The peculiar cluster HS 327 in the Large Magellanic Cloud: can OH/IR stars and carbon stars be twins?

11 Pages·0.97 MB·English

by  Jacco Th. van Loon

Astronomy & Astrophysics manuscript no.
(will be inserted by hand later)
The peculiar cluster HS 327 in the Large Magellanic Cloud: can
⋆
OH/IR stars and carbon stars be twins?
1 2 3 1 4 5
1 Jacco Th. van Loon , A.A. Zijlstra , L. Kaper , G.F. Gilmore , C. Loup , J.A.D.L. Blommaert
0
0
2 1 Instituteof Astronomy,Madingley Road, Cambridge CB3 0HA,United Kingdom
2 UMIST, P.O.Box 88, Manchester M60 1QD,United Kingdom
n 3 Astronomical Institute, Universityof Amsterdam, Kruislaan 403, NL-1098 SJ Amsterdam, The Netherlands
a 4 Institut d’Astrophysiquede Paris, 98bis Boulevard Arago, F-75014 Paris, France
J
5 ISO Data Centre, Astrophysics Div., Science Dept. of ESA, Villafranca del Castillo, P.O.Box 50727, E-28080
8
Madrid, Spain

1
v Received date; Accepted date
7
0 Abstract. The obscured OH/IR star IRAS05298−6957 in the LMC was recently noticed to be member of the
1 small double cluster HS 327 that also contains a carbon star (van Loon et al., 1998, A&A 329, 169). Hencethey
1
are coeval and have (nearly) the same progenitor mass, which can only be understood if Hot Bottom Burning
0 (HBB) has preventedIRAS05298−6957 from being a carbon star.
1 Wepresentextensivevisualandnear-IRphotometricdatafor>104starsinandaroundHS327,andspectroscopic
0
dataforsomeofthebrightestAGBstarsamongstthese.Colour-magnitudediagramsareusedtoestimatetheage
/
h for thecluster and its members, and luminosities are derived for the stars for which spectra havebeen obtained.
p The age for IRAS05298−6957 and the carbon star is estimated to be ∼ 200 Myr. This corresponds to a Main-
o- Sequence progenitor mass ∼4.0 M⊙ — the first direct measurement of the lower mass threshold for HBB. This
r agrees with stellar evolution models that,however, fail to reproduce thelow luminosity of thecarbon star.
t
s
a Keywords.Stars:carbon–Stars:evolution–Stars:AGBandpost-AGB–openclustersandassociations: general
: – Magellanic Clouds – Infrared: stars
v
i
X
r 1. Introduction The relative efficiency of the 3rd dredge-up is higher
a
for lower metallicity, leading to the prediction of a large
Starsofintermediatemass(∼1to8M⊙)evolvealongthe populationofcarbonstarsintheMagellanicClouds(Iben
Asymptotic Giant Branch (AGB) before ending their life
1981). Hence, the lack of observed carbon stars with lu-
as a white dwarf (Iben & Renzini 1983). The energy pro-
minosities greater than M = −6 mag, a full magni-
bol
duction during the final part of the AGB ascentnormally
tude below the AGB-tip luminosity, came as a surprise
takes place in a hydrogen shell surrounding the degener-
(Iben 1981;Costa& Frogel1996).Severalreasonsforthis
atedcarbon-oxygencore,butepisodicallyaninnerhelium
havebeensuggested,ofwhichthemostpromisingarethat
shellignites(thermalpulse=TP)whichmaycausemate-
carbon star formation is avoided by nuclear processing of
rialenrichedwith the products ofnuclearprocessingsuch
carbonintooxygenandnitrogenatthebaseoftheconvec-
as carbon and s-process elements to enter the convective
tive mantle for the mostmassiveAGB stars(Hot Bottom
mantle. This is called 3rd dredge-up, and may result in a
Burning=HBB:Iben&Renzini1983;Woodetal.1983),
photosphericC/Oratioexceedingunity,creatingacarbon
or that luminous carbon stars become invisible at wave-
star. Thus the most luminous AGB stars are expected to
lengths shortward of ∼ 1µm due to obscuration by a cir-
be carbon stars. These concepts seem to be confirmed by
cumstellar dust shell as a result of intense mass loss on
the observation that in clusters in the Large Magellanic
the TP-AGB. Recent theoretical work confirms the oc-
Cloud(LMC)noM-typestarsarefoundbrighterthanthe
currence of HBB in massive AGB stars with LMC metal-
brightest carbon star in that cluster (Aaronson & Mould
licity (Boothroyd et al. 1993; Frost et al. 1998; Marigo
1985; Westerlund et al. 1991).
et al. 1998),and luminous obscuredcarbon stars have re-
Send offprint requests to: Jacco van Loon, e-mail: centlybeenfoundintheLMC(vanLoonetal.1997,1998,
[email protected] 1999a,b; Trams et al. 1999a).
⋆
based on observations obtained at theEuropean Southern
Observatory (La Silla, Chile)
2 van Loon et al.: Can OH/IRstars and carbon stars betwins?
TheluminousOH/IRstarIRAS05298−6957(Woodet 2.3. I-band spectroscopy with EMMI at the ESO/NTT
al. 1992) was noticed by van Loon et al. (1998) to be sit-
EMMI atthe ESO 3.5mNTT atLa Silla,Chile,wasused
uated in the core of the small cluster HS 327, and hence
on February 4, 1996, to obtain low-resolution (R∼ 500;
it may be possible to estimate its age by comparison of
grism#4) spectra between ∼0.61and 1.04µm. The goal
the Colour-Magnitude Diagram (CMD) with theoretical
was to obtain a spectrum of IRAS05298−6957, but this
isochrones.They also serendipitously discovereda carbon
was unsuccessful. However, four stars in the vicinity of
star in the same cluster. This observation suggests that
IRAS05298−6957wereidentifiedfromtheirredcolourson
carbon stars and (oxygen-rich) OH/IR stars may be co-
V and I-band acquisition images: (V −I) ∼ 3 to 4 mag,
eval.More TP-AGB stars haverecently been identified in
compared to normal red giants that have (V − I) ∼ 1
LMC clusters by Tanab´e et al. (1997).
to 2 mag. One of these is a carbon star, which spec-
We have obtained optical and near-IR imaging pho-
trum was published and discussed in van Loon et al.
tometryfortheHS327clusterandsurroundings,and(lim-
(1998). Here the results for the other three stars are pre-
ited) spectroscopic observations for some of the brighter
sented. All spectra were exposedfor 5 minutes. The CCD
stars,inorderto deriveanageforthe starsinthe HS327
frames were corrected for the electronic offset (bias) and
cluster.Theresultsarepresented,andtheimplicationsfor
for the relative pixel response (flatfield). Wavelength cal-
AGB evolution and carbon star formation are discussed.
ibration was performed relative to He+Ar lamp spectra.
The sky-subtracted spectra were then corrected for the
2. Observations wavelength dependence of the instrumental response as
measuredfromaspectrumofthestandardstarLTT1020,
2.1. Gunn gri-band imaging with the Dutch 0.9m
and for atmospheric continuum extinction.
ThedirectimagingcameraattheDutch0.9mtelescopeat
LaSilla,Chile,wasusedonthe sixnightsofDecember 25 2.4. I-band spectroscopy with EFOSC ii at the
to30,1996,toobtaindeepimagesofa3.77′×3.77′ region
ESO/3.6m
around the cluster HS 327, through Gunn g (λ = 5148
0
˚A, ∆λ = 81 ˚A), r (λ = 6696 ˚A, ∆λ = 103 ˚A) and i EFOSC ii at the ESO 3.6m telescope at La Silla, Chile,
0
(λ =7972 ˚A, ∆λ=141 ˚A) filters (Thuan & Gunn 1976; was used on July 5, 2000, to obtain a low-resolution
0
Wade et al. 1979). The total integration time amounts to (R∼ 400; grism #12) spectrum between 0.60 and 1.03
3h45m per filter, split into 5 minutes exposures to avoid µm of a very red point source that was discovered on
saturation and to allow for refocussing in order to reach the K -band image, near HS 327. The spectrum was ex-
s
thebestimagequality.Thepixelsmeasure0.442′′×0.442′′ posed 25 minutes. Data reduction was analogous to the
on the sky, and stellar images on the combined (shift- EMMI spectra, but the observed standard star was LTT
added) frames have a FWHM of ∼ 1.5′′, though some 7379.An appropriatefilterwasappliedto removecharges
individual frames show stellar images of <∼1.1′′ FWHM. from the impact of energetic particles. Flexure was found
TheCCDframeswerereducedusingstandardprocedures to be negligible. The thin, back-illuminated CCD #40 in
withintheESO-MIDASpackage.Thephotometrywascal- use with EFOSC ii causes severe fringing at wavelengths
ibratedbymeansofregularobservationsofstandardstars λ>∼0.8µm. The telluric emission lines of the sky at dawn
(see Appendix A and Sect. 3.3 for more details). were especially bright and difficult to remove: preference
was given to an accurate stellar continuum determina-
tionforλ<∼0.7µm,whichinthecrowdedfieldallowedonly
2.2. Ks-band imaging with SOFI at the ESO/NTT
small patches of sky near the object to be taken.
SOFI at the ESO 3.5m NTT at La Silla, Chile, was used
onFebruary15,1999,toobtainaK -band(λ =2.162µm,
s 0
3. Analysis
∆λ = 0.275µm) image of HS 327 in order to identify the
counterpart of IRAS05298−6957 and to derive accurate
3.1. Cluster morphology
bolometricluminositiesforthebrighteststarsinthevicin-
ity of HS 327. Nine dithered images, each of 10×2 sec- The open cluster HS 327 (Hodge & Sexton 1966) is poor
ond integration time were combined into one image with but rather compact, measuring ∼ 0.5′ across (∼ 7 pc at
an effective integration time of 3 minutes, at an airmass the 50 kpc distance to the LMC). Close inspection of the
of 1.322. Standard near-IR observing and data reduction region revealed an accompanying loose cluster of stars,
techniques were employed. The sky background was de- that we callHS 327-E(East),ata distance of ∼45′′ from
rived from the median of the individual frames, and sub- the western cluster HS 327-W originally listed by Hodge
tracted.Thepixelsmeasure0.292′′×0.292′′onthesky,and &Sexton(Figs.1&2).TheHS327-Ecomponentmaybe
stellarimagesonthefinalframehaveaFWHMof∼1.0′′. identified with the small faint cluster KMK 59 (Kontizas
For photometric calibration the NICMOS standard star er al. 1988), but there is some confusion: SIMBAD’s
S 121-E (K = 11.781±0.005 mag) was observed at an (J2000.0) α = 5h29m23sδ = −69◦55′18′′ would indeed
s
airmassof1.224by combining five shifted images,eachof identify KMK 59 with HS 327-E, but Kontizas et al.
