Nature of W51e2: Massive Cores at Different Phases of Star Formation

0.48 MB·English

by  Hui Shi

DRAFTVERSIONJANUARY17,2010
PreprinttypesetusingLATEXstyleemulateapjv.11/12/01
NATUREOFW51E2:MASSIVECORESATDIFFERENTPHASESOFSTARFORMATION
HUISHI1,JUN-HUIZHAO2,J.L.HAN1,
DraftversionJanuary17,2010
ABSTRACT
Wepresenthigh-resolutioncontinuumimagesoftheW51e2complexprocessedfromarchivaldataoftheSub-
0
1 millimeterArray(SMA)at0.85and1.3mmandtheVeryLargeArrayat7and13mm.Wealsomadelineimages
0 and profiles of W51e2 for three hydrogenradio recombinationlines (RRLs; H26α, H53α, and H66α) and ab-
2 sorption of two molecular lines of HCN(4-3) and CO(2-1). At least four distinct continuum componentshave
beendetectedinthe3′′regionofW51e2fromtheSMAcontinuumimagesat0.85and1.3mmwithresolutionsof
n
0.3′′×0.2′′and1.4′′×0.7′′,respectively.Thewestcomponent,W51e2-W,coincideswiththeultracompactHIIre-
a
J gionreportedfrompreviousradioobservations.TheH26αlineobservationrevealsanunresolvedhyper-compact
ionized core (<0.06′′ or <310AU) with a high electron temperature of 1.2×104K, with the corresponding
7
1 emissionmeasureEM>7×1010pccm- 6 andtheelectrondensityNe >7×106 cm- 3. TheinferredLymancon-
tinuumfluximpliesthattheHIIregionW51e2-Wrequiresanewlyformedmassivestar,anO8staroraclusterof
] B-typestars,tomaintaintheionization. W51e2-E,thebrightestcomponentat0.85mm,islocated0.9′′eastfrom
R
the hyper-compactionized core. It has a total mass of ∼140M according to our spectral energy distribution
⊙
S analysisanda largeinfallrate of>1.3×10- 3M⊙yr- 1 inferredfromtheabsorptionofHCN. W51e2-Eappears
. tobetheaccretioncenterinW51e2. Giventhefactthatnofree- freeemissionandnoRRLshavebeendetected,
h
p the massive core of W51e2-E appears to host one or more growing massive proto-stars. Located 2′′ northwest
- fromW51e2-E,W51e2-NWisdetectedinthecontinuumemissionat0.85and1.3mm. Nocontinuumemission
o hasbeendetectedatλ≥7mm. Alongwith themaseractivitiespreviouslyobserved,ouranalysissuggeststhat
tr W51e2-NWisatanearlierphaseofstarformation. W51e2-Nislocated2′′northofW51e2-Eandhasonlybeen
s detectedat1.3mmwithalowerangularresolution(∼1′′),suggestingthatitisaprimordial,massivegasclump
a
intheW51e2complex.
[

Subjectheadings:HIIregions–ISM:individualobjects(W51e2)–stars: formation
2
v
1 1. INTRODUCTION 2004;Ketoetal.2008).
0 Observationsofhydrogenradiorecombinationlines(RRLs)
Massivestarsareformedindense,massivemolecularcores.
1 were carried out toward W51e2 (Mehringer 1994; Ketoetal.
Detailed physical processes in star-forming regions have not
1 2008;Keto&Klaassen 2008). Mehringer(1994) failedto de-
beenwellstudieduntilrecenthigh-resolutionobservationswere
. tect the H92α line in this region, which might be attributed
1 available at submillimeter wavelengths. High angular resolu-
0 tionobservationsarenecessarytounveilthephysicalenvