The Initial Mass Functions of Four Embedded Stellar Clusters

0.31 MB·English

by  A. Leistra

Draftversion February5,2008
PreprinttypesetusingLATEXstyleemulateapjv.6/22/04
THE INITIAL MASS FUNCTIONS OF FOUR EMBEDDED STELLAR CLUSTERS
A. Leistra1, A. S. Cotera2, J. Liebert1
Draft versionFebruary 5, 2008
ABSTRACT
We present near-infrared J, H, and K images of four embedded stellar clusters in the Galaxy.
We find a significant fraction of pre-main-sequence stars present in at least one of the clusters. For
the clusters dominated by main-sequence stars, we determine the initial mass function (IMF) both
by using the K luminosity function and a global extinction correction and by deriving individual
extinction corrections for each star based on their placement in the K vs. H −K color-magnitude
6 diagram. BasedonourIMFs we finda significantdiscrepancybetweenthe meanIMFderivedvia the
0 different methods, suggesting that taking individual extinctions into accountis necessaryto correctly
0
derive the IMF for an embedded cluster.
2
Subject headings: open clusters and associations: general — stars: formation — stars: luminosity

n function, mass function
a
J
1. INTRODUCTION (Leistra et al. 2005). Claims that variations in the IMF
4
Embedded clusters are increasingly recognized as vi- exist,whether basedonindividualextreme clusters such

1 tal sites of star formation for both low- and high-mass as the Arches or a general analysis of the data (Scalo
v stars. Recent studies indicate that clusters may ac- 1998), must thus be handled with care to compare only
3 results based on similar methodology.
count for 70-90% of star formation and that embed-
8 The final release of the Two Micron All Sky Sur-
ded clusters (those still partially or fully enshrouded
0 vey (2MASS) has fostered studies (e.g. Dutra & Bica
in their natal molecular cloud) may exceed the num-
1 2000, 2001; Bica et al. 2003; Ivanov et al. 2002) which
ber of older, non-embedded open clusters by a fac-
0 can probe a much larger portion of the Galaxy for pre-
tor of ∼20 (Elmegreen et al. 2000; Lada & Lada 2003).
6
viously unknown embedded stellar clusters and signifi-
0 The stellar content of embedded clusters within well-
cantly increased the number of known embedded clus-
/ known star formation regions can now be probed, where
h high extinction (AV & 10) prohibits studies at opti- ters. A compilation of some of these results along
p with previously known embedded clusters is presented
cal wavelengths. The IMF of such clusters has gener-
- in Porras et al. (2003) who find that ∼80% of the stars
o ally been found to be consistent with a Salpeter value
intheirsamplearefoundin“largeclusters”ofmorethan
r with a slope of Γ = −1.35 (e.g Okumura et al. 2000;
st Blum, Damineli, & Conti 2001; Figuerˆedo et al. 2002) 100 stars, despite the rarity of such clusters. However,
thesestudiesarenotfoolproof,andcompilationsof“em-
a althoughoutliershavebeenfoundaswell,generallywith
: flatter slopes than the Salpeter value (e.g. Porras et al. bedded clusters”basedpurely onthe 2MASS data with-
v
out followup must be treated with caution. The studies
i 1999).
X Althoughnear-infrared(NIR) spectralclassificationof basedsolelyonstellardensitycriteria(e.g.Dutra & Bica
2000, 2001) have been found in followup work by dif-
r massive stars is possible (Hanson et al. 1996), in most
a cases determinations of the IMF from NIR data rely ferent groups (e.g. Dutra et al. 2003; Leistra et al. 2005;
Borissova et al. 2005) to have only about a 50% success
heavily, if not exclusively, on photometry and use spec-
rate toward the inner Galaxy where the stellar back-
troscopyonlytoobtainreliablemassesofthefewbright-
ground is high. We have performed an independent
est and most massive stars in a cluster if at all. Since
searchofthe2MASSarchive,searchingthePointSource
these results thus depend on stellar evolutionary models
Catalog for regions of higher stellar density than the
aswellasdetailsofthehandlingofextinction,thisraises
background(determinedlocallywithina5′radius)which
concerns about to what extent the IMF depends on the
are redder in H −K than the local field. This selects
methodologyemployed. Massey(2003)citestheexample
for embedded clusters, with the color criteria helping to
of NGC 6611, where two separate analyses of the same
eliminate chance superpositions and regions of low ex-
data (Hillenbrand et al. 1993; Massey et al. 1995) using
tinction. A largebackgroundradius and the use of color
different treatments of extinction produced IMFs differ-
selection are critical to the automated identification of
ing by more than the formal 1σ errors would suggest
embedded clusters, but even color selection can fail in
(Γ=−1.1±0.1andΓ=−0.7±0.2.) Similarly,the IMF
regions of high backgroundstellar density, where patchy
for the G305+0.2embedded clusterdiffers by morethan
extinction can mimic clusters.