10×2 second integration time. themselves give (converted to J2000.0) α=5h28m56sδ =
van Loon et al.: Can OH/IRstars and carbon stars betwins? 3
Fig.1. Dutch 0.9m Gunn-i (left) and NTT/SOFI K -band (right) images of the region around the LMC cluster HS
s
327. North is up and East to the left; the edge measures 3′40′′ on the sky.
Fig.2. Brightness contoursin Gunn g (left), i (middle) and K (right). The i-band map is overlaidwith the brightest
s
stars, of which the carbon star and the other three stars of which spectra were taken are labelled, as well as the
serendipitously discovered red star. The cross indicates the position of IRAS05298−6957. Orientation and field size
are the same as in Fig. 1.
−69◦55′28′′, placing KMK 59 ∼ 2.5′ to the West of HS rareagain,andonlyoneofthese (star“C”)causesapeak
327-E.The SecondGenerationDigitized Sky Survey does rivalling the cluster contours. What is most intriguing is
not show anything obvious at Kontizas et al.’s position. that HS 327-W is blue, whilst HS 327-E is red.
The morphology of the cluster pair becomes apparent
whenbrightnesscontoursareconstructedfromtheimages The brightest stars are overlaid on the i-band image
after smoothing by a Gaußian filter with σ = 17.5′′ (Fig. (Fig. 2) to aid in comparing the overall morphology with
2). HS 327 stands out most conspicuously in the g-band the location of individual sources. The stars for which
becausebrightblue fieldstarsarerelativelyrare.Thevis- spectra have been obtained are labelled: the carbon star
ible star density is highest in the i-band, where several (“carbon”:vanLoonetal.1998),threeM-typestars(“A”,
individual bright(red) starsadd detailed structure to the “B” and “C”),and a veryred carbonstar (“red”)discov-
morphologyoftheclusteraswellassomepeaksnotassoci- ered on the K -band image, as well as the location of the
s
atedwith the cluster.Inthe K -bandbrightstarsbecome OH/IR star IRAS05298−6957(cross).
s
4 van Loon et al.: Can OH/IRstars and carbon stars betwins?
Fig.3. Dutch 0.9m Gunn-i (left) and NTT/SOFI K -
s
band (right) images of the immediate surroundings of
IRAS05298−6957. North is up and East to the left; the
edge measures 24′′ on the sky. The cross marks the posi-
tion of the near-IR counterpart of IRAS05298−6957.
The near-IR counterpart of IRAS05298−6957 is very
prominent in the K -band image, but (completely) invisi-
s
ble in the i-bandimage (Fig.3). Notice the blue star very
nearIRAS05298−6957.The30secondsi-bandacquisition
image obtained with EFOSC ii for the spectroscopy of
Fig.4.NTT/EMMIspectraofM-typestarsnearHS327.
theveryredstarinthesamefieldreachedthesamedepth
thankstothegreatercollectingareaoftheESO/3.6mtele-
scope and the ∼1′′ stellar images. carbon star (“red”) are considered not to belong to the
cluster complex, as they are all situated well outside the
faintestbrightnesscontourthatisolatesthedoublecluster
3.2. Cluster membership
from the surrounding field.
IRAS05298−6957, the carbon star and possibly also star
“A” are seen in projection against the Eastern part of
3.3. Spectral types
the HS 327 cluster complex. How likely is their physical
association with the HS 327-E cluster? Spectra are now available for the brightest and reddest
BesidesIRAS05298−6957itself,therearefivesimilarly IRpointsourcesinthe fieldaroundHS327,whichcanbe
brightmid-IRpointsourceswithinaradiusof10′fromHS usedtoconfirmthattheseareAGBstarsandtodetermine
327-E,andtheseareprobablynotallobscuredAGBstars. whether they are oxygen-rich(M-type) or carbon stars.
This corresponds to a surface density of < 70 obscured Thespectraofstars“A”,“B”and“C”arepresentedin
AGBstarspersquaredegree,ora<∼6%chancecoincidence Fig. 4 in arbitrary flux units, together with their J2000.0
of an obscured AGB star within 1′ of HS 327-E. coordinates.Thespectraareclassifiedbycomparisonwith
Similarly, there arenine knowncarbonstars including Turnshek et al. (1985). They are all M-type: strong TiO
the cluster member, and one additional possible carbon absorption bands dominate, and absorption by VO be-
star, within a radius of 10′ from HS 327-E. This corre- comes noticable for the cooler star “C”. The Ca ii triplet
sponds to a surface density of ∼ 100 carbon stars per is clearly visible in all three stars, and reflects slight red-
square degree, or a 10% chance coincidence of a field car- shifts consistent with these stars being members of the
bonstarwithin1′ofHS327-E.Thenearestknowncarbon LMC. The Ca ii strength and the estimated luminosities
starotherthantheclustermemberortheveryredcarbon (Section 3.8 and Table 2) indicate they are AGB stars.
star is SHV0530080−695949at 2.7′. TheOH/IRstarIRAS05298−6957wasalreadyknown
Star“A”isamoderatelyluminousM-typestar,which to be anoxygen-richAGB star:spectra in the 3 to 14 µm
are less rare objects than OH/IR stars or carbon stars. region indicate the presence of oxygen-rich circumstellar
Indeed,thetwosimilarstars“B”and“C”arefoundwithin dust (van Loon et al. 1999a; Trams et al. 1999b), and a
2′ from HS 327. With three such objects in a field >10× 1.6 GHz spectrum of OH maser emission suggests LMC
the projected size of the cluster complex, the chance of membership and a wind velocity of ∼11 km s−1.
