A Spitzer Space Telescope far-infrared spectral atlas of compact sources in the Magellanic Clouds. II. The Small Magellanic Cloud

0.53 MB·English

by  Jacco Th. van Loon

Draftversion January25,2010
PreprinttypesetusingLATEXstyleemulateapjv.11/10/09
A SPITZER SPACE TELESCOPE FAR-INFRARED SPECTRAL ATLAS OF COMPACT SOURCES IN THE
MAGELLANIC CLOUDS. II. THE SMALL MAGELLANIC CLOUD
Jacco Th. van Loon1, Joana M. Oliveira1, Karl D. Gordon2, G. C. Sloan3, and C. W. Engelbracht4
1 AstrophysicsGroup,Lennard-Jones Laboratories,KeeleUniversity,StaffordshireST55BG,UK
2 SpaceTelescopeScienceInstitute, 3700SanMartinDrive,Baltimore,MD21218, USA
3 DepartmentofAstronomy,CornellUniversity,Ithaca, NY14853, USA
4 StewardObservatory,UniversityofArizona,933NorthCherryAvenue,Tucson, AZ85721, USA
Draft version January 25, 2010
0
1
ABSTRACT
0
2 We present far-infrared spectra, λ=52–93 µm, obtained with the Spitzer Space Telescope in the
Spectral Energy Distribution mode of its MIPS instrument, of a selection of luminous compact far-
n
infrared sources in the Small Magellanic Cloud. These comprise nine Young Stellar Objects (YSOs),
a
thecompactHiiregionN81andasimilarobjectwithinN84,andtworedsupergiants(RSGs). Weuse
J
thespectratoconstrainthepresenceandtemperatureofcooldustandtheexcitationconditionswithin
5
the neutral and ionized gas, in the circumstellar environments and interfaces with the surrounding
2
interstellar medium. We compare these results with those obtained in the LMC. The spectra of the

sources in N81 (of which we also show the ISO-LWS spectrum between 50–170 µm) and N84 both
]
A display strong [Oi] λ63-µm and [Oiii] λ88-µm fine-structure line emission. We attribute these lines
to strong shocks and photo-ionized gas, respectively, in a “champagne flow” scenario. The nitrogen
G
content of these two Hii regions is very low, definitely N(N)/N(O) < 0.04 but possibly as low as
h. N(N)/N(O)<0.01. Overall,the oxygenlines anddustcontinuum areweakerin star-formingobjects
p in the SMC than in the LMC. We attribute this to the lower metallicity of the SMC compared to
- that of the LMC. Whilst the dust mass differs in proportion to metallicity, the oxygen mass differs
o less; both observations can be reconciled with higher densities inside star-forming cloud cores in the
r
SMC than in the LMC. The dust in the YSOs in the SMC is warmer (37–51 K) than in comparable
t
s objects in the LMC (32–44 K). We attribute this to the reduced shielding and reduced cooling at the
a
low metallicity of the SMC. On the other hand, the efficiency of the photo-electric effect to heat the
[
gas is found to be indistinguishable to that measured in the same manner in the LMC, ≈ 0.1–0.3%.

1 This may result from higher cloud-core densities, or smaller grains, in the SMC. The dust associated
v with the two RSGs in our SMC sample is cool, and we argue that it is swept-up interstellar dust,
7 or formed (or grew) within the bow-shock, rather than dust produced in these metal-poor RSGs
8 themselves. Strong emission from crystalline water ice is detected in at least one YSO. The spectra
4 constitute a valuable resource for the planning and interpretation of observations with the Herschel
4 Space Observatory and the Stratospheric Observatory For Infrared Astronomy (SOFIA).
.
1 Subject headings: circumstellar matter — stars: formation — stars: mass loss — supergiants —
0 Magellanic Clouds — infrared: stars
0
1
: 1. INTRODUCTION for cooling. These diagnostic signatures became widely
v
accessiblewithinthe Milky Way,byvirtue ofthe Kuiper
i Aboutthecycleofgasanddustthatdrivesgalaxyevo-
X Airborne Observatory (KAO, see Erickson et al. 1984)
lution, much can be learnt from the interfaces between
and the Long-Wavelength Spectrograph (LWS, Clegg et
r the sources of feedback and the interstellar medium
a al. 1996) onboard the Infrared Space Observatory (ISO,
(ISM), and between the ISM and the dense cores of
Kessler et al. 1996).
molecular clouds wherein new generations of stars may
The gas-richdwarf companions to the Milky Way, the
form. These regions are characterized by the cooling
LargeandSmallMagellanicClouds(LMC andSMC)of-
ejecta from evolved stars and supernovae, and clouds
fer a unique opportunity for a global assessment of the
heated by the radiation and shocks from hot stars, in
feedback into the ISM and the conditions for star for-
supernova remnants and young stellar objects (YSOs)
mation, something which is much more challenging to
embedded in molecular clouds.
obtain for the Milky Way due to our position within it.
The interaction regions with the ISM lend themselves
The LMC and SMC are nearby (d ∼ 50 and 60 kpc, re-
particularlywellto investigationin the infrared(IR) do-
spectively: Cioni et al. 2000; Keller & Wood 2006) and
main,notablyinthe50–100µmregion;cooldust(∼20–
already the scanning survey with the IR Astronomical
100 K) shines brightly at these wavelengths, and sev-
Satellite (IRAS) showed discrete sources of far-IR emis-
eral strong atomic and ionic transitions of abundant el-
sioninthem(Schwering&Israel1989);itwasusedtode-
ements (viz. [Oi] at λ = 63 µm, [Oiii] at λ = 88 µm,
scribethe diffusecooldustandgasaswell(Stanimirovi´c
and [Niii] at λ = 57 µm) provide both important di-
et al. 2000). The stars, star-forming regions, and ISM
agnostics of the excitation conditions and a mechanism
are also lower in metal content than similar components
[email protected] oftheGalacticDisc,ZLMC ∼0.4Z⊙ andZSMC ∼01–0.2
2 van Loon et al.