the errors between the value derived from the K lumi-
In §2 we present the observations and data reduction
nosity function (KLF; Γ=−1.5±0.3) and that derived
for four embedded clusters found in the 2MASS Point
using the color-magnitude diagram (Γ = −0.98± 0.2)
Source Catalog, in §3 we present the K-band luminos-
1 Steward Observatory, University of Arizona, 933 N. Cherry ityfunctions (KLF)andinitialmassfunctions (IMF)for
Ave.,Tucson,AZ85721 the clusters, and in §3.5 we address the issue of system-
2 SETI Institute, 515 N. Whisman Road, Mountain View, CA aticdifferencesbetweendifferentmethodsofderivingthe
94043
IMF for embedded clusters, for our clusters as well as
Electronicaddress: [email protected],[email protected], [email protected]
2 Leistra et al.
IMFs from the literature. trend with location on the chip was observed in the cal-
ibration for any of the clusters, though the scatter be-
2. OBSERVATIONS& DATAREDUCTION
tweenthe PISCESand2MASSmagnitudesbecomessig-
We selected five young stellar cluster candidates from nificant in the outermost 15 ′′; we thus exclude these
the 2MASS Point Source Catalog based on color and sources from the analysis.
density criteria as described in (Leistra et al. 2005). We
selected regions with a higher stellar density than the 3. ANALYSIS
locally defined field with redder H −K color than the
3.1. Sh 2-217 Cluster
field to select embedded cluster candidates. The cluster
candidates were observed using the PISCES instrument WepresentaK-bandimageoftheclusternearSh2-217
(McCarthy et al. 2001) on the 6.5m MMT on Jan10-11, in Figure 1. The cluster is nearly circular in projection
2003. PISCES uses a 1024x1024 HAWAII array with a and is quite dense; even in the highest-resolution indi-
platescaleof0′.′185/pixelontheMMT,providinga3′×3′ vidual pointings we obtained it suffers from crowding in
field of view. Images of all cluster candidates were ob- thecentralregions. Theclusterextendsovermost,ifnot
tained in J, H, and K filters to a limiting magnitude of all, of the field of view. We present the K image rather
J =19.5,H =18.5,K =18. Fourofourcandidates(all than a color frame because the seeing was significantly
excepttheonenearSh2-217)wereindependentlyidenti- better at K.
fiedasclustercandidatesbyBica et al.(2003),whoused This cluster was analyzed in the NIR by
criteria based only on stellar density without color con- Deharveng et al. (2003), who discuss the large un-
siderations. Our deeper, higher-resolution images sug- certainties in the distance to Sh2-217. Based on the
gest that four of these candidates are genuine stellar Lyman continuum fluxes from the main exciting star of
clusters, while the fifth, near Sh 2-258, contains only a Sh2-217 (located several arcminutes outside our field of
few stars in the PISCES images and could be either a view; the cluster is located on the periphery of the H II
verysmallcluster ora chancesuperposition. We present region) they adopt a distance of 5.0±0.8 kpc., which is
our results for the four confirmed clusters in this paper. consistent with the kinematic distance to the associated
Seeing conditions when the images were obtained were molecular gas. The cluster is coincident with a peak in
variable, ranging from 0′.′5 to 1′.′1 at K. the 8 µm emission as measured by the MSX mission,
All images were reduced and combined using IRAF suggesting dust is still present in the cluster.
routines. The distortions of the PISCES camera were The K vs H −K color-magnitude diagram is shown
mapped by imaging the globular cluster NGC 4147 and in Figure 2. The stellar density (determined in K where
mapping the observed locations to the USNO-B known theseeingwasbest)doesnotplateauinthe3′×3′fieldof
coordinates,thenconstructingatransformationfunction view,suggestingthatthe true field-stardensity levelhas
using the IRAF task geomap and correcting the images notbeenreached,andclusterstarsarestillpresentoutto
using IRAF geotran. There are too few USNO-B stars the edges of the field. As a result, in order to correctfor
in the heavily extincted regions we observed to provide foreground contamination, we selected an adjacent field
suitabledistortioncorrectionsfromthefieldsthemselves. of the same size from 2MASS to use as a comparison
Theindividualimages,takenwithaspiralditherpattern, field. We assumedthe luminosity function ofthe field to
were then combined. We fit a PSF to each image using be the same as that of the outer portions of the cluster
the IRAF task psf, allowing the PSF to vary across the (excluding the inner regions to minimize the results of
field to compensate for residualdistortions. Photometry crowding and mass segregation) in order to extrapolate
was then done using IRAF-DAOPHOT. from the limiting magnitude of 2MASS to that of our
Photometric calibration was performed using the images. We then binned the field starsby K andH−K
2MASS magnitudes offield stars. Stars used for calibra- with a bin size of 0.5 magnitudes and randomly selected
tion were selected to be well-separated from other stars the appropriate number of stars for removal from each
and from nebulosity in the PISCES images to ensure binintheclusterregion. Thisissimilartotheprocedure
that they were uncontaminated in the lower-resolution employed by, among others, (Blum et al. 2000) (who do
2MASS images, and to have magnitudes bright enough notdescribe anextinctioncorrection;this lackofcorrec-
to have good photometry in 2MASS (K < 14). We tionisequivalenttoassumingacommonextinction)and
chosetousearelativelylargenumberofcalibrationstars Figuerˆedo et al. (2002), and is the method we employed
ratherthanselectingthefewmostisolatedstarstoreduce in (Leistra et al. 2005). The resulting statistically cor-
effects of potential variability and photometric outliers rected CMD is shown in Figure 3. A total of 62 stars
among the calibration stars. The scatter in the photo- were removedin this procedure,out of an initial total of
metriccalibrationderivedfromcomparisonto2MASSis 236. The fairly wide distribution in H −K for cluster
the dominant source of photometric error, contributing stars is likely due to a combination of factors, notably
two to three times the measurement errors as reported an actual spread due to differential extinction to differ-
by DAOPHOT. DAOPHOT errors were ∼ 0.03 magni- ent regions of the cluster and to the greater influence of
tudeswhilethecalibrationuncertaintieswere∼0.1mag- crowding in H (where the seeing was poorer). Although
nitudes. Quoted errors in the 2MASS photometry were we find that individually correcting extinctions gener-
negligible,withmoststarshavinganerrorof±0.003mag allyprovidesasuperiorestimationofthe IMFcompared
or less in all bands. Thus, the quoted error should be with using only the K data and a single extinction for
consideredanoverestimatewhenconsideringtherelative the cluster asa whole, in this situation the lowerquality
photometry of stars within either cluster; the calibra- (in particular the poorer seeing and consequently more
tion errors from comparison to the 2MASS photometry severecrowding)oftheH-banddataleadsusexpectthat
will shift allour measurements by the same amount. No the K luminosity function (KLF) will produce a better
Embedded Stellar Clusters: II 3
estimate of the IMF for this cluster than the CMD will. evenin K. To make this transformation,we firstcorrect
We have previously compared these two methods of de- the observed K for distance and extinction. Without
terminingtheIMFforembeddedclustersinLeistra et al. spectra, we cannot obtain a precise estimate for the ex-
(2005);inthatcase,wefoundtheygavedifferentresults, tinction; instead, we compute an averageextinction cor-
with the “CMD” method (which we anticipate will be rectionbasedontheobservedJ−H andH−K colorsof
morereliableinmostcases,especiallywherevariableex- the brighter stars. Since there is little difference in the
tinction is present) yielding a flatter slope. intrinsic H −K color of stars of spectral type F5 and
earlier,this estimate of an averageextinction is not sen-
3.1.1. The KLF sitivetominorerrorsinthedistanceestimate. Usingthe
stellar evolutionary models of Meynet & Maeder (2003)
In order to obtain a robust determination of the KLF
for solar metallicity, we relate the mass for each track
and the IMF, we need to determine the completeness of
to an absolute K magnitude for a star on the ZAMS.
our data. To do this, we performed artificial star tests.
We transformed Lbol to K using the bolometric correc-
We inserted five artificial stars, each of the same magni-
tionsfromVacca, Garmany, & Shull (1996)for the early
tude, at a time into the cluster region, then ran IRAF-
spectraltypesandMalagnini et al.(1986)forlaterspec-
DAOPHOTwiththesameparametersasweusedforthe
tral types. We then use the intrinsic V −K colors from
initialanalysis. Thisprocedurewasrepeated50timesfor
Bessell & Brett (1988) for A-M stars and from Wegner
each magnitude bin (∆m = 0.5), for a total of 250 arti-
(1994) for O andB stars. Finally we interpolate linearly
ficial stars added in each bin in H and in K. The stars
between the masses available on the evolutionary tracks
were added in small numbers at a time to avoid hav-
to find the masses corresponding to our magnitude bins,
ing the artificial stars significantly change the crowding
and fit a power law to the resulting mass function. The
characteristicsandthus influence the completeness. The
IMF slope we derive by this method is Γ=−2.7±0.25,
artificial star tests indicate a high level of completeness
excluding bins corresponding to K > 17.5 where in-
down to K = 17.5. The actual completeness is most
completeness becomes significant. The slope we fit to
likely slightly lower, since in the crowded central region
the KLF itself for sources detected in both H and K is
of the cluster our method may produce false positives
0.35±0.04.
when the artificial star is placed on top of a real star of
We have previously used multi-color photometry in
approximatelythesamemagnitude. Despitethisconcern
conjunction with near-IR spectroscopy of the bright-
wehaveusedthecalculatedincompletenessincorrecting
est stars to determine the IMF for embedded clusters
the KLF; however, we have excluded the K = 17.5 bin
(Leistra et al. 2005). Although we do not have spec-
fromconsideration,both becausethis effect willbe most
tra in this case, we can still derive extinctions for in-
pronounced at faint magnitudes, and because statistical
dividual stars based on their near-IR colors. We use
uncertainties in the incompleteness will be significant.
the same evolutionarymodels and conversionsfrom the-
Knowingourincompleteness,wecancalculatetheKLF
oretical to observed quantities described for the KLF
forthe cluster. Figure4showsboththeuncorrectedand
method. TheZAMSderivedfromtheevolutionarytracks
completeness-corrected versions of the KLF, as derived
of Meynet & Maeder (2003) is overplotted on the dis-
from all sources detected in K. The slope of the KLF is
tance and extinction-corrected CMD in Figure 5. The
0.31±0.04, with no extinction correction applied.