encountering star “A” within the cluster boundaries by The spectrum for the “red” star is presented in Fig. 5
coincidence is <∼30%. in arbitrary flux units, together with the J2000.0 coordi-
Thus it is highly likely that both the OH/IR star nates.The reddenedcontinuumshowsweakabsorptionof
IRAS05298−6957andtheclustercarbonstararephysical thestrongestCNseriesandofC near0.77µm,identifying
2
membersofHS327-E,andprobablytheM-typeAGBstar this star unambiguously as a carbon star. The star may
“A”too.The M-type stars“B”and“C”andthe veryred be only moderately carbon-richand/or rather warm.The
van Loon et al.: Can OH/IRstars and carbon stars betwins? 5
ments. Incompleteness is mainly due to blending and/or
positional discrepancies as a result of crowding. The i-
band point-sourceextractionreachesa completenesslevel
of 98% at i ∼ 18.75 mag, dropping to 55% and 5% at
i ∼ 21.75 and 23.75 mag, respectively. For stars with
i>∼22.25magtheerroronthephotometryis>∼1mag,with
magnitudes systematically too bright by >∼0.1 mag. The
g and r-bands go ∼0.75 mag deeper. The K-band images
reach K ∼19 mag.
s
Point-sourceswerecross-correlatedbetweenthediffer-
Fig.5. ESO 3.6m/EFOSC ii spectrum of the very red ent filters after geometric transformation (rotation and
lineartranslation),usinganiterativeschemewithagrow-
carbon star near HS 327.
ing search radius, and rejecting extended sources on the
basis of the sharpness parameter returned by ALLSTAR.
absenceofthe Caiitripletaround0.86µmmightindicate
that the photosphere is of low metallicity altogether. It
is difficult to assign a spectral sub-type, but the overall 3.5. Colour-magnitude diagrams
appearance and the lack of any conspicuous features that
prove otherwise suggest that it is of N-type, i.e. carbon The i versus (g −i) and K versus (i−K ) CMDs are
s s
enriched by 3rd dredge-up on the TP-AGB.(cf. Turnshek giveninFig.6fortheHS327clusterandsurroundingfield.
et al. 1985; Barnbaum et al. 1996). The cluster boundaries were defined by a rectangularbox
For the spectrumof the cluster carbonstar the reader encompassing the faintest solid i-band contour in Fig. 2.
is referredto vanLoonet al.(1998).They arguethat itis The red stars for which spectra have been obtained are
anAGB star,which agreeswith the estimated luminosity labelled, and the location of the AGB, RGB and Main
(Section 3.8 and Table 2). Sequenceareindicated.TheK -bandimagebarelyreaches
s
asfaintastheMainSequence.Thereddeningvectorisde-
termined by convolving the extinction curve compiled by
3.4. Multi-object photometry
Mathis (1990)witha Vegamodel(Kurucz1993),yielding
Multi-object photometry was performed on the images extinction coefficients A /A = 1.094, 0.796, 0.621 and
λ V
using an implementation of DAOPHOT (version ii) and 0.112forthegri-andK -bands,respectively.Starsofother
s
ALLSTAR (Stetson 1987) within ESO-MIDAS. spectral types have somewhat different extinction coeffi-
The Point Spread Function (PSF) was represented by cients, especially at shorter wavelengths. Kurucz (1993)
ananalyticMoffat functionwith β =2.5 plus a quadratic and Fluks et al. (1994) spectra are used to transform the
lookuptabletotakeintoaccountvariationsintheshapeof isochronesofBertellietal.(1994)intoCMDs,takinginto
thePSFacrosstheframe.ThisPSFwasestablishedusing account the filter and CCD response curves. The metal-
bright, relatively isolated stars in an iterative scheme in- licity ofthe intermediate-agepopulationis assumedto be
volvingsubtractionofcontaminatingneighbourstars.The [metals/H]=−0.4 (typical for the LMC).
multi-object photometry itself comprised between several The field contains stars with ages ranging from ∼ 0.1
and a dozen passes (per filter) through DAOPHOT & to ∼10 Gyr. The age of the oldest stars is not very accu-
ALLSTAR, eachtime subtracting already measuredstars
ratebecauseisochronesforstarsofsuchagesdonotdiffer
in order to also find and measure fainter and/or blended
much in colours and the photometry becomes incomplete
stars.Thisresultedinthe detectionof∼2.4×104 sources
above the Main Sequence turn-off for stars older than a
in each of the gri-bands, and ∼ 1.0×104 stars in the K-
few Gyr. Also, the reddening of the stars with respect to
band in the 3.7×10−3 square degree area around HS327.
the isochrones is a priori unknown, although the general
The photometric calibration was established by com-
look of the CMDs suggests a visual extinction of no more
parison of aperture photometry on the standard star im-
than a few 0.1 mag.
ages with aperture photometry on a (few) dozen bright
stars in the images of the HS327 area (after subtract-
ingcontaminatingneighbourstarsusingPSFfitting).The 3.6. Colour-colour diagrams
multiple-object PSF photometry was then scaled accord-
ingly. The calibration of the gri-band photometry, cor- Colour-colourdiagramshavethepotentialtoidentifystars
recting for airmass dependencies of the photometric zero with particular spectral signatures, such as carbon stars.
point, is mainly based on the December 25 night but the Inoxygen-richcoolstarstheg-,r-andi-bandsincludeTiO
photometric zero point has been checked for the other absorption,with the i-band also including VO absorption
nightstoo(seealsoAppendixA).Photometriccalibration that is present in the coolest stars. In carbon stars the
errors are unlikely to exceed a few %. g-band includes C absorption, and the r- and i-bands
2
The completeness and photometric errors were esti- include CN absorption (stronger in the i-band). The K -
s
mated by a restricted number of artificial star experi- band mainly measures the continuum.
