). For each association, V
bandmagnitudeswerecorrectedfornebularemission([OIII]λλ4959,5007).bergss−1cm−2
˚A−1
We performed circular aperture photometry on each A correction for foreground extinction, based on the
source using the IDL routine APER. Aperture sizes for Schlegel et al. (1998) extinction maps, (foregroundE(B-
eachassociationandfield object weredetermined by ex- V) = 0.016), was applied according to the parameteri-
amining where the curve of growth (in flux vs. aperture zation of Cardelli et al. (1989), including the update for
radius) began to flatten, whereas aperture sizes for the the near-UVgivenby O’Donnell(1994). Giventhe loca-
clumpsweredeterminedbyeye. Foreachassociationand tion of the isolated associations far outside the optically
field object, sky annuli were positioned approximately 1 luminous area of NGC 1533,we excluded any correction
arcsecond away from each of the objects’ centers. Be- for the internal extinction. Additionally, fluxes for the
cause the clumps are so closely spaced in the image, we associations were corrected for the strong emission lines
subtractedtheskymeasurementoutsidethecorrespond- [OIII]λλ4959,5007present in the F555Wbandpass. We
ing associationas a best estimate. As evidenced by Fig- used the STSDAS package SYNPHOT to simulate pho-
ure 2, the circular apertures containing the clumps do tometryusinganactualspectrumofassociation1(to be
notaccountforallthe lightinthe association. We chose presented in a future paper) both with and without the
the aperture sizes to contain the maximum amount of nebular emission lines. The procedure for this correc-
emissionwhile notoverlappingwith otherapertures. By tion is as follows: (1) We measured the Hα flux for all
masking out the apertures containing the clumps, and threeHIIregionsintheNGC1533SINGGnarrow-band,
running APER on what remained, we determined fluxes continuum-subtracted image; (2) We measured the Hα
for the diffuse light components (DLC) of each associa- emission line flux in the spectrum of association 1; (3)
tion. This component simply represents the light losses We divide quantity (1) by quantity (2) for each associa-
resultantfromthe small apertures,andthus may not be tiontoobtainaflux correctionfactorforeachindividual
genuine diffuse emission. Thus, the DLC is essentially association;(4) Using SYNPHOT, we measure the total
thedifference betweenthetotalassociationflux,andthe flux of association1’s spectrum containedin the F555W
sum of the fluxes of the individual clumps. bandpass;(5)We subtractthe emissionlinesinthisarea
8
of spectrum, and remeasure the total flux in the same Hubble Deep Field galaxy number densities, we ex-
area with SYNPHOT; (6) We subtract (5) from (4) to pect 1-2 × 105 galaxies per square degree in WFPC2,
get an uncalibrated flux of just the emission lines in the corresponding to a limiting I-band magnitude of 25-
F555Wbandpass; (7) Finally, we calibrate and scale the 26 (Casertano et al. 2000). Field galaxy number den-
totalnebularemission-linefluxfrom(6)bytheindividual sities are lower in the far-UV, roughly 3 × 104 galaxies
correction factor determined in (3). per square degree to a limiting AB magnitude of 26.5
These emission-line corrections were 0.26, 0.17, and (Teplitz et al. 2006). Considering these number densi-
0.27 AB magnitudes (17%−28% of the total flux) for ties, we would then expect 5-10 redbackgroundgalaxies
associations 1, 2 and 5, respectively. We note that these to appear in our 29′′×26′′ field of view, and 0-1 galax-
emission line corrections assume that the flux ratios of ies with emission in the UV. We detect and analyze 11
Hαto [OIII]emissionlinesarethe sameforeachassoci- objects in the vicinity of association 5, none of which
ation. Theerrorduetopotentialreasonablevariationsin appear in the UV images, and the majority of which
theseline ratiosis negligiblewhencomparedtothe total appear quite red in V-I. Of the 11 objects in the vicin-
corrections. Furthermore, this correction has the effect ity of association 5, six are very obviously extended or
that the associations are redder in V-I after correction elliptical red sources,with V-I colors(Vega magnitudes)
than one would expect based on the V-I colors of the between0.7and1.8(aspreviouslymentioned,oneobject
individual clumps, which were not corrected for nebular is detected only in F814W). Three of the remaining five
emission due to the limited (∼1.5 arcsecond) resolution compactsourceshaveV-Icolorsinasimilarlyredrange,
of the SINGG Hα images. while two exhibit fairly blue V-I colors, 0.04 and 0.19.