ZAMSliesinthemiddleofthedistributionofstarsdueto
the method used to estimate an averageextinction. The
3.1.2. The IMF
scatter around the ZAMS is rather large, and is likely
We derived an IMF for the Sh 2-217 cluster by two due to a combination of variable extinction in the clus-
methods. For both methods we used a distance to the ter region and poor photometry in H, especially in the
cluster of 5 kpc (Deharveng et al. 2003). In general central portion of the cluster.
we believe that the “CMD method” for deriving the Since extinction appearsto vary acrossthe cluster, we
IMF,whichusesindividually-derivedextinctionsforeach impose an extinction limit on the sample used for the
star in the cluster, to be more reliable than the “KLF CMD computation of the IMF. Since the most massive
method”whichassumesacommonextinctiontoallstars stars can be seen to greater extinction than less massive
in the cluster, since variable extinction is frequently ap- stars, neglecting to impose this constraint will produce
parent in the NIR images of embedded clusters. How- anoverlyflatIMF.Thusweneedtosimultaneouslylimit
ever, in this case our K data is superior to our H data our sample by extinction and by mass. We use a mean
duetothe differenceinthe seeing,whichwas∼0.′′ inK extinction to the Sh2-217 cluster of AV = 9 mag as de-
and ∼1.2′′ in H. This suggests that despite the general terminedbyindividuallyde-reddeningsourcesuntilthey
drawbacks of the KLF method it may be preferable in reachthemainsequence. Withtheexceptionofafewex-
this situation; including the H-band data adds nothing treme outliers that mostlikely suffer from poorphotom-
if it is of poor quality. At the least, using both methods etry,mostsourceswithhigherextinctionshaveAV <15.
will provide information on potential systematic effects At a distance of 5 kpc with our limiting magnitudes, we
inthe IMFdeterminationthatdepend onthe methodol- can observe stars earlier than G4 to an extinction of 9
ogy used. mag and earlier than F0 to an extinction of 15 mag.
The KLF method of determining the IMF is some- Atadistanceof5kpcwithourlimitingmagnitude,an
times employed even when multi-color photometry and extinction of AV = 9 limits us to G4 and earlier stars,
spectra are available (e.g. Blum et al. 2000), and is sim- while AV = 15 limits us to F0 and earlier. We select
ply a transformation from K magnitude bins to mass AV = 10 and G2 as our limits; stars with higher ex-
bins. Themajorproblemwiththisisthatofvariableex- tinction or later spectral type cannot be detected over
tinction,whichformanyembeddedclustersissignificant the entire range of mass or extinction included and thus
4 Leistra et al.
are excluded. Approximately 34 stars have a higher ex- Arches cluster (Stolte et al. 2002)), we expect the core
tinctionthanthis,includingfivewithextremecalculated IMF to be quite flat. When we re-derive the IMF for
extinctions (AV > 50) that most likely suffer from poor this cluster using their radius with our data, we derive
photometry and have unrealistic colors. This extinction an IMF of Γ = −1.55±0.22, statistically indistinguish-
limit will also have the effect of excluding stars with able fromour originalresult. Porraset al.(1999) do not
K-band excess from the IMF determination, since such comment on issues of confusion or field star contamina-
sources would appear to be at high extinction. This will tion, so we cannot evaluate how much of an effect it is
tend to push the IMF to flatter values, since lower-mass likelytohaveontheirresult;weexpectcrowdingtobea
sources spend more time as IR-excess objects and thus moresignificantissue,anda misidentificationofblended
aremorelikelytoberuledoutbythiscriterion. However, sources as single stars by Porras et al. (1999) could ac-
we consider this to be a better approach than including count for their finding a steeper IMF than we do using
the sources since: 1. the number of sources detected in the same method.
all three bands in this cluster showing near-IR excess is
3.2. IRAS 06058+2158 Cluster
small, suggesting such sources will not significantly in-
fluence the mass; 2. the majority are of a low enough We presenta three-colorcomposite of the cluster near
mass to fall below the completeness limit, and are thus the IRAS source 06058+2138 in Figure 6. Bik et al.
excluded anyway regardless of the method; 3. includ- (2005) obtained VLT spectra in K of several NIR point
ing them (since not all sources are detected in J) would sources near the IRAS point source, which is located
weaken the extinction limit and tend to again force the near the center of the cluster, and identified two can-
IMFtoflattervalues. InclusterswhereIR-excesssources didate massive YSOs and an embedded early-B star.
dominatewedonotfitanIMF(seeSection3.2.1forfur- The spectrophotometric distance they derive from the
ther discussion). B star is 1.0-1.5 kpc. This cluster is much more heav-
Individual extinctions are derived by moving the stars ily embedded than the Sh 2-217 cluster, with significant
along the direction of the reddening vector until they lie nebular emission and prominent dust lanes. Numerous
on the ZAMS. Once the stars have been corrected indi- OH and methanol masers have been detected in this re-
viduallyforextinction,theyareplacedinmassbins. We gion, which along with the IRAS point source suggest
fit a power law to the data, excluding masses < 1.1M⊙ ongoing star formation. (see, e.g., Caswell et al. (1995),
from the fit since they cannot be seen over the entire Szymczak et al. (2000)). A peak in the 8 µm emission
range of extinctions in the cluster. The IMF slope we is observed in the MSX data, extending from the region
derive by this method is Γ = −1.61±0.2. As for the ofNIR nebulosity to the southeastto the isolatedbright
KLF method, the quoted errors represent only the for- source.
mal errors in the fit and should be considered an under- The embedded cluster here is described in the com-
estimate. pilation of Lada & Lada (2003), who quote a dis-
The difference between these two values for the IMF tance of 1.5 kpc (Carpenter et al. 1993). However,
emphasizesthatformalstatisticalerrorssignificantlyun- Hanson, Luhman, & Rieke (2002) describe a UCHII re-
derestimate the true uncertainties in the IMF. In this gion associated with the IRAS point source, and quote
case, as was the case in Leistra et al. (2005), the CMD adistanceof2.2kpc (K¨ompe et al.1989),andBik et al.