6 van Loon et al.: Can OH/IRstars and carbon stars betwins?
Fig.6. CMDs of i versus (g −i) (top) and K versus (i−K ) (bottom) for the double cluster HS 327 (left) and
s s
surrounding field (right). Stars whose spectra have been taken are labelled, and the location of the AGB, RGB and
Main Sequence are indicated. Isochrones (Bertelli et al. 1994) are plotted for [metals/H]=−0.4, and ages of 108, 109
and 1010 yr.
The carbon stars indeed exhibit somewhat different There is only a small mismatch between the synthetic
colours from the M-type stars (Fig. 7). Carbon stars may colours for M0 to M3 stars andthe observedredgiant se-
have bluer (g −i) colours than M-type stars if they do quence. This may be due to differences in metallicity or
nothavevery strongC bands thatabsorbin the g-band, slight inaccuracies in the synthetic models. The observed
2
becausetheg-bandintensityofM-typestarsissuppressed starsredderthantheseearly-Mtypestars—includingthe
byTiOabsorption.Ascanclearlybeseeninthesynthetic, three M-type stars A, B and C — are not represented by
solar-metallicity M-type spectra from Fluks et al. (1994), the synthetic sequence. This may be due to the spectral-
the coolest M-type stars have relatively red (i−K ) for typedependenceoftheinter-andcircumstellarreddening,
s
their (g −i) colours due to the TiO and VO absorption which is especially important for intrinsically red objects
that affects the i-band. The same happens for the carbon towards the blue wavelengths. Large-amplitude pulsation
stars,butnowitisduetotheCNabsorptioninthei-band. of AGB stars may be responsible for weaker TiO absorp-
van Loon et al.: Can OH/IRstars and carbon stars betwins? 7
Fig.8. SEDs of the red stars near HS 327.
Fig.7.(i−K )versus(g−i)colour-colourdiagramforthe
s
stars in the field of HS 327.The M-type and carbonstars
too close to IRAS05298−6957 to have been included
for which spectra have been obtained are indicated by
in the Loup et al. catalogue that was compiled us-
circlesandsquares,respectively.SyntheticM-typespectra
ing a homogeneous automatic multiple-object photome-
from Fluks et al. (1994) are plotted too (crosses, solid
try procedure. Photometry is obtained by aperture pho-
line).
tometry on the original images. The magnitudes fol-
low the IRAS/ISO convention, with zero magnitudes of
the LW1 (4.5 µm), LW2 (6.7 µm), LW10 (12 µm) and
tion bands in the i-band (Schultheis et al. 1998). It may
LW3 (14.3 µm) filters corresponding to 181.8, 89.5, 34.7
be notedthatspectraareavailablefornearlyalllate-type
and 20.7 Jy, respectively. Additional near- and mid-IR
AGBstarsinthefield,andthattherearenostrongcandi- photometry and spectroscopy for IRAS05298−6957 can
datesforcarbonstarsthathaveescapednoticeotherthan
be found in Trams et al. (1999). This object was also
the two identified.
detected by MSX (Price & Witteborn 1995) at 0.27
Jy in Band A (Egan et al. 1999), centred at 8.3 µm
3.7. Mid-IR photometry (range: 6.8 to 10.9 µm). Inspection of the MSX im-
ages (http://www.ipac.caltech.edu/ipac/msx/msx.html)
Mid-IR images at wavelengths of 4.5, 6.7, 12 and 14.3 µ reveals that IRAS05298−6957 was also detected in Band
exist that cover the HS 327 area. These were obtained D, centred at 14.6 µm, and that the red field carbon star
with the CAM instrument onboard ISO for the ISOGAL was marginally detected at ∼ 0.03 Jy in Band A (be-
(Omont et al. 1999) and mini-survey (Loup et al. 1999) low the nominal sensitivity limit). This is all consistent
projects.Thedetailsofobservationanddatareductionas with the ISO photometry. The optical, near- and mid-IR
well as the catalogue with point source photometry will photometry in Table 1 was gathered at different epochs,
be described in a subsequent paper by Loup et al. whichisimportantbecauseIRAS05298−6957—andpos-
Only very few objects in the HS 327 area are bright siblytheotherredstarsaswell—isalong-periodvariable
enoughinthe mid-IR to have beendetected by these ISO star with a K-band amplitude of ∼ 2 mag (Wood et al.
observations. Amongst these, IRAS05298−6957, the car- 1992). LW2 and LW10 observations are available for two
bon star, the “ABC” stars and the very red star are all epochs, for which the photometry was averaged.
detected in at least one of the mid-IR passbands, which ThephotometricSEDsofthesixredstarsnearHS327
canbeusedtobetterdeterminetheirbolometricluminosi- are plotted in Fig. 8. IRAS05298−6957 is optically invis-
ties. These stars also constitute six of the seven brightest ible because of the severe circumstellar extinction by its
stars in the K-band. Besides these stars that we had al- massive dust envelope, whose emission produces a strong
readyidentifiedasbrightredgiantsonthebasisofshorter IR excess. The cluster carbon star does not suffer from
wavelengthdata,noothersuchstarsappearinthemid-IR much circumstellar extinction, but some IR excess emis-
images of the HS 327 area. sionisvisibleatthelongestwavelengths.Thefieldcarbon
The mid-IR photometry is summarised in Table 1. star (“red”) does experience significant extinction, but it
Stars A and B and the carbon star are too faint and/or is yetunclearwhether it is ofinterstellaror ofcircumstel-
8 van Loon et al.: Can OH/IRstars and carbon stars betwins?