Finally,forcomparisonwithpopulationsynthesismod- The lack of UV emission from these sources combined
els and previous work, we transformed the ST magni- withthenumberbeingconsistentwithourpredictionfor
tudes to the Vega magnitude system, and the filters background galaxy detections leads us to classify all 11
to the standard Johnson-Cousins system following the as backgroundemitters.
method outlined by Sirianni et al. (2005) for both the We detect and analyze19 objects in the vicinity of as-
F555W and F814W data. Because magnitude systems sociations 1 and 2. In contrast to the association 5 field
convergein the V band,andbecause the F555Wfilter is objects, many of these field objects appear bright in UV
quite similar to the Johnson-Cousins V band, its trans- andV,suggestingthatthestarformationintheSEpart
formation produces virtually no change. Furthermore, of NGC 1533’s H I ring extends beyond the initially de-
the F250W filter has no Johnson-Cousins counterpart, tected isolated H II regions. Furthermore, most of the
and it is impractical to convert AB magnitudes to Vega surrounding blue field objects lie clustered within 500
magnitudes in this part of the spectrum. Throughout parsecsofthenearestassociation(1or2),whilethelikely
thispaper,wewillquoteV-IcolorsintermsofVegamag- background galaxies are spread more evenly throughout
nitudes in the standard Johnson-Cousins filters. This the image. While we cannot say for certain which of the
magnitude system allows for the greatest ease in com- fieldobjectsaroundassociations1and2arebackground
paring our results with both previous work and existing galaxies,wecanpredictthatthe 10sourcesnotdetected
population synthesis models. However, UV-V and UV-I in F250W, 5 of which have V-I colors greater than 1.0,
colors and UV magnitudes will remain in the HRC fil- aremorelikelytobedistantbackgroundgalaxies. These
ter bandpasses and in AB magnitudes. The faintness of sourcesalsoappeartobe moreextended. Moreover,this
the clumps andfield objectscombinedwith considerable numberofsourcesisconsistentwiththepredictionbased
pixel-pixel noise in the read noise limit produces errors on the background galaxy density of deep WFPC2 im-
in the photometry as high as 0.25 magnitudes. The re- ages.
sulting Vega magnitudes are presented in Tables 2, 3, Wedonote,however,thatweareunabletodistinguish
and4aswellasthe V-IandUV-Vcolors,alongwiththe extendedlikelybackgroundgalaxiesfrompotentiallyold
measured FWHM of each source. (∼ 1 Gyr), extended open clusters, which can have V-I
colors up to 1.4 and absolute magnitudes ranging from
2.4. Source Properties and Contamination by
-1 to -5.1 in the V band (Lata et al. 2002). Nor can we
Background Galaxies
differentiate unresolved likely background galaxies from
Our sample of field objects and clumps totals 48 indi- globularclusters,whichhaveV-Icolorsrangingfrom0.85
vidual sources altogether, 34 of which are located in the to 2.65, and absolute V-band magnitudes from -1.5 to -
3 kpc × 3 kpc field around associations 1 and 2. We 9.5(Harris 1996). Still, suchold, extended open clusters
estimate that the associations span 40 - 75 parsecs in are rare in the Milky Way (e.g. Lata et al. 2002; Lada
their full visible extent, similar to large Milky Way stel- & Lada 2003), and globular clusters tend to lie closer to
larassociations. Figure3givesthelocationofallobjects the galaxy center (within ∼ 20 kpc) than the star for-
identifiedinthetwoHRCimages,withthefieldcontain- mation seen here (Harris 1996), although a few are as
ing associations 1 and 2 on the top frame, and that of far as 100 kpc from Galactic center. Still another possi-
association5 on the bottom frame. Of the 48 individual bility remainsthatsomeofthese objectscouldrepresent
sources,only 17 are resolved. 9 of the 11 field objects in isolated red supergiants.
theassociation5fieldarewell-resolvedextendedsources. Tables3and4presentthe photometricpropertiesand
The V-I colors of the clumps range from -0.7 to 0.2 in opticalsizesforalldetectedfieldobjects. Inthesetables,
V-I,andthoseofthe fieldsourcesrangefrom-0.9to 1.8. we have identified those field objects which are the best
IfallthesourceswereatthesamedistanceasNGC1533, candidates for distant background galaxies (extended,
the absoluteV-bandmagnitudesrangefrom-3.4to-7.3. red, and/or no UV-emission) as F˙, and those blue field
Almost certainly,some ofthe field objects in the HRC objects which are most likely to be associated with the
images represent distant background galaxies. From
Isolated OB Associations 9
ongoing star-formation around NGC 1533 as F. Figure
3 shows three-color (RGB) images of the field objects
and background galaxies. The top panel shows the field
objects surrounding associations 1 and 2, and the bot-
tompaneldisplaysthefieldobjectsofassociation5. The
labels on each image correspond to the names listed in
Tables 3 and 4 (“Assn” for the association; “F” for field
object, and “F˙” for likely background galaxy). We do
includealloftheseobjectsinthe subsequentanalysisfor
completeness’ sake (except where their red colors push
them outside of adopted plotting ranges), and we urge
the reader to consider the 10 “dotted” objects in Table
3 and the 11 objects in Table 4 (also identified in the
figures) as potential background galaxies.
InFigures5and6,wecomparetheindividualsources’
absolute magnitudes and V-I and UV-V colors with Fig. 5.— Color-magnitude diagram for the blue field objects
and clumps, with V-I in Vega magnitudes. The dashed and solid
those of single star MK spectral types and a 50 M⊙ lines shown are calibrations of MK spectral types from Schmidt-
instantaneous-burstStarburst99population(see3.1). In Kaler(1982). ThedottedlinerepresentstheV-Icolorevolutionof
orderto plot MK spectraltypes in MV versusUV-V, we aStarburst99 50M⊙ population withZ=0.4Z⊙. Thepopulation
convertedtheirU-bandabsoluteVegamagnitudestothe synthesis model was runfor aninstantaneous burst Salpeter IMF
F250W filter in AB magnitudes using the SYNPHOT. (α = 2.35, Mup = 100, Mlow = 0.5), and employs Padova group
isochrones.
To arrive at Starburst99 UV-V colors we calculated the
F250W AB magnitude using the output synthetic spec-
trum along with SYNPHOT. In many cases, the colors
andabsolutemagnitudesoftheobjectsarewell-matched
bysingleOorBstars. Thissimilaritydoesnotprovethat
field objectsand clumps are single stars,but thatO and
Bstarsdominatethe radiationoutputinmanycases. In
other cases, the expected magnitude-colorevolution of a
Starburst99 population (run for an instantaneous burst
Salpeter IMF withα = 2.35, M = 100, M = 0.5,
up low
Padova group isochrones, and Z=0.4Z⊙) with an initial
mass of 50 M⊙ matches the data. We note that a 50
M⊙ population with a 100 M⊙ upper mass limit is un-
physical, and results in fractional stars. We have simply
usedaninitialstarburstmassof 50M⊙ to appropriately
scale the absolute magnitude and illustrate a degener-
acy. The colors are unaffected by the mass of the initial
Fig. 6.— Color-magnitude diagram for the field objects and
starburst. Wediscussthe implicationsoffractionalstars clumps,withUV-VinABmagnitudes. Thedashedandsolidlines
and stochastic IMFs in section 3.3. For completeness, shown are calibrations of MK spectral types from Schmidt-Kaler
we have included the bluest (in V-I) background galaxy (1982). ThedottedlinerepresentstheUV-Vcolorevolutionofthe
sameStarburst99populationshowninFigure5.