Z (cf.discussioninMaeder,Grebel&Mermilliod1999). The uncertainties in the extracted spectrum stem
⊙
This offers the possibility to assess the effect metallicity mostly from inaccuracies in the sky-subtraction. The
has on the dust content and on the heating and cool- statistical scatter was quantified, upon which the error-
ing processes, and to study these in environments that bars(e.g.,asplottedinFig.4)arebased. However,larger
aremoresimilartothoseprevailingintheearlyUniverse deviations may result from complex spatial structure of
than the available Galactic examples (cf. Oliveira 2009). the skyemission,andthis isnotpossibleto quantify. As
The Spitzer Space Telescope marriessuperbsensitivity thisisimportantforanoveralljudgmentofthereliability
with exquisite imaging quality, able to detect the far-IR of the spectrum, a “quality” flag was decided on the ba-
emission from a significant fraction of the total popula- sis ofa subjective assessmentofthe reliabilityof the sky
tionsofYSOs,massiveredsupergiants(RSGs),etcetera. subtraction and spectrum extraction. This is listed in
The telescope also carried a facility, the MIPS-SED, to Table 2, along with other descriptors of the MIPS-SED
obtain spectra at 50–100µm, and we used this to target data. Where possible inouranalysis,we haveaccounted
representative samples of luminous 70-µm point sources formeasuredvariationsbetweenadjacentspectralpoints
in the LMC and SMC. The LMC spectra are presented in the computation of errorbars on derived quantities,
in paper I of this two-part series (van Loon et al. 2010); rather than to rely solely on the errorbars derived from
here we presentthe results ofthe SMC observationsand the statistical noise.
a comparison with the LMC results. In our analysis of the MIPS-SED data we shall also
make use of associated photometry, from S3MC at 24
2. OBSERVATIONS and 70 µm with MIPS.
2.1. Data collection and processing 2.2. Target selection
Ourdatasetcompriseslow-resolutionspectraobtained
The targets were selected on the basis of the following
using the Spectral Energy Distribution (SED) mode of
criteria: (i)pointsourceappearanceat70µm, and(ii)a
the Multiband Imaging Photometer for Spitzer (MIPS;
minimum flux density at 70 µm of F (70) > 0.3 Jy. To
Rieke et al. 2004) onboard the Spitzer Space Telescope ν
reduce the large sample of potential targets to within a
(Werner et al. 2004), taken as part of the SMC-Spec
reasonable time request, a further requirement was that
program (PI: G.C. Sloan). The spectra cover λ=52–93
at the time of proposal submission there was a Spitzer
µm, at a spectral resolving power R ≡ λ/∆λ = 15–25
IRSspectruminthearchiveorplannedtobetaken. Due
(two pixels) and a cross-dispersion angular resolution of
tothelimitationsonobservingtimewecouldnotsample
13–24′′ Full-Width at Half-Maximum (sampled by 9.8′′
awide rangeofobjecttypesand/orphotometricproper-
pixels). The slit is 20′′ wide and 2.7′ long, but 0.7′ at
ties (as we did in the LMC sample), and we have delib-
one end of the slit only covers λ > 65 µm as a result
erately restricted ourselves to generally luminous YSOs
of a dead readout. To place the angular scales into per-
and RSGs. They do, however, sample different regions
spective, 20′′ ≡6 pc at the distance of the SMC. This is
within the SMC (see Fig. 1). Table 3 summarizes the
characteristic of a SNR, star cluster, or molecular cloud
MIPSphotometricpropertiesoftheselectedtargets,and
core; it is smaller than a typical Hii region, but larger
Fig. 3 shows them in the F (70) vs. F (70)/F (24) dia-
than a typical PN. ν ν ν
gram in comparison to the LMC sample.
The target list, Table 1, is described in §2.2 and §3,
and their distribution on the sky is displayed in Fig. 1. 3. COMMENTSONINDIVIDUALOBJECTS
The background spectrum was measured at one of four
Intheremainderofthispaper,weshallrefertoobjects
possible chop positions, chosen to be free of other dis-
from the Henize (1956)catalogas “N[number]”; the full
crete sources of 70-µm emission. This depends on the
designation would be “LHA115-N[number]”. Sources
time of observation. Fig. 2 shows 70-µm close-ups, ex-
with only an S3MC designation are abbreviated follow-
tracted from the combined SAGE-SMC Spitzer Legacy
Program(Gordonetal.2010,inpreparation)andS3MC ing the IRAS convention(where the last digit of the RA
part derives from decimal minutes). Table 1 describes
Spitzer survey (Bolatto et al.2007),with the Astronom-
allMIPS-SEDtargets,withliteraturereferenceschecked
ical Observation Request (AOR) footprints overlain.
until Summer 2009.
The raw data were processed with the standard
Some targetshavebeen studied before,and briefsum-
pipeline version S16.1.1, and the spectra were extracted
maries of their nature are given below. Essentially
andcalibratedusing the dat software,v3.06(Gordonet
nothing was known previously about IRAS00429−7313
al. 2005). In some cases, small shifts in the centroids
(#1) and S3MC00540−7321 (#6). All sources except
ofwell-detected,unresolvedspectrallinesarenoticeable;
BMB-B75 had alreadybeen recognisedas 170-µmpoint
these always amount to less than about a micrometer,
sources in Wilke et al. (2003). The S3MC sources
i.e. within the spectral sampling, and no attempt was
were first identified as candidate YSOs by one of us (J.
made to correct for this (cf. figure 5 and its discus-
Oliveira);allYSOsinoursamplehavebeenconfirmedto
sion in §5.1.2). Spectra were extracted from the on-off
be YSOs with ground-basedIR observationsand Spitzer
background-subtracted frame. The extraction aperture
IRS spectra (Oliveira et al., in prep.).
wasfivepixelswideinthecross-dispersiondirection,and
the (remaining) background level was determined in a-
3.1. IRAS00430−7326 (#2)
few-pixel-wide apertures at either side of, and at some
distance from, the extraction aperture. The extracted TheIRsourceisseenprojecteduponanebulouscluster
spectrum was corrected to an infinite aperture and con- (Bica&Schmitt1995;deOliveiraetal.2000),andisem-
verted to physical units, providing an absolute flux cali- beddedwithinthecompactHiiregionSMC-N10(Henize
brated spectrum (cf. Lu et al. 2008). 1956) = DEMS11 (Davies, Elliott & Meaburn 1976);
Far-IR spectra of compact sources in the SMC 3
TABLE 1
Description ofcompactsources in the SMCastargetsforMIPS-SED.