Table 1. Optical, near- and mid-IR photometry for the six red stars in the HS327 area. Coordinates have ∼ 1′′
accuracy.
Star α (J2000.0) δ (J2000.0) g r i Ks [4.5] [6.7] [12] [14.3]
Cluster members:
IRAS05298−6957 5h29m24.5s −69◦55′14′′ >∼22.00 >∼22.00 >∼22.75 10.81 7.47 5.54 4.76 3.68
carbon 5h29m25.1s −69◦54′53′′ 16.67 15.07 13.77 10.60 10.18 9.94 9.46 8.74
A 5h29m24.2s −69◦55′42′′ 16.87 15.60 13.86 11.18 11.23 11.52 >∼11.40 >∼10.80
Field stars:
red 5h29m11.2s −69◦56′14′′ 20.07 17.23 15.89 11.15 9.18 8.75 8.55 8.60
B 5h29m36.3s −69◦56′05′′ 17.36 15.83 13.93 11.20 11.10 10.80 10.77 11.18
C 5h29m24.8s −69◦56′25′′ 16.70 15.12 12.93 9.75 9.74 9.60 9.60 9.71
Table 2. Spectral types and luminosities. once more during the time series. The relative photomet-
ric calibrationbetween the different epochs was improved
Star Spectrum log(L/L⊙) Mbol by subtracting a baseline constructed from 146 stars that
Cluster members: had been recovered at ≥ 30 epochs. All lightcurves were
IRAS05298−6957 OH/IR >4.3 <−6.0 thencheckedbyeyeforvariability.Noconvincingvariable
carbon C4.5 3.8 −4.7 starcandidatescouldbeidentified:aselectionof∼50best
A M3 3.7 −4.5
cases were all consistent with variations due to inaccura-
Field stars:
cies in the source extraction and photometry.
red C 3.6 −4.3
B M5 3.7 −4.4
C M7 4.1 −5.6
4. Discussion
4.1. The age of the HS 327 cluster
larorigin.ThethreeMgiantsarenotnoticeablyreddened
nordotheyexhibitIRexcessemission,andtheirmass-loss It is difficult to assess ages of the individual cluster stars.
rates must therefore be very low: M˙ <∼10−8 M⊙ yr−1 (see The locations in the CMD of the cluster carbon star and
van Loon et al. 1999b). The derivation of accurate mass- star “A” suggest ages of t < 1 Gyr for these stars, de-
loss rates requires additional photometry at wavelengths pending on the exact amount of interstellar reddening.
of λ∼20 to 60 µm. For IRAS05298−6957 an age cannot be derived directly
from the CMD because of the difficulty to correct for its
3.8. Luminosities circumstellarextinction.It’sluminosityoflog(L/L⊙)>∼4.3
and likely evolutionary state at the very tip of its AGB
Bolometric luminosities (Table 2) are estimated by in- suggests an age of t<∼500 Myr (Bertelli et al. 1994).
tegrating under the SED. The luminosity derived for The best indicators for the cluster age in the optical
IRAS05298−6957 here is consistent with the more com- CMD are: (i) The compact red clump at (g −i) = 1.1,
pletederivationofthebolometricluminositygivenbyvan i = 18.2 and (i − K ) = 1.2, K = 16.8 represents
s s
Loon et al. (1999b): Mbol = −5.21 to −6.72 mag. It is stars with ages 1<∼t < 10 Gyr; (ii) The bulk of the Main
certainly the most luminous AGB star in the field. Both Sequence stars for which photometry could be extracted
theclusterandfieldcarbonstarsareafewtimeslesslumi- have ages of a few Gyr, but the data is incomplete for
nous,withluminosities typicalfor opticallybrightcarbon Main Sequence stars that are older; (iii) An extension of
stars(Costa&Frogel1996).Theredfieldstarislesslumi- the Main Sequence to younger stars with ages t>∼2×108
nous than obscured carbon stars detected by IRAS (van yr; (iv) Luminous AGB stars with ages of a few ×108 yr.
Loon et al. 1997, 1998, 1999a,b). It is difficult to disentangle field and cluster due to
the low contrast in stellar density (Figs. 1 & 6). In order
to estimate whether the cluster belongs to the dominant
3.9. Search for short-period variable stars
fieldpopulationofafewGyr,ortothesparseyoungerfield
The imaging data comprises six nights of intensive obser- population of a few hundred Myr, statistics are obtained
vation, which would, in principle, allow the detection of about the relative frequency of stars of different classes
variability on timescales of a few days and with ampli- as defined in the CMD (Fig. 9): Main Sequence stars of a
tudes of the order of several tenths of a magnitude, e.g. few Gyr (MS ), t∼500 Myr to 1 Gyr (MS ) and t∼200
0 1
Cepheids.Tothisaimweinvestigatedthe 46individualr- to 500 Myr (MS ); red clump stars, RGB stars of similar
2
bandimages,usingonlyasinglepassthroughDAOPHOT agesastheredclumpstars,andAGBstarsthatgenerally
& ALLSTAR. The 4th epoch (December 25) was taken represent ages of several hundred Myr. The counts are
as a reference, for it contained the largest number of ex- normalisedtotheirsum(Table3),andcomparedbetween
tractedstars(4868).Ofthese,3706wererecoveredatleast clusterandfieldtakingintoaccountthedifferencesinarea.
van Loon et al.: Can OH/IRstars and carbon stars betwins? 9
where M2 = M3 = 3.3 M⊙ the current (200 Myr) Main-
Sequence Turn-Off mass, and M4 = 4.0 M⊙ the Main-
Sequence progenitor mass corresponding to the current
tipoftheAGB(Bertellietal.1994).AssumingaSalpeter
Initial Mass Function (Salpeter 1955)
dN
=ξM−2.35 (1)
dM
the observed NMS = 12 for M1 = 2.2 M⊙ (500 Myr:
MS ) predicts N =5, and the observed N =81 for
2 post MS
M1 = 1.75 M⊙ (1 Gyr: MS1+MS2) predicts Npost = 14.