method gives a noticeably flatter result than the KLF (2005) obtain a spectroscopic distance of 1.0-1.5 kpc.
method. Clearly the individual extinction correction With no a priori reason to prefer one distance over the
leads to the conclusion that more massive stars are other,andnouncertaintiesassociatedwitheither,weuse
present than an average correction does. This could the averagedistance of 1.5 kpc for our analysis.
be due to the effects of mass segregation, or to an in-
3.2.1. The KLF
correctly chosen extinction limit (so that we truly are
samplingmassivestarsmorecompletelythanlowermass We present color-color and color-magnitude diagrams
stars). It is difficult to understand which of these ef- for this cluster in Figures 7 and 8. The “cluster region”
fects is most important without obtaining spectra for a was defined to coincide with the extent of the near-IR
significant number of stars in the cluster. nebulosity, and field stars were statistically corrected as
This cluster is also analyzed by Porras et al. (1999), described in 3.1. The CMD shows objects spanning a
who use a slightly different distance (5.8 kpc) and ex- rangeofextinctions,withrelativelyfewobjectswithcol-
tinction(<AV >=5.3±3.7)andderiveanIMFslopeof ors consistent with unextincted main-sequence stars re-
Γ=−0.59basedon54sourcesusingtheJ vsJ−H CMD maining in the statistically corrected data. Since the
to individually correct extinctions and compare with a cluster in this case did not fill the field of view, we were
theoretical JLF. This is a significant discrepancy from able to use the data fromthe non-clusterportions of the
our result with either method. A number of factors may regionas ourfield, eliminating the needfor anoff-source
contribute to this difference. The most significant, how- 2MASS field and extrapolation based on the KLF. The
ever, is likely to be due to a different choice of cluster J−H vs. H−K color-colordiagramshowsthatthema-
boundaries. Their quoted cluster radius corresponds to jorityofsourcesinthe clusterregionexhibit K-bandex-
only 50′.′9, compared to ours of ∼ 80′′. This suggests cess and fall in the regionpopulated by reddened CTTS
that their IMF will be more weighted towardthe cluster and YSOs, suggesting they may be pre-main-sequence
corethanours. Thevaluetheyquoteforafield+cluster objects. BecausetheamountofK-bandexcessisaffected
IMF is Γ=−2.71±0.24,much steeper and in fact quite by many factors (see, e.g.,Meyer, Calvet, & Hillenbrand
closetothevalueweobtainusingtheKLF.Ifthecluster (1997)),derivingmassesfortheseobjectsisdifficult. We
suffers from mass segregation, as we would expect given thus derive only a KLF for this cluster, and do not con-
that it is observed even in very young clusters (e.g. the vert it to an IMF or derive an IMF from the CMD. We
Embedded Stellar Clusters: II 5
determine and correct for our incompleteness as in 3.1. distanceof3.5kpc (Wouterloot & Brand1989)is shown
The KLF we derive using a statistically-corrected sam- in Figure 12.
ple of all sources detected in K, with no attempt to cor- Using the color-magnitude diagram based method of
rectfor extinctiondue to the uncertaintyof the intrinsic derivinganIMFasdescribedabove,foralimitingextinc-
H−K colorofthe YSOsthatarepresent,hasaslopeof tionofAV =25,wefindanIMFslopeofΓ=−0.9±0.25.
0.30±0.03. If only sources detected in H are included, UsingtheKLFmethod(withnoextinctioncorrection,to
the KLF declines for K >14.5 since the high extinction mimic the results of a study with only single-color pho-
meansthefainterobjectsarelesslikelytobedetectedin tometry available) we arrive at Γ = −2.6±0.3. Even
H. afterallowingfortheerrorstobe largerthanquoteddue
to uncertainty in the photometry and the conversion to
3.2.2. Pre-Main-Sequence Objects mass, these two slopes are inconsistent with each other,
suggestingthatsystematiceffectsinoneorbothmethods
A total of 37 out of 58 sources (63%) detected in all
dominate over the statistical errors.
three bands show a K-band excess in the color-color di-
agram, suggesting they are pre-main sequence objects. 3.4. Sh 2-288
This is a lower limit on the actual pre-main-sequence
The near-IR cluster image (Figure 13) of the cluster
fraction of the cluster, since objects with a sufficiently
near Sh2-288 shows a cluster with a dense core, crossed
high IR excess may be detected in K but not in J or H.
near the center by a dust lane. The center of the clus-
Atotalof49sourcesweredetectedinK withinthe clus-
ter is unresolved in our images, taken with a seeing of
terregionthatwereundetectedinJ,H,orboth. Adding
0.7′′. This region was previously identified as an em-
in these sources would give a PMS fraction of 80%. The
bedded cluster by (Dutra & Bica 2001). In their cata-
latterfigureisanupperlimit,sincesomeoftheK−-only
log of outer-Galaxy HII regions, Rudolph et al. (1996)
detectionsarelikelytobeknotsofnebularBrγ emission
quote widely disparatedistances,with a radiokinematic
or heavily extincted backgroundstars. Comparing these
distance of 7.2 kpc and a photometric distance of 3.0
values to the near-IR excess fraction of embedded clus-
kpc(Brand & Blitz1993). Thekinematicdistancewould
ters of known ages presented in Haisch, Lada, & Lada
make the sources we observe (several with K < 12) ex-
(2001), we conclude that the age of the IRAS 6058 clus-
tremely massive, it is thus far more likely that the pho-
ter is less than 3 Myr.
tometric distance is correct, and we have used the pho-
We observe a few sources with colors even red-
tometric distance for our analysis.
der than the reddened extension of the CTTS locus.