candidates inFigure 5,using NGC 1533’sdistance mod-
the stellar populations, but does not account for some
ulustocalculatetheirabsolutemagnitudes. Thereddest
of the details of stellar evolution. In section 3.3 we dis-
candidates do not lie within the ascribed color range.
cuss the limited applicability of bothmodels to ourdata
3. STELLARPOPULATIONSOFASSOCIATIONS given the low luminosities of the isolated associations,
Ideally,stellarpopulationsynthesismodels,alongwith and masses below what is required for a fully-populated
integrated colors and luminosities, can provide a solu- IMF.
tion for the masses and ages of the stars populating
3.1. Comparison with Starburst99 Model
the clusters. These models aim to account for most of
the known stellar evolutionary processes, and integrate To accurately designate a metallicity for the Star-
over various sets of assumed initial conditions. In the burst99 model, we used an optical spectrum of associ-
following analysis, we describe two approaches we took ation1,estimating itsmetallicity tobe roughly∼0.4Z⊙
in an effort to constrain the stellar populations of the (Werk et al., in preparation). Figure 7 plots various di-
isolated associations. We compare the associations’ col- agnostics from ages 1 - 16 Myr for a Starburst99 syn-
ors and luminosities to output from Starburst99 models thetic population for Z= 0.4Z⊙, an instantaneous burst
(Leitherer et al. 1999), and we perform a simple, inde- of 106M⊙, and a Salpeter IMF with a lower limit of 0.5
pendentcalculationusingthe associations’totalUVand M⊙ and an upper limit of 100 M⊙. It also shows the
Hα luminosities. Both approaches employ single-age, correspondingvalues calculatedfor associations1,2 and
single-metallicitymodelswithaSalpeterIMF.Whilethe 5. The error bars indicate photometric errors.
Starburst99 model accounts for details of stellar evolu- In Figure 7a, we show the temporal evolution of
tion, it cannot account for anything but a scaled-down the ratio Log (F /F ) for the Starburst99 model
Hα UV
IMF over the full range of stellar initial masses. Our in- and the three associations. Hα and UV luminosities
dependent calculation is more quantized with regard to serveas reasonablestar formationindicators (Kennicutt
10
TABLE 3
Photometry Results and Sizes for Association 1 & 2 Field Objects
ID R.A. Dec. Fλ (F250W) V-Ia UV-V mV MV FWHM
hms dms 10−18 cgsb Vega AB pc
F˙1 041014.54 -561144.84 – 0.05+.23 – 25.52+.14 -5.92 11.38
−.26 −.16
F˙2 041015.51 -561143.35 – 1.38+.32 – 26.63+.28 -4.81 7.81
−.43 −.39
F3 041012.73 -561145.58 1.36±0.14 -0.23+.04 1.17+.19 23.92+.03 -7.52 5.91
−.04 −.23 −.03
F˙4 041014.64 -561139.58 – 1.01+.17 – 26.55+.14 -4.89 7.88
−.19 −.16
F5 041015.10 -561137.89 1.02±0.13 -0.40+.34 -0.95+.20 26.35+.15 -5.09 5.60
−.45 −.23 −.17
F6 041014.82 -561136.00 0.90±0.14 -0.61+.36 -0.74+.22 26.27+.15 -5.17 5.95
−.51 −.25 −.17
F˙7 041014.24 -561136.35 – 2.03+.10 – 25.24+.10 -6.20 6.02
−.11 −.11
F8 041014.80 -561135.18 0.56±0.11 – -1.07+.25 27.12+.17 -4.32 3.97
−.30 −.20
F9 041014.89 -561134.69 0.75±0.13 0.34+.21 -0.67+.23 26.41+.15 -5.03 5.62
−.25 −.26 −.17