These very rough estimates agree reasonably well with
the observed N ∼ 15 (“AGB”). There are two TP-
post
AGB stars (the carbon star and the OH/IR star) where
less than one would be expected, but this is a bias as the
cluster was studied exactly for reason of the presence of
these two stars.
4.2. Cluster binarity
The LMC contains several populous binary clusters
(Dieball et al. 2000; Dieball & Grebel 2000). HS 327 ap-
Fig.9. HRD with all stars in and around HS 327, with pears double (see also Pietrzyn´ski & Udalski (2000a,b),
indicated the areas for counting the different classes of showing that binarity also occurs amongst sparse open
stars. clusters. The moderate age of HS 327 implies that such
clusters can resist tidal disruption for some while, al-
Table 3. Relative number of the different classes of stars though we do not know of course how many of HS 327’s
compared to their total, n = Ni/Ntot, both for the clus- members have already been stripped off in the past.
ter (Ntot = 480) and the field (Ntot = 6739) around HS There is some indication for a difference in the stel-
327. The cluster-over-field density, ρ = Ni,cluster/Ni,field, lar content of the West and East components: HS 327-E
is corrected for the relative area (0.000334 and 0.003401 is redder, or rather brighter in the K band. An analy-
s
⊓⊔◦, respectively). sis as performed in Table 3 but now comparing the East
and West components of HS 327 suggests that HS 327-W
Cluster Field ρ mightbe somewhatyoungerthanHS 327-E.Perhapsstar
MS0 0.529±0.041 0.541±0.011 0.71±0.05 formation in the West part was triggered by momentum
MS1 0.056±0.011 0.049±0.003 0.84±0.17 input into the ISM by massive evolved stars in the East
MS2 0.025±0.007 0.014±0.001 1.33±0.41 part. This result is not statistically significant, though,
Clump 0.125±0.017 0.138±0.005 0.66±0.09
and Pietrzyn´ski & Udalski (2000a,b) derive ages that are
RGB 0.233±0.024 0.247±0.007 0.68±0.07
equal to within ∼50 Myr.
AGB 0.031±0.008 0.012±0.001 1.86±0.52
4.3. Progenitor masses of carbon stars and OH/IR
Although the statistics are not striking, the different
stars
classes of stars all suggest that the cluster is associated
with the younger population of a few hundred Myr (that The obscured oxygen-rich AGB star IRAS05298−6957 is
is also presentin the field), andthat the older population more luminous than the cluster carbon star, which in its
of a few Gyr belongs entirely to the field. The statistics turn is more luminous than the oxygen-rich cluster star
of the AGB stars and youngest Main Sequence stars in “A”. Because these three stars have formed simultane-
particular favours the interpretation of HS 327 having an ously, the sequence in luminosities strongly suggests a se-
age of t∼200 Myr.This agreeswith the age of t∼160± quence in evolutionary state as a result of a sequence in
20 Myr as derived by Pietrzyn´ski & Udalski (2000a,b), progenitormass,withIRAS05298−6957the mostevolved
which happens to correspond to an epoch of intensified and with the most massive progenitor. How can the se-
star formation that also seems to have occurred in the quence in chemistry be understood?
field surrounding HS 327. Intermediate-mass AGB stars become carbon stars
AGB stars seem to be relatively abundant within the as 3rd dredge-up becomes effective on the upper AGB.
cluster. As a simple consistency check, the number N This process is counteracted upon in massive AGB stars
MS
of Main-Sequence stars within a certain mass interval by carbon burning at the base of the convective mantle
[M ,M ] may be compared to the number N of post- (HBB). Clearly, the progenitor mass of the cluster car-
1 2 post
Main Sequence stars within the mass interval [M ,M ], bon star must be high enough to support 3rd dredge-up
3 4
10 van Loon et al.: Can OH/IRstars and carbon stars betwins?
but too low for the onset of HBB, whilst the progenitor Table A.1.PhotometricstandardstarsfortheGunnsys-
of IRAS05298−6957 was sufficiently massive for HBB to tem.
occur. Star “A” has yet to experience (enough) thermal
pulses to become a carbon star. Star literature observed
The cluster age of t ∼ 200 Myr implies a progenitor g r i g r i
mass of MMS∼4.0 M⊙ for stars currently at the tip of BD+21 607 9.25 9.26 9.27 9.25 9.26 9.27
HD 19445 8.088 8.070 8.070 8.09 8.07 8.07
the AGB (Bertelli et al. 1994). The range in progenitor
Ross 374 10.851 10.772 10.85 10.77 10.74
massesofstarsalongthe AGB isless than0.1M⊙.Hence Ross 683 11.40 11.08 11.38 11.12 11.04
the firstdirectmeasurementforthe thresholdmassabove
Ross 786 10.06 9.81 10.03 9.81 9.73
which HBB occurs indicates MHBB>∼4.0 M⊙. Ross 889 10.43 10.51 10.20 10.27 10.35
This agreeswellwith currentstellarevolutionmodels.