The 8 µm image from the MSX missionshows a peak
Meyer, Calvet, & Hillenbrand (1997) observe sources
coincident with the near-IR nebulosity; there is not a
with similar colors, and suggest re-radiation by an ex-
significant amount of 8 µm emission from regions dark
tended envelope as an explanation.
intheNIR.Thissuggeststhattherearenotasignificant
number of sources so deeply embedded that they cannot
3.3. IRAS 06104+1524 Cluster
be seen in K present in this cluster.
Thenear-IRclusterimage(Figure9)oftheclusternear The K versus H −K color-magnitude diagram of the
IRAS06104+1524showsaclearseparationintotwosub- cluster near the HII region Sh2-288 (Figure 14) clearly
clusters separated by ∼ 2′. The southwest subcluster is shows the effects of variable extinction; the stars sep-
dominatedbytwocloselyspacedbrightsourceswhilethe arate into two groups, one nearly unextincted and one
northeastsubclusterisdenserandisnotdominatedbya with ∼AV =5. We correct for field star contamination
single object. A ridge of marginally higher density than using the region of the field outside the cluster region as
the surroundingfieldappearsto liebetweenthesubclus- described above. The J −H versus H −K color-color
ters, though it is not apparent whether this is a real diagram (Figure 15) shows few stars separated from the
feature. The MSX 8 µm image similarly shows two sep- main sequence locus by more than 2σ, suggesting that
arate peaks, with no indication of a connection. These most stars in this cluster are on the main sequence. Ex-
aretreatedasseparateclustersbyBica et al.(2003),and treme outliers in the color-color diagram were inspected
there are IRAS point sources associated with each of individually; in general, they lie in the crowded central
them(IRAS06104+1524andIRAS06103+1523,respec- region of the cluster and most likely suffer from poor
tively). The IRAS point sources both have kinematic photometry due to the different PSFs in H and K that
distances of 3.5 kpc (Wouterloot & Brand 1989), sug- resulted from variations in seeing. Such sources were
gesting the subclusters are related. The radio kinematic excluded from analysis of the KLF and the IMF. Addi-
distances derived to the two sources are the same, and tionally, the brightest source in the cluster, which lies
theangularseparationiscomparabletothesizeofeither in the most crowded central region and has a FWHM
subclump; in addition, a slight overdensity of stars can slightly broader than most sources in the field, was ex-
be seen in the K image. These factors suggest that, at cluded since it is quite likely to be a blend of multiple
the very least, these clusters are related; they may differ sources. We consider that the effects on the IMF are
inage,but they arelikely to be partofthe same general likelytobeworseifablendisincludedthanifanysingle
star-formation event. We see no difference between the star, even the most massive, is excluded.
two apparent in either the CMD or the color-color dia- Using the photometric distance of 3.0 kpc from
gram (Figures 10 and 11); if they do differ in age, it is (Rudolph et al.1996),wederiveanIMFfromtheKLFof
beyondtheabilityofourdatatodiscern. Duetothisap- Γ=−1.95±0.62. Tobettercomparethetwomethodsof
parentassociationandthe smallnumber ofstarsin each deriving the IMF, we included only those sources which
cluster, we analyze the two together as a single cluster. were also detected in H, so that the same datasetwould
The CMD after statistical correction and adjusting to a be used for both the KLF and CMD methods of deriv-
6 Leistra et al.
ing the IMF.We individually de-reddenedsourcesinthe 4. SUMMARY
H −K CMD until they were on the main sequence, im-
We present NIR images of four embedded clusters in
posing an extinction limit as before, and derive an IMF
the outer Galaxy. In the case of the cluster near IRAS
ofΓ=−1.62±0.65. Giventhe largeuncertainties,these
06058+2158 the number of stars with NIR excess indi-
resultsareentirelyconsistent. Thebetteragreementmay
catesapre-main-sequencefractionbetween60%and80%
be because the extinction bias is less in the latter case.
and an age of less than 3 Myr; the other three clusters
show less nebular emission and fewer stars with NIR ex-
3.5. Comparison of Methods for IMF Determination
cess indicating an older age. We compute the IMF for
A summary of the IMFs derived for the three clusters
the three clusters dominated by main-sequence stars, in
without significant numbers of pre-main-sequence stars
each case using both a KLF-based method relying on a
andthesimilarresultsfrom(Leistra et al.2005)isshown single extinction value for the cluster and using only K
in Table 1. In each case, there is a significant difference
banddataandaCMD-basedmethodwhereanindividual
betweentheIMFderivedfromtheKLFandthatderived
extinction value is calculated for each star. We found a
fromthe CMD.Does this reflectonlyuncertainties,oris
statisticallysignificantdifference betweenthe twovalues
one method in general more reliable than the other? A
in two of the three cases, prompting us to examine IMF
simple analysis would suggest that the CMD method is
values ofembedded clusters fromthe literature to deter-
more reliable, simply because it uses more information;
mine whether systematic effects are at work. We found
theextinctionclearlyvariesacrossmanyembeddedclus-
asignificantdifferenceinthe meanvalueofthe IMFsfor
ters (of those analyzed here, most notably Sh 2-288 and
embeddedclustersderivedfrommethodsthathandleex-
IRAS 06058+2158),and accounting for this should pro-
tinction individually compared with those that adopt a
vide a more robustestimate of the true IMF. We cannot
singlevalueforthe extinction. Althoughalargersample
say for certain that this is the case, however, without
wouldhelptomakethisclaimmorerobust,sincemanyof
obtaining spectra for most of the stars in each cluster,
the results come froma single study (Porras et al. 1999)
so that we can classify them spectroscopically and ob-
and methodological details of that work could affect the
tain individual masses. We note that in each case, the
results,weconsiderittobesignificantenoughthatIMFs
IMF we derivefromthe CMD by individually correcting
obtained by different methods should not be compared
the extinction for each object is flatter than that we de-
in an attempt to search for variations in the IMF from
rive from the KLF by assuming a single extinction for
region to region.