# Principalnamea Alternativename Objecttypeb RAandDec(J2000) References
1 IRAS00429−7313 YSO 04451.86−725734.2 36,42,48,55
2 IRAS00430−7326 N10,LIN60 YSO 04456.30−731011.6 1,2,6,9,11,16,21,22,40,41,42,48,55,57
3 S3MC00464−7322 c IRAS00446−7339 ? YSO 04624.46−732207.3 4,16,27,30,42,46,47,48,55
4 GM103 IRAS00486−7308, [GB98]S10 RSG 05030.62−725129.9 17,18,19,20,39,48,49,50,55
5 BMB-B75 RSG 05212.82−730852.8 3,7,16,50
6 S3MC00540−7321 d YSO 05402.30−732118.7 42,55
7 S3MC00541−7319 e YSO 05403.36−731938.3 2,42,55
8 S3MC01051−7159 f IRAS01035−7215 ? YSO 10507.25−715942.7 1,2,8,11,16,21,33,41,42,44,45,48,55
9 IRAS01039−7305 DEMS129,[MA93]1536 YSO 10530.22−724953.8 9,14,36,41,42,48,50,55,57,58
10 IRAS01042−7215 [GB98]S28 YSO 10549.30−715948.8 18,20,48,50,55
11 S3MC01070−7250 g IRAS01054−7307 ? YSO 10659.66−725043.1 2,9,41,42,55
12 N81 IRAS01077−7327, DEMS138 Hii 10912.67−731138.4 1,2,6,9,10,11,12,13,15,16,21,22,23,24,
25,26,28,29,30,31,32,34,37,38,41,43,
48,51,52,53,54,55
13 S3MC01146−7318 h (inN84) YSO/Hii 11439.38−731829.3 2,5,16,35,41,42,55,56
References. — 1. Beasleyet al. (1996); 2. Bica& Schmitt(1995); 3. Blanco, McCarthy & Blanco(1980); 4. Bot et al. (2007); 5. Bratsolis,
Kontizas&Bellas-Velidis(2004); 6. Charmandaris,Heydari-Malayeri&Chatzopoulos(2008);7. Cionietal.(2003); 8. Copetti(1990);9. Davies,
Elliott&Meaburn(1976);10. Dennefeld&Stasin´ska(1983);11. deOliveiraetal.2000;12. Dufour&Harlow(1977);13. Dufour,Shields&Talbot
(1982); 14. Evans et al. (2004); 15. Filipovi´cet al. (1997); 16. Filipovi´cet al. (2002); 17. Groenewegen(2004); 18. Groenewegen & Blommaert
(1998); 19. Groenewegenetal.(1995); 20. Groenewegenetal.(2000); 21. Henize(1956); 22. Henize&Westerlund(1963); 23. Heydari-Malayeri,
LeBertre&Magain(1988); 24. Heydari-Malayerietal.(1999); 25. Heydari-Malayerietal.(2002); 26. Heydari-Malayerietal.(2003); 27. Hodge
(1974); 28. Indebetouw, Johnson & Conti (2004); 29. Israel & Koornneef (1988); 30. Israel et al. (1993); 31. Kennicutt & Hodge (1986); 32.
Koornneef&Israel(1985);33. Kron(1956);34. Krtiˇcka(2006);35. Lindsay(1961);36. Loupetal.(1997);37. Mart´ın-Hern´andez,Vermeij&van
derHulst(2005);38. Martinsetal.(2004);39. McSaveneyetal.(2007);40. Meynadier&Heydari-Malayeri(2007);41. Meyssonnier&Azzopardi
(1993);42. Oliveiraetal.(inprep.);43. Pageletal.(1978);44. Pietrzyn´ski&Udalski(1999);45. Pietrzyn´skietal.(1998);46. Rubio,Lequeux&
Boulanger(1993);47. Rubioetal.(1993b)48. Schwering&Israel(1989);49. vanLoonetal.(2001);50. vanLoonetal.(2008);51. Vermeij&van
derHulst(2002);52. Vermeijetal.(2002);53. Wilcots(1994a);54. Wilcots(1994b);55. Wilkeetal.(2003);56. Wisniewski&Bjorkman(2006);
57. Woodetal.(1992);58. Zijlstraetal.(1996).
a Namesofthetype“N[number]”are“LHA115-N[number]”infull.
b Usedacronyms: Hii=regionofionizedHydrogen;RSG=RedSupergiant;YSO=YoungStellarObject.
c AbbreviationofS3MCJ004624.46−732207.30(followingtheIRASconvention).
d AbbreviationofS3MCJ005402.30−732118.70(followingtheIRASconvention).
e AbbreviationofS3MCJ005403.36−731938.30(followingtheIRASconvention).
f AbbreviationofS3MCJ010507.25−715942.70(followingtheIRASconvention).
g AbbreviationofS3MCJ010659.66−725043.10(followingtheIRASconvention).
g AbbreviationofS3MCJ011439.38−731829.26(followingtheIRASconvention).
TABLE 2 TABLE 3
Description of MIPS-SEDdataof compact MIPS photometricdataof MIPS-SEDtargetsin the
sourcesin the SMC. SMC.
# AORKey Integrationa Quality Extraction # AORTarget Fν(24) σ(24) Fν(70) σ(70)
[mJy] [mJy] [mJy] [mJy]
1 27535360 6×10 good on–off
2 27535104 8×3 good on–off 1 IRAS00429−7313 588.27 0.20 1291 6
3 27532288 7×10 good on–off 2 IRAS00430−7326 717.69 0.54 7168 30
4 27536640 20×10 ok on–off 3 S3MC00464−7322 166.67 0.10 1191 12
5 27536384 20×10 good on–off 4 GM103 100.00 0.07 121 8
6 27532032 5×10 good on–off 5 BMB-B75 52.55 0.06 481 4
7 27531776 3×10 good on–off 6 S3MC00540−7321 472.56 0.17 1502 6
8 27531520 8×3 good on–off 7 S3MC00541−7319 812.79 0.27 2423 11
9 27536128 3×10 good on–off 8 S3MC01051−7159 2402.03 0.55 7138 40
10 27535872 6×10 good on–off 9 IRAS01039−7305 796.12 0.24 2353 6
11 27531264 20×10 good on–off 10 IRAS01042−7215 569.61 0.14 1351 10
12 27535616 8×3 good on–off 11 S3MC01070−7250 49.71 0.07 440 2
13 27531008 3×10 good on–off 12 N81 1705.20 0.32 9190 41
13 S3MC01146−7318 144.06 0.18 4340 45
a Totalon-sourceintegrationtime,Ncycles×t(s).
Henize & Westerlund (1963) estimated a nebular mass
∼4 M⊙. It was detected as a compact radio continuum This IR object is possibly associated with the dark
source by Filipovi´c et al. (2002). Meynadier & Heydari- cloud #7 in Hodge (1974), which was also the strongest
Malayeri (2007) recognised it as a very-low excitation 12CO(1–0) and 13CO(1–0) detection in the ESO/SEST
“blob”,possibly poweredby less massive stars;however,
key programme (Israel et al. 1993), SMC-B1, which is
Charmandaris,Heydari-Malayeri&Chatzopoulos(2008)
split up into SMC-B12 and 3 each containing an esti-
conclude that the mid-IR properties resemble simply a mated few 104 M (Rubio, Lequeux & Boulanger 1993;
⊙
scaled-down version from more prominent Hii regions.