The lower and higher mass limits for carbon star forma-
tion in the LMC are currently believed to be M ∼ 1.2 and BD+21 607 are used as primary standards because
low
and Mhigh ∼ 4 M⊙, respectively (Groenewegen & de thesearetheonlystarsinoursetthathaveknowni-band
Jong 1993; Marigo et al. 1999). Ventura et al. (1999) magnitudes.Ross683andespeciallyRoss889showedthe
show that stars with MMS∼3.8 M⊙ first go through a largest discrepancies with respect to the full set of stan-
short(∆t∼30,000yr)J-type(13C-enhanced)carbonstar dard stars, and their use as standard stars may therefore
phase before HBB converts them back into oxygen-rich be questioned. Ross 374 and Ross 786 were found to be
stars. Because of the short lifetime and the lack of evi- in good agreement with their literature magnitudes. The
dence for 13C enhancement, it is unlikely that the cluster magnitudes determined here are accurate to ∼0.01 mag.
carbon star is currently going through this peculiar evo-
lutionaryphase.It is possible thatIRAS05298−6957may
References
have done so in the past, though.
However,astherefereerightfullypointedout,itisdif- Aaronson M., Mould J., 1985, ApJ288, 551
ficult to reconcile the low luminosity of the cluster car- BarnbaumC.,StoneR.P.S.,KeenanP.C.,1996,ApJS105,419
bon star with a 4 M⊙ TP-AGB star. Stellar evolution Bertelli G., Bressan A., Chiosi C., Fagotto F., Nasi E., 1994,
A&AS106, 275
models (Marigo et al. 1999) predict that for a star with
Boothroyd A.I., Sackmann I.-J., Ahern S.C., 1993, ApJ 416,
MMS∼4.0 M⊙ the onset of the TP-AGB occurs around
762
M ∼ −6 mag. The luminosity of the cluster carbon
bol Costa E., Frogel J.A., 1996, AJ 112, 2607
star of Mbol ∼ −4.7 mag would rather suggest MMS<∼3.0 Dieball A.,Grebel E.K., 2000, A&A358, 897
M⊙. The star may have been fainter than average due to Dieball A.,Grebel E.K., Theis C., 2000, A&A 358, 144
either stellarsurfacepulsations,apost-TPluminositydip
EganM.P.,etal.,1999,in:AstrophysicswithInfraredSurveys:
or both, but it is unlikely that this would account for a A Prelude to SIRTF, eds. M.D. Bicay, R.M. Cutri & B.F.
difference of ∼ 1.3 mag. It is also possible, but unlikely, Madore. ASPConf.Ser. 177, p404
thatthecarbonstarisnotaclustermember.Wetherefore Fluks M.A., Plez B., Th´e P.S., et al., 1994, A&AS105, 311
suggestthat currentstellar models over-estimatethe core FrostC.A.,CannonR.C.,LattanzioJ.C.,WoodP.R.,Forestini
mass(i.e. luminosity)atwhichAGB starswith M ∼4.0 M., 1998, A&A332, L17
MS
M⊙ start to experience 3rd dredge-up. Groenewegen M.A.T., deJong T., 1993, A&A267, 410
Hodge P.W., Sexton J.A., 1966, AJ 71, 363
Acknowledgements. Wearegratefulforgenerousallocation by Iben I., 1981, ApJ246, 278
ESO of Director’s Discretionary Time. This work has benefit- Iben I., RenziniA., 1983, ARA&A21, 271
ted from the use of the SIMBAD database, operated at CDS, Kontizas E., Metaxa M., Kontizas M., 1988, AJ 96, 1625
Strasbourg, France, and the Second Generation Digitized Sky Kurucz R.L., 1993, Kurucz CD-ROM. Smithsonian
Survey,producedat theSpaceTelescope ScienceInstituteun- Astrophysical Observatory,Cambridge MA
derU.S.GovernmentgrantNAGW-2166andbasedonphoto- Loup C., Cioni M.R., Blommaert J.A.D.L., 1999, in: The
graphic data obtained usingthe Oschin Schmidt Telescope on UniverseasSeenbyISO,eds.P.Cox&M.F.Kessler.ESA-
Palomar Mountain and theUK Schmidt Telescope. We thank SP427, p369
Fernando Comer´on for help with the SOFI observations and Marigo P., Bressan A., Chiosi C., 1998, A&A331, 564
forreadinganearlierversionofthemanuscript.Wealsothank Marigo P., Girardi L., Bressan A.,1999, A&A344, 123
the referee for her/his valuable suggestions. O Jacco se sente Mathis J.S., ARA&A28, 37
muito afortunado ser g´emeos unidos com o anjinho Joana. OmontA.,TheISOGALCollaboration,1999,in:TheUniverse
as Seen byISO,eds.P. Cox & M.F. Kessler. ESA-SP427,
p211
Appendix A: Photometric standard stars Pietrzyn´ski G., Udalski A., 2000a, ActaAstron. 50, 337
Pietrzyn´ski G., Udalski A., 2000b, Acta Astron. 50, 355
The standard stars used for the photometry in the Gunn
Price S.D., Witteborn F.C., 1995, in: Airborne Astronomy
system are listed in Table A1, together with their mag-
SymposiumontheGalacticEcosystem:FromGastoStars
nitudes from the literature (Thuan & Gunn 1976; Wade to Dust, eds. M.R. Haas, J.A. Davidson & E.F. Erickson.
et al. 1979) and from our observations during the five ASPConf.Ser. 73, p685
nightsfromDecember25to 29,1996.Ofthese,HD19445 Salpeter E.E., 1955, ApJ121, 161