the entire cluster. This suggeststhatmore massivestars
Truly reliable IMFs for embedded clusters will most
may preferentially lie in more heavily extincted regions
likely require spectra for a large number of stars in the
in embedded clusters. Resolving this seeming discrep-
clusters; we are continuing to try to obtain spectra for
ancy would require spectra for a large number of cluster
thesesourcestobettercharacterizethemassivestarpop-
members, so that the IMF derivedfrom spectralclassifi-
ulation and the IMF of these clusters.
cation of stars can be compared to that derived via the
different photometric methods.
Weexaminetherelationbetweenderivedmassandex-
tinction (with the extinction limit imposed) for the Sh2-
217 cluster, where we have the most data, in Figure 16.
Such a relation appears to be present, albeit at low sig-
nificance. This effect is opposite in sign to what would
be expected from massive stars clearing their immediate
environment more rapidly than their lower-mass coun-
terparts,but the stars weobservearemostly of interme-
diate mass, rather than truly high mass, such that their
winds are not as significant; the effects of mass segrega-
tion, placing the more massive stars at denser regions in
the cluster, appear to dominate over the effects of clear-
ing in these clusters. Since the clusters are numerically
dominatedbylow-massstars,theaverageextinctionwill
be mostly determined by the average for the low-mass
stars,changedslightlybytheaverageforhigh-massstars.
If more massive stars are indeed preferentially found at
higherextinctionasourresultssuggest,thiswouldmean
the majority of (low-mass) stars are over-corrected for
extinction by a small amount when using a single value,
while a few (high-mass) stars are under-corrected (thus
lowering the derived mass) by a large amount; thus, the
effectsofover-correctingandunder-correctingextinction
do not fully cancelout, since the large under-corrections
would be more likely to move stars between mass bins
than the small over-corrections. If there is no relation
betweenmassandextinction,wewouldexpectthesetwo
results to cancel, since the average extinction for low-
massstarswouldbethesameasthatforhigh-massstars.
Embedded Stellar Clusters: II 7
WethankDonMcCarthyforuseofandassistancewith makesuseofdataproductsfromtheTwoMicronAllSky
the PISCES instrument. We thank Phil Massey and Survey (2MASS), which is a joint project of the Univer-
Michael Meyer for comments that improved this paper. sity of Massachusetts and IPAC, funded by NASA and
This research has made use of the SIMBAD database, NSF.
operated at CDS, Strasbourg, France. This publication
REFERENCES
Bessell,M.S.&Brett,J.M.1988, PASP,100,1134 Ko¨mpe,C.,Baudry,A.,Joncas, G.,& Wouterloot, J.G.A.1989,
Bica, E., Dutra, C.M., Soares, J., & Barbuy, B. 2003, A&A, 404, A&A,221,295
223 Lada,C.J.,Alves,J.,&Lada,E.A.1996,AJ,111,1964
Bik, A., Kaper, L., Hanson, M.M., & Smits, M. 2005, A&A, 440, Lada,C.J.&Lada,E.A.ARA&A,41,57
121 Leistra,A.,Cotera,A.S.,Liebert,J.&Burton,M.2005,AJ,130,
Blum,R.D.,Conti,P.S.,&Damineli,A.2000,AJ,119,1860 1719
Blum,R.D.,Damineli,A.,&Conti,P.S.2001,AJ,121,3149 Malagnini, M. L., Morossi, C., Rossi, L. & Kurucz, R. L. 1986,
Borissova, J., Ivanov, V. D., Minniti, D., Geisler, D. & Stephens, A&A,162,140
A.W.2005,A&A,435,95 Massey, P., Johnson, K. E., & DeGioia-Eastwood, K. 1995, ApJ,
Brand,J.&Blitz,L.1993A&A,275,67 454,151
Caswell, J. L., Vaile, R. A., & Forster, J. R. 1995, MNRAS, 277, Massey,P.2003, ARA&A,41,15