Rubio et al. 1993b). The mm-continuum emission from
B12 was much brighter and more extended than that
3.2. S3MC00464−7322 (#3)
from B13 (Bot et al. 2007).
4 van Loon et al.
Fig.1.—All13MIPS-SEDpointsourcesplottedontopoftheMIPS70-µmSAGE-SMC+S3MCimage. Themainstar-forming,dense-
gas body of the Small Magellanic Cloud stretches from North-East to South-West. The compact Hii region N81 (#12) and Hii regions
N83+N84(#13)assumeratherisolatedpositionswithintheShapleyWing(Shapley1940), totheEast.
3.3. GM103 (#4) d (Cioniet al. 2003). A 3–4µm spectrum was presented
invanLoonetal.(2008);itresemblesthatofacoolgiant
GM103isidentifiedwithIRAS00486−7308,nearN36
star. There is a compact radio continuum source within
(Henize1956). Aground-basedN-bandspectrumshowed
a few arcsec, which was only detected at 2.37 GHz (not
the 10-µm silicate feature in emission (Groenewegen et
at 1.42,4.80 or 8.64 GHz), at a level of 8 mJy (Filipovi´c
al. 1995); modelling of the SED yielded a mass-loss rate
et al. 2002).
of M˙ ∼ 10−5 M yr−1 and a luminosity near the AGB
⊙
maximum associated with the Chandrasekhar limit for
3.5. S3MC00541−7319 (#7)
the core mass. Groenewegen & Blommaert (1998) pre-
sented an optical spectrum, of late-M spectral type; ap- This source is seen projected against the star cluster
parently cooler than the M4 type which Groenewegen [H86]144 (Bica & Schmitt 1995).
has assumed in other works. McSaveney et al. (2007)
presented an analysis of the pulsation properties. They 3.6. S3MC01051−7159 (#8)
confirmed the long pulsation period of P =1070 d, at a
ThisIRsourcesitsinthe middleoftheclusterOGLE-
large near-IR amplitude of ∆K = 1.3 mag, and a lumi-
CLSMC147 (Pietrzyn´ski et al. 1998), which has an es-
nosity near the AGB maximum. From pulsation mod-
timated age of t ≈ 13+12 Myr (Pietrzyn´ski & Udalski
elling they derived a (current) mass of Mpuls ≈ 6 M⊙. −7
Van Loon et al. (2008) presented a 3–4 µm spectrum, 1999). There are several compact Hii regions nearby,
resembling that of a cool luminous star. the closest of which is #1520 in Meyssonnier & Az-
zopardi (1993), at 6′′. These sources form part of the
much larger, loose stellar association NGC395 (Kron
3.4. BMB-B75 (#5)
1956;Bica&Schmitt1995)associatedwithN78(Henize
An M6-type star (Blanco, McCarthy & Blanco 1980), 1956) = DEMS126 (Davies et al. 1976), which contains
it has a luminosity near the AGB maximum, and shows a further two compact Hii regions (Indebetouw, John-
Mira-typevariabilitywithaverylongperiodofP =1453 son & Conti 2004) and is an extended source of radio
Far-IR spectra of compact sources in the SMC 5
1 2 3 4 5 6
7 8 9 10 11 12
13
Fig.2.— Close-upsof the 70-µm emissioncentred oneach of the 13 MIPS-SED targets, withoverplotted the AOR footprints (on- and
off-sourceslitorientations). AllimageshaveNorthupandEasttotheleft,andmeasure10′ oneachside. Theintensityscaleislinear,but
adjustedindividuallysuchastofacilitateanassessmentoftherelativebrightnessofthetargetcomparedtothebackground.
spectrum (van Loon et al. 2008).
3.8. IRAS01042−7215 (#10)
This very red IR source (Groenewegen & Blommaert
1998) has long been considered as a candidate dust-
enshroudedAGBstarorRSG.Groenewegenetal.(2000)
modelled the SED assuming a spectral type of M8, but
the obtained luminosity is rather low (103 L ). This
⊙
raisesdoubtsaboutitbeingacool,evolvedstar. Indeed,
it was recognized as a YSO by van Loon et al. (2008),
on the basis of water ice absorption and Brα and Pfγ
emission lines on a very red continuum in the 3–4 µm
spectrum.
3.9. S3MC01070−7250 (#11)
Thissourcecanbeidentifiedwithafaint,compactHα
emission-lineobject,#1607inMeyssonnier&Azzopardi
(1993), which is situated in a stellar association within
the Hii region DEMS133 (Davies et al. 1976; Bica &
Schmitt 1995).
3.10. N81 (#12)
Fig. 3.—Fν(70)vs.Fν(70)/Fν(24)diagram,withMIPSphotom- The IR source IRAS01077−7327 is associated with
etryfromtheS3MCcatalogfortheSMC-SpecMIPS-SEDtargets the rather isolated,very brightcompact Hii regionN81
(squares). TheSAGE-SpecMIPS-SEDtargetsintheLMCarealso
(Henize 1956) = DEMS138 (Davies et al. 1976). A
overplotted(dots).
comprehensive HST optical imaging and spectroscopic-
continuum emission (Filipovi´c et al. (2002). Curiously,
imaging study was performed by Heydari-Malayeri et
at7′′,oppositeto[MA93]1520,isthebackgroundgalaxy
al. (1999, 2002), who identified several O-type stars
2MFGC779 (Mitronova et al. 2004). within a few arcseconds in the core of the Hii region.
These stars are young, possible pre-main sequence, in-
3.7. IRAS01039−7305 (#9)
ferred from their low luminosity and weak stellar winds.
This red IR object is a compact source of Hα emis- Nonetheless,they inducea“Champagneflow”(Heydari-
sion, #1536 in Meyssonnier & Azzopardi (1993), identi- Malayeri et al. 1999), with a shock-compressed front in
fied with DEMS129 (Davies et al. 1976), and probably one direction (West), and an ionized tail in the oppo-
an early-B type stellar object (#2027 in Evans et al. site direction (East). This picture was confirmed with
2004). It was recognized as a YSO for its Brα and Pfγ high-resolutionradio continuumimages,by Indebetouw,
emission lines on top of a red continuum in the 3–4 µm Johnson&Conti(2004)andMart´ın-Hern´andez,Vermeij
6 van Loon et al.