F10 041013.55 -561136.42 1.59±0.14 -0.27+.18 -0.86+.13 25.78+.09 -5.66 7.15
−.21 −.14 −.10
F11 041014.74 -561134.77 0.35±0.14 -0.10+.07 2.04+.36 24.56+.04 -6.88 5.48
−.08 −.55 −.04
F˙12 041014.41 -561133.89 – 0.79+.24 – 26.95+.20 -4.49 5.34
−.30 −.25
F˙13 041013.74 -561134.02 – 1.49+.27 – 26.94+.25 -4.50 5.64
−.34 −.32
F14 041013.70 -561133.91 0.68±0.17 -0.89+.52 -0.44+.31 26.28+.19 -5.16 4.66
−.93 −.39 −.23
F˙15 041014.05 -561132.99 – 1.36+.27 – 27.28+.25 -4.16 5.60
−.34 −.32
F˙16 041012.86 -561133.76 – 0.67+.29 – 26.61+.24 -4.83 8.02
−.36 −.30
F˙17 041013.49 -561131.89 – -0.32+.20 – 25.77+.09 -5.67 5.19
−.24 −.10
F18 041014.19 -561130.99 2.41±0.16 -0.27+.08 -0.19+.08 24.66+.04 -6.78 5.62
−.09 −.09 −.04
F˙19 041013.85 -561127.36 – -0.01+.27 – 26.92+.16 -4.52 5.78
−.33 −.19
a V-IcolorshavebeentransformedVegaMagnitudesinthestandardJohnson-Cousinsfiltersystemusingthe
prescriptionofSiriannietal. (2005). Absolutemagnitudesandphysicalsizeswerecalculatedusingadistance
modulusof31.44(D=19.4Mpc,Blakeslee,privatecommunication). b ergss−1 cm−2 ˚A−1
TABLE 4
Photometry Results and Sizes for Association 5 Field
Objects
ID R.A. Dec. V-I mv FWHMa
hms dms Vega pc
F˙20 041016.52 -560624.07 1.22+.28 25.27+.25 14.68
−.36 −.33
F˙21 041016.86 -560604.30 1.71+.22 25.64+.21 15.47
−.28 −.27
F˙22b 041016.00 -560611.62 – – 17.48
F˙23 041015.06 -560618.80 0.04+.16 26.00+.09 6.95
−.18 −.10
F˙24 041014.77 -560617.57 0.19+.16 25.99+.11 9.99
−.18 −.12
F˙25 041014.48 -560620.22 1.56+.21 24.30+.19 24.73
−.25 −.23
F˙26 041014.66 -560610.95 0.85+.07 24.80+.06 9.91
−.09 −.07
F˙27 041014.04 -560623.07 2.25+.28 27.14+.27 5.89
−.38 −.37
F˙28 041014.20 -560618.31 0.71+.11 24.81+.08 17.90
−.11 −.08
F˙29 041014.02 -560616.17 1.34+.13 26.09+.11 6.36
−.13 −.12
F˙30 041014.96 -560600.15 1.79+.32 24.83+.30 24.97
−.43 −.42
a Physical sizes are given in parsecs for straightforward comparison with
clumpsandotherfieldobjects. b WedonotdetectF˙22inanyfilterother
thanF814W,althoughwe consideritarealsource, withan I bandmag-
nitudeof23.14(Vegamagnitudes).
1998). While Hαtracesthe ionizingradiationfromstars The evolution of UV-V, UV-I, and V-I colors of the
more massive than 15M⊙, the UV luminosity originates same Starburst99 population is plotted in three color-
from stars more massive than 3M⊙. The ratio Log color diagrams for the same range in ages in Figures
(F /F ) decreases steeply with time, as massive ion- 7b, 7c, and 7d. The Starburst99 UV magnitudes in
Hα UV
izing stars die out, and late-type O stars and B stars the HRC F250W bandpass were generated by the SYN-
produce the majority of the UV flux. The best-match PHOTpackagefortheoutputStarburst99spectra. Seen
ages to the evolution of Log (F /F ) are 3.2 Myrs, in7b,7c,and7d,associations1,2,and5areconsistently
Hα UV
4.5 Myrs, and 3.0 Myrs for associations 1, 2, and 5 re- bluer than the model colors, except for association 2 in
spectively. In this plot, the horizontal error bars repre- V-I.Thus,wecannotconstraintheagesusingsimplythe
sent the minimum possible error given the photometric colorsoftheassociations. Aswewilldiscuss,theinability
errors. oftheStarburst99modeltomatchtheassociationcolors