210 McCarthy, D.W., Ge, J., Hinz, J.L., Finn, R.A., & deJong, R.S.
Carpenter,J.M.,Snell,R.L.,Schloerb,F.P.,&Skrutskie,M.F. 2001,PASP,113,353
1993, ApJ,407,657 Meynet,G.&Maeder,A.2003, A&A,404,975
Comeron,F.,Rieke,G.H.,&Rieke,M.J.1996,ApJ,473,294 Meyer,M.R.,Calvet,N.,&Hillenbrand,L.A.1997,AJ,114,288
Cotera, A. S., Erickson, E. F., Colgan, S. W. J., Simpson, J. P., Muench, A.A., Lada, E.A., & Lada, C.J. 1999, in 33rd ESLAB
Allen,D.A.,Burton,M.G.1996,ApJ,461,750 Symposium,StarFormationfromtheSmalltotheLargeScale,
Deharveng, L., Zavagno, A., Salas, L., Porras, A., Caplan, J., & eds.F.Favata, A.Kass,&A.Wilson,301
Cruz-Gonzalez, I.2003, A&A,399,1135 Nagata, T., Woodward, C. E., Shure, M., & Kobayashi, N. 1995,
Dutra,C.M.&Bica,E.2000,A&A,359,9 AJ,109,1676
Dutra,C.M.&Bica,E.2001,A&A,376,434 Nu¨rnberger,D.E.A.,&Petr-Gotzens,M.G.2002,A&A,382,537
Dutra, C. M., Ortolani, S., Bica, E., Barbuy, B., Zocalli, M., & Okumura, S., Atsushui, M., Nishihara, E., Watanabe, E., &
Momany,Y.2003,A&A,408,127 Yamashita,T.2000,ApJ,543,799
Elmegreen, B. G., Efrernov, Y., Pudritz, R. E., & Zinecker, H. Porras, A., Cruz-Gonzalez, I., & Salas, L. 1999, in 33rd ESLAB
2000, in Protostars & Planets IV: University of Arizona Press; Symposium,StarFormationfromtheSmalltotheLargeScale,
eds.Mannings,V.,Boss,A.P.,&Russell,S.S.,841 eds.F.Favata, A.Kass,&A.Wilson,491
Figer, D. F., Morris, M., & McLean, I. S. 1996, in The Galactic
Center: Astronomical Society of the Pacific Conference Series; Porras, A., Christopher, M., Allen, L., DiFrancesco, J., Megeath,
edGrebel,R.,263 S.T.,&Myers,P.C.2003,AJ,126,1916
Figer, D. F., Kim, S. S., Morris,M., Serabyn, E., Rich, R. M., & Rudolph,A.L.,Brand,J.,deGeus,E.J.,&Wouterloot, J.G.A.
McLean,I.S.1999, ApJ,525,750 1996,ApJ,653,669
Figuerˆedo,E.,Blum,R.D.,Damineli,A.,&Conti,P.S.2002,AJ, Scalo,J.1998inTheStellarInitialMassFunction,ASPConference
124,2739 SeriesVol.142,eds.G.Gilmore&D.Howell,p.201
Figuerˆedo,E.,Blum,R.D.,Damineli,A.,&Conti,P.S.2005,AJ, Stolte, A.,Grebel, E. K.,Brandner, W., Figer,D. F. 2002, A&A,
129,1523 394,459
Haisch,K.E.,Lada,E.A.,Lada,C.J.2001,ApJ,553,153
Szymczak, M.,Hrynek,G.,&Kus,A.J.2000, A&AS,143,269
Hanson,M.M.,Conti,P.S.,&Rieke,M.J.1996, ApJS,525,750
Vacca,W.D.,Garmany,C.D.,&Shull,J.M.1996,ApJ,460,914
Hanson, M. M.,Luhman, K.L., & Rieke, G. H. 2002, ApJS, 138,
Wegner,W.1994,MNRAS,270,229
35
Wouterloot, J.G.A.&Brand,J.1989,A&A,80,149
Hillenbrand,L.A.,Massey,P.,Strom,S.E.,&Merril,M.K.1993,
AJ,106,1906
Homeier,N.L.,&Alves,J.2005, A&A,430,481
Ivanov, V. D., Borissova, J., Pessev, P., Ivanov, G. R. & Kurtev,
R.2002, A&A,394,1
8 Leistra et al.
TABLE 1
IMFs forembedded stellar clusters
Cluster IMF:CommonAV IMF:Individual AV Distance(kpc) Source
Sh2-288 Γ=−1.95±0.82 Γ=−1.62±0.5 3.0 Thiswork
IRAS06104+1524 Γ=−2.49±0.3 Γ=−1.38±0.6 3.5 Thiswork
Sh2-217 Γ=−2.7±0.25 Γ=−1.61±0.2 5.0 Thiswork
G305.3+0.2 Γ=−1.5±0.3 Γ=−0.98±0.2 4.0 PaperI
Fig. 1.— K-band image of the Sh 2-217 cluster center (North=up, East=left). Image is approximately 120′′on a side. The stellar
densitydoesnotplateauinourentire3′FOV,suggestingtheclusteroutskirtscontinueatleasttotheedgeofourimage.
Embedded Stellar Clusters: II 9
Fig. 2.— K vs. H−K color-magnitude diagram for the cluster near Sh2-217. Average DAOPHOT error bar is the size of the plot
symbolsorsmaller. Overalluncertaintiesincludingcalibration(whichincludetermsthatwillnotaffecttherelativepositionofthepoints)
areindicatedbythesymbolintheupperright.
10 Leistra et al.
Fig. 3.— Statistically-corrected K vs. H−K color-magnitude diagram for the cluster near Sh2-217. Statistical correction was done
based on the 2MASS Point Source Catalog for an adjacent field, extrapolating to the our limiting magnitude based on the luminosity
function.
Fig. 4.— K luminosityfunctionforthe Sh2-217cluster. (Solidismeasured; dashediscorrected forincompleteness.) Since thecluster
fillsthefieldofviewandthecomparison2MASSdatadonotgoasdeep, wedonotshowafieldsampleforcomparison.