& vander Hulst (2005). The latter estimated anionized YSO-like objects near more mature massive stars, e.g.,
mass of M ≈56 M , rather less than Henize & West- theproplydσOri-IRS1(vanLoon&Oliveira2003). The
ion ⊙
erlund (1963), who had estimated M ≈ 660 M from identificationofthissourceinN84suggeststhatstarfor-
ion ⊙
optical spectroscopy. The detection of H in its excited mation is still on-going there.
2
S1(1–0) transition at λ=2.121 µm (Koornneef & Israel
1985) confirmed the presence of a mild shock. CO was 4.2. Classification of the MIPS-SED spectra
alsodetected(Israeletal.1993),thoughweakly. Vermeij
A simple classification scheme based on the spectral
et al. (2002) presented ISO SWS and LWS spectra, and
appearance,“TheKeeleSystem”,wasfirstintroducedin
we compare with their measurements (see §5).
Paper I. The primary spectral type is defined as follows:
3.11. S3MC01146−7318 (#13) • An upper-case letter denotes the continuum slope,
foraspectrumexpressedinF asafunctionofλ: C
The nature of this source is unclear. It is seen pro- ν
= rising (e.g., relatively cold dust); F = flat (this
jectedagainsttherichstarclusterNGC460(Kron1956)
includes spectra with a peak mid-way the MIPS-
=IRAS01133−7333(Loupetal.1997),witharatherun-
certainageestimatedtobe t∼20(±14)×106 yr(Hodge SED range); W = declining (e.g., relatively warm
dust);
1983). The cluster has plenty of (candidate) Be stars
(Wisniewski&Bjorkman2006),yetthenearestrecorded
• Following the upper-case letter, a number denotes
objects are 6′′ away. One of these is an emission-line
the presence of the [Oi] and [Oiii] lines: 0 =
star, #507 in Lindsay (1961) = #1792 in Meyssonnier
no oxygen lines are present; 1 = the [Oi] line is
& Azzopardi (1993), which is a candidate Be star (Wis-
present, but the [Oiii] line is not; 2 = both [Oi]
niewski&Bjorkman2006). Unrelatedtothatsourcebut
and [Oiii] lines are present; 3 = the [Oiii] line is
notmuchfurtheraway(8–14′′)isanextendedradiocon-
present, but the [Oi] line is not.
tinuum source (Filipovi´c et al. 2002). The cluster is em-
bedded in the bright Hii region N84A (Henize 1956) = Asecondaryclassificationisbasedonadditionalfeatures:
DEMS151(Daviesetal.1976),forwhichCopetti(1990) a lower-case letter “b” may follow the primary type in
estimated an ionized mass of Mion ≈ 1200 M⊙. Testor the presence of a bump in the λ∼70–80 µm region.
& Lortet (1987), pioneering CCD imaging, argued for We have classified all spectra (Table 4), erring on the
an extended period of sequential star formation in this side of caution with respect to the detection of spectral
part of the SMC, and the presence of unevolved O stars lines. So, for instance, an object with spectral type C0
in the N84 region was taken as a sign of continued star may still display weak oxygen lines in a higher-quality
formation. Weak 12CO(1–0) emission appears to be as- spectrum.
sociatedwith it, though the bulk of emissionarises from
further North-West (Mizuno et al. 2001). Hodge (1974) 5. DISCUSSION
identified a dark cloud in this region. More detailed in- Wefirstdiscusstheoxygenfine-structureemissionlines
vestigationsofthemolecularanddustdistributioninthe (§5.1), followedby the nitrogenline (§5.2), and then the
N83/N84regionwereconductedbyBolattoetal.(2003) dustcontinuum(§5.3)anddiscretefeaturespossiblydue
and Lee et al. (2009), but S3MC01146−7318 does not to ice, molecules, or minerals (§5.4). At the end (§5.5),
appear to be associated with any conspicuous feature in we summarize the population MIPS-SEDcharacteristics
their maps. andcomparewiththoseintheLMCpresentedinPaperI.
4. RESULTS 5.1. Oxygen
The MIPS-SED spectra of all 13 targets are presented
The diagnosticvalue ofthe oxygenlines wasdescribed
inFig.4. Alltargetswereconsiderablybrighterthanthe
in detail in Paper I (cf. Tielens & Hollenbach 1985). We
surroundingskyemission,exceptGM103(of whichnev-
recall that the [Oi] line at λ = 63 µm is an important
ertheless a reliable spectrum could be extracted). Table
cooling line in relatively dense and neutral or weakly-
4 summarizes properties derived from the Spitzer data.
ionizedgas,anditisenhancedifslowshocksarepresent.
One or two fine-structure emission lines may be seen,
The[Oiii]lineatλ=88µmisameasureofthe electron
[Oi] 3P(1–2) and [Oiii] 3P(0–1), at λ = 63.2 and 88.4
density in ionized gas such as that which occupies Hii
µm, respectively. These are discussed in §5.1. There is
regions.
no convincing detection of the [Niii] 2P(1/2–3/2) tran-
ThelineluminositiesintheSMC sourcescanbe found
sition at λ=57.3 µm, and we discuss the implication in
in Table 4, for an assumed distance of 60 kpc. They
§5.2. There is evidence for additional discrete features
were computed in identical fashion to those of the LMC
in the spectra of some objects, but their identification is
sourcesinPaperI,bysumming thethreespectralpoints
uncertain. They are discussed in §5.4. The slope of the
centered on the line (data registered at 60.81,62.52 and
continuum is an indication of the dust temperature and
64.23 µm for the [Oi] line, and at 86.46, 8817 and 89.88
is discussed in §5.3.
µmforthe[Oiii]line)aftersubtractingacontinuumob-
tainedbylinearinterpolationbetweenthespectralpoints
4.1. Clarification of the nature of S3MC01146−7318
immediatelyadjacenttotheintegrationinterval. Theer-
The MIPS-SED spectrum of this object (Fig. 4, #13) ror was computed by adding in quadrature the errors in
looks very much like that of N81 (Fig. 4, #12), so it is the three spectral points that were summed, and three
likely an (ultra)compact Hii region too. Although N84 timestheerrorinthemeanofthetwocontinuumanchors
has been suggested to be a somewhat evolved molecu- (to account for the error in the continuum estimate at
lar cloud complex, it is not unprecedented to encounter each of the three spectral points).
Far-IR spectra of compact sources in the SMC 7
Fig.4.—MIPS-SEDspectraofall13targetsintheSMC.Verticaldashedlinesindicatethepositionsofthe[Oi]and[Oiii]fine-structure
emissionlinesatλ=63and88µm,respectively.
5.1.1. Oxygen in star-forming regions and YSOs radicallydifferentfromthelinefluxeswedetermined(Ta-
ble 4).
Theoxygenlinesintheobjectsassociatedwithstarfor-
Vermeij et al. did not publish the ISO-LWS spectrum,
mation are weak. In fact, on closer inspection, the only
soweextractedthepipeline-processedspectrumfromthe
really convincing detections are in S3MC01051−7159,
ISOArchive. Itiscomposedofindividualsegments,with
N81,andS3MC01146−7318(Fig.4,#8,#12and#13),
large jumps in flux level from one segment to the other.
where both the [Oi] and [Oiii] lines are clearly dis-
We used the overlap regions to calibrate them relative
cernible. IntheLMCsourcesassociatedwithstarforma-
to one another, and applied a global correctionfactor to
tion,[Oi]wasmoreoftendetected;thisispartlybecause
bring the flux level in line with that in our MIPS-SED
the detection threshold (in L ) was about half that in
⊙ spectrum. TheresultisshowninFig.5,bothattheISO-
theSMCobservations,butmostLMCsourceswereabove
LWS spectral resolution and convolved with a Gaussian
the SMC threshold too.
of width 3.5 µm (MIPS-SED spectral resolution). The
5.1.2. Shocks and ionized gas in the compact Hii regions continuum is more reliably measured in the MIPS-SED
N81 and S3MC01146−7318 in N84 spectrum;theoxygenlinesarebetterresolvedintheISO-
In the case of N81, the MIPS-SED aperture comfort- LWS spectrum. The [Oiii] λ88-µm line was recorded in
ablyincludesthe entireradiosourceassociatedwiththis twooverlappingLWSsegmentsbutatratherdifferentin-
Hiiregion. PreviousobservationsmadebyVermeijetal. tensities; nevertheless the convolvedline profile matches
(2002)usingtheISO-LWSemployedamuchlargeraper- that in the MIPS-SED spectrum quite well. The [Oi]
ture, with a circular 80′′ diameter; there is not much λ63-µmlineismorediscrepant,butthisisduetoabump
more emission than captured in the MIPS-SED slit and intheISO-LWSspectrumlongwardoftheoxygenline—
thusthespectraarecomparable. Vermeijetal.measured it could be due to water ice, but the correspondingISO-
F([OI])=37×10−16Wm−2andF([OIII])=74×10−16 LWS segmentdrops suspiciouslyat that side, and it was
W m−2, or 414 and 829 L , respectively, which is not not recorded in the MIPS-SED spectrum.
⊙
8 van Loon et al.
TABLE 4
Quantitiesderived fromthe MIPS-SEDspectra of compactsources in the SMC.
# AORTarget Typea αb Tdust c L([OI])d L([OIII])d L(FIR)e Notes
[K] ———————— [L⊙] ————————
1 IRAS00429−7313 F0 −0.007±0.021 48+−11 69±23 <15 1.814−+00..00438×104
2 IRAS00430−7326 F0 −0.022±0.010 48+−11 67±31 80±25 1.323−+00..003218×105 f
3 S3MC00464−7322 C1 0.673±0.020 37+−11 55±17 <17 3.06−+00..1186×104 f?
4 GM103 C0 0.78±0.06 35+−11 <19 54±7 4.44−+00..3217×103
5 BMB-B75 C0 0.32±0.11 42+−22 <14 41±10 6.2−+00..54×103
6 S3MC00540−7321 F0 −0.016±0.036 48+−11 <23 47±18 1.770−+00..00437×104 f?
7 S3MC00541−7319 W0b −0.147±0.017 51+−11 <18 43±22 3.34−+00..0065×104
8 S3MC01051−7159 W2b −0.080±0.018 50+−11 287±56 51±45 1.341−+00..002263×105
9 IRAS01039−7305 C0b 0.121±0.029 45+−11 47±28 <25 3.54−+00..1110×104
10 IRAS01042−7215 F3b 0.21±0.05 44+−11 <23 63±17 2.66−+00..0098×104
11 S3MC01070−7250 C1 0.32±0.07 42+−11 22±7 14±9 6.84−+00..2274×103
12 N81 F2 0.21±0.05 44+−11 518±27 748±38 1.56−+00..0055×105
13 S3MC01146−7318 C2b 0.35±0.05 41+−11 539±22 405±28 1.038−+00..00439×105
a Thespectralclassificationschemeisdescribedin§4.2.
b Thespectralslopeisdefinedas: α≡2.44 FFνν((8855))−+FFνν((5555)).
c Thedusttemperaturewasestimatedfromthespectralslopeasdescribedin§5.3.1.
d Upperlimitstothelineluminositiescorrespondto1-σvalues.
e Thefar-IRluminositywasestimatedbasedonthedusttemperatureasdescribedinPaperI,assumingadistanceof60kpc.
f Possiblebroadcrystallinewatericebandbetweenλ=60–70µm.
the photo-ionizedgas, but this is not resolvedin our ob-
servations). ThedetectionoftheH S1(1–0)lineat2.121
2
µm in N81 (Koornneef & Israel 1985) is also indicative
of a mild shock.
The source S3MC01146−7318, in N84, is virtually
identical to N81, only a little bit fainter. We thus con-
jecture that it is of a similar nature, i.e. a compact Hii
region with a champagne flow structure. Stars must re-
cently have formed here, and are now blowing the cloud
apart through their mechanical and radiative feedback.
5.1.3. Oxygen in evolved objects
No oxygen line emission is seen in either GM103 or
BMB-B75 — the apparently significant detection of
[Oiii] is an anomaly caused by the poor signal near
the edge of the MIPS-SED range (these are two of the
faintest targets). This is not very surprising as RSGs
lack a potent excitation mechanism for these lines, and
even in the LMC the [Oi] line was tentatively detected
only in the most luminous RSG, WOHG064 (Paper I).
5.2. Nitrogen
Evidencehas been presentedofa low nitrogencontent
Fig. 5.—Top: ISO-LWSspectrumofN81. Bottom: Comparison
in relation to the abundance of other common metals
betweentheMIPS-SEDspectrumandtheISO-LWSspectrumafter
such as oxygen, if the overall metallicity is low (Rubin
convolvingthelatterwiththeMIPS-SEDinstrumentalprofile.
et al. 1988; Simpson et al. 1995; Roelfsema et al. 1998;
The ISO-LWS spectrum of N81 (Fig. 5) also includes Rudolph et al. 2006). This hinges on evidence in the
the [Oiii] λ52-µm line, and the important [Cii] λ158- Magellanic Clouds, which is rather limited. Any addi-
µm cooling line. The latter is probably sampled from a tionalorrefinedmeasurementsinthe MagellanicClouds
larger region than that sampled by the MIPS-SED; the are most welcome to solidify — or weaken — the ob-
[Oi] λ63-µm line is expected to dominate the cooling in served trend.
the denser inner part of the cloud (see §5.5.2). InPaperIweusedthebestMIPS-SEDspectraofLMC
The strength of the oxygen lines of such different ion- compact Hii regions to estimate that the nitrogen-to-
ization stage indicates the coincidence of both ionized oxygenratio, N(N)/N(O) .0.1 and possibly as low
LMC
gas and shocks. This is consistent with the idea of the as . 0.03, which is much below the 0.1–0.4 typical in
“champagne flow” (note that in this scenario there is a the Galactic Disc. Like in the LMC sample, no nitrogen
distinct spatialsegregationbetweenthe shockedgasand line is detected in the SMC sample. Can we still place
Far-IR spectra of compact sources in the SMC 9
interesting upper limits? This depends on the strength Bar. It thus appears that the diffuse ISM is cold, below
of the reference oxygenline, preferably the [Oiii] line at 30 K, but the temperature rises above 40 K within the
88 µm as the ionization potential of the lower level of cores of star-forming regions. Compact sources in star-
this transition is similar to that of the [Niii] line at 57 forming regions in the SMC may have more compact,
µm. The [Oiii] is not phenomenally bright in any SMC hence warmer dust envelopes than such sources in the
source and we are anticipating N(N)/N(O) < 0.05. LMC, possibly as a result of diminished attenuation of
SMC
But the spectra of N81 and S3MC01146−7318are very the destructive effect of the interstellar radiationfield in
good, displaying strong [Oiii] lines. the metal-poor, dust-depleted SMC.
Assuming a detection level of the [Niii] line at 57 µm In contrast, the dust in the two RSGs in our SMC
similar to the 1-σ noise level near the [Oi] line at 63 sample is much cooler than that in similar sources in
µm,wewouldget[Niii]/[Oiii].0.04inN81and.0.05 the LMC (Paper I). In fact, the dust temperature of the
in S3MC01146−7318. To place this improvement in RSGs in the SMC is indistinguishable from that of the
context, previous data on N81 obtained with ISO-LWS YSOs. We explore this in more detail in the following
by Vermeij et al. (2002) yielded a limit of F([NIII]) < subsection.
38 × 10−16 W m−2, or 426 L , and thus (see above)
⊙ 5.3.2. Cool dust around red supergiants: swept-up ISM?
[Niii]/[Oiii]<0.5. Vermeij & van der Hulst (2002) esti-
matedanelectrondensityn ∼few×102 cm−3. Atsuch The two bright RSGs in the LMC, WOHG064 and
e
modestelectrondensitywewouldderive(Liuetal.2001) IRAS05280−6910, had MIPS-SED spectra which de-
N(N)/N(O) . 0.03. However, if the true electron clined even more steeply than a Rayleigh–Jeans tail of
SMC
density is higher, the limit may be reduced to as low as a hot black-body, indicative of dust emission but at a
N(N)/N(O) <0.01. This compares favourably with temperatureover100K.ThetwoRSGsintheSMCsam-
SMC
the abundance estimates obtained by Dufour, Shields & ple exhibit far-IR emission associated with dust of only
Talbot (1982), N(N)/N(O)SMC = 0.036, and suggests Tdust =35–42K.Suchcooldustwasoccasionallyseenin
that,indeed,thenitrogencontentinthemetal-poorSMC less massive evolved stars in the LMC, notably carbon
is even lower than that in the LMC. stars,andtheexplanationputforwardthenwasthatthis
is swept-up interstellar dust rather than dust produced
5.3. Dust continuum by these stars (see Paper I).
Perhaps this scenario also applies to the two RSGs in
AsinPaperI,wedefinethecontinuumslopeasfollows:
the SMC sample (cf. van Loon et al. 2008): if the far-
Fν(85)−Fν(55) IR emission from the warm circumstellar dust is weak
α≡2.44 , (1)
F (85)+F (55) —due to a low dust-to-gasratioin the metal-poor wind
ν ν
— then the far-IR emission may be dominated by that
suchthatα=0inaflatF spectrumandα=−1inthe
ν from cool ISM dust (though the dust-to-gas ratio in the
Rayleigh–Jeans approximation to the long-wavelength
ISM, too, is low in the metal-poor SMC). In the case
tailofaPlanckcurve. Thesevalueswerecomputedfrom
of GM103, the sky surrounding the point source shows
the mean values of the three spectral points centered at
complex far-IR emission arising from the ISM near this
55 and 85-µm, respectively, and are listed in Table 4.
star (see Figs. 1 & 2). It would not therefore come as a
The errorwas computed by propagationof the standard
surprise that this star has collected ISM dust in a bow-
deviations in the two sets of three spectral points used.
shock as it has plowed through the relatively dense ISM
5.3.1. Dust temperatures in this part of the SMC. This does not appear to be the
casefortheotherRSG,though: BMB-B75isinfactseen
The dust temperature characterizing the MIPS-SED
against a relatively “empty” bubble (see Figs. 1 & 2).
range is a powerful discriminant between warm circum-
Alternatively, the dust may have formed — or grown
stellar envelopes of evolved stars and cold molecular
(by condensationonto pre-existing grains)— within the
clouds in star-forming regions. We have estimated the
bow-shock itself. Indeed, these slow shocks (few tens of
dusttemperaturebycomparisonwithasinglegrey-body: km s−1) bear similarity to those traveling through the
Fλ =Bλ(Tdust)(cid:0)1−e−τλ(cid:1), (2) molecular atmospheres of dust-producing pulsating red
giant stars (Bowen 1988). This would naturally explain
where Bλ(Tdust) is a Planck emission curve at the dust dust near BMB-B75 in the absence of diffuse ISM dust.
temperature T , the optical depth τ = τ λ−β, and
dust λ V 5.4. Ice, molecules, and minerals
β = 1–2, here assumed β ≡ 1.5 (cf. Goldsmith, Bergin
& Lis 1997). Using the curve in figure 6 in Paper I, Thebroad(severalµm)emissionfeaturesseeninsome
we converted the α values into dust temperatures. The ofthespectraarealmostcertainlyduetoeitherminerals
results are listed in Table 4. or ices. At the spectral resolution of MIPS-SED, the
The temperatures of YSOs and compact Hii regions detection of molecules is rare, though not entirely ruled
in our SMC sample are T ≈ 37–51 K, or T ≈ 45 out (cf. Paper I and §5.4.2).
dust dust
K on average. This is considerably warmer than sources
5.4.1. Water ice
in the LMC (Paper I), which have T ≈32–44 K.
dust
On the basis of IRAS data, at lower spatial resolution Crystalline water ice has significant opacity in the 60–
than the Spitzer data around 70 µm, Stanimirovi´c et al. 70 µm region. The feature is generally broad, but the
(2000)estimated T ≈29 K for N81 (which they may peakwavelengthandprofileshapecanvaryconsiderably
dust
not have resolved), but reaching T ≈ 37 K in N88; (e.g., Malfait et al. 1999; Dijkstra et al. 2006).
dust
elsewhere in the Shapley Wing they derived T ≈26– The bump between λ ∼ 60–70 µm in the spectrum of
dust
27 K, and varying between T ≈28–30 K in the SMC IRAS00430−7326is too sharpto be the peak ofa single
dust
10 van Loon et al.
black-body,buttoobroad(anddisplaced)tobethe[Oi]
line. We thus suggestthis is a strong emissionfeature of
crystallinewaterice. Thiswouldnotbesurprisingasthis
isalsothe objectwiththestrongestwatericeabsorption
at λ∼3 µm in the Spitzer-based SMC sample observed
in that way (J.M.Oliveira et al., in prep.). What is sur-
prisingisthatthe dustisnotverycold,butperhapsthis
isthereasonfortheicetohavecrystallizedtothedegree
we witness.
S3MC00464−7322istheYSOwiththecoldestdustin
our MIPS-SED sample. The bump around λ ≈ 64 µm
is slightly displaced to longer wavelengths than the [Oi]
line seen in other objects, and it is thus possible that in
thiscaseitisduetocrystallinewatericeandnotoxygen.
A third YSO, S3MC00540−7321exhibits an emission
featurecloserto60µm,whichmightalsobeduetocrys-
talline water ice.
5.4.2. Molecules
GM103 displays the same sharp peak around 79 µm
that was also noticed in the MIPS-SED spectra of the
SNR N49 and PN or high-mass star IRAS05047−6644
in the LMC (Paper I). Then, it was suggested that it
Fig.6.— Averaged MIPS-SED spectra of 11 objects associated
may be due to a blend of the cluster of relatively strong
withstarformationintheSMCand22comparableobjectsinthe
transitionsofwatervapourandOHemissionlinesaround LMC. Clearly, the SMC objects lack dust and oxygen, and their
that wavelength (cf. Lerate et al. 2006). The detection dustiswarmer,comparedtothemoremetal-richLMCobjects.
in the spectrum of GM103 is of rather low significance.
The lower dust and oxygen content is not surprising
However, if the swept-up dust scenario is true (§5.3.2)
giventhatthemetallicityoftheISMintheSMCislower
then a slow (C-type) shock could be held responsible for
thanthatofthe LMC.Toquantify the differenceindust
the prolific formation of water vapour (Bergin, Neufeld
content, we compare the average far-IR luminosities be-
& Melnick 1998).
tween the 11 SMC and 22 LMC star-formation objects,
where we remind the reader that we computed the far-
5.4.3. Minerals
IR luminosity due to the modified black-body emission
Broademissionfeaturesare sometimesseenaround75 fromdustatatemperatureT (seePaperIfordetails).
dust
µm, most clearly in IRAS01042−7215 and possibly in This luminosity is a function ofthe dustmass, M , as
dust
S3MC01051−7159. These are both YSOs, just like the well as the dust temperature:
YSO N159S in the LMC in which a similar bump was
seen (Paper I). L(FIR)∝Mdust(Tdust)4+β. (3)
Although the carrier of this feature is not yet known,
Assuming, as before, a value for β ≡1.5, we thus obtain
itisreassuringthatasimilarfeatureappearsintheSMC
hM i /hM i ≈3.8.
andLMCspectra. InPaperIwesuggestedthathydrous dust LMC dust SMC
Thedifferenceindustcontentisverysimilartothefac-
silicates provide a promising identification.
tor 2–4 difference in metallicity between the SMC and
LMC, and suggests that the total dust mass of a star-
5.5. Comparison of the SMC with the LMC
forming molecular cloud core scales in proportion to its
5.5.1. Dust and oxygen content metallicity. Whilst unsurprising at first, this seems at
Thereare11objectsinourSMCsamplethatareasso- odds with the even lower values for the dust-to-gas ra-
ciated with star-formation (YSOs and compact Hii re- tio that have been measured in the diffuse ISM of the
gions). There are 22 similar objects in the LMC sam- SMC (∼ 1/30 that in the Galaxy, Stanimirovi´c et al.
ple (Paper I). It thus becomes meaningful to compare 2000), and with the smaller sizes of CO clouds in the
these samples to identify fundamental differences be- SMC (Lequeux et al. 1994). Graingrowthwithin clouds
tween these two galaxies. If differences are found, an might enhance the dust-to-gas ratio over that typically
obviousreasonforthesemightbethedifferenceinmetal- encountered in the diffuse ISM; this was recently sug-
licity between these galaxies. gested to happen in molecular cloud cores within the
A comparison of the average MIPS-SED spectra be- metal-poor tail of the SMC (Gordon et al. 2009). To
tweentheSMCandLMCstar-formationobjectsimmedi- explain more efficient growth in the SMC would require
atelyrevealsthreeconspicuousdifferences(Fig.6). First, thedensitieswithinmolecularcloudcoresintheSMCto
thecontinuuminthe SMCisfainter. Thiscouldimply a be larger than within those in the LMC, as the growth
lower dust content in the SMC. Second, the continuum timescale τ ∝N−1 (Zhukovska & Gail 2009).
H
in the SMC is flat, as opposed to the rising continuum For the oxygen lines we obtain a luminosity ratio of
in the LMC. This could imply that the dust is warmer hL(O)i /hL(O)i ≈ 1.7, i.e. very similar to the
LMC SMC
in the SMC, as we quantified earlier. Third, the oxygen ratio of far-IR luminosities (which is ≈ 1.6) but less
lines are weaker in the SMC. This could reflect a lower than expected from a simple scaling with metallicity.
oxygen abundance in the SMC. Higher densities in cloud cores in the SMC would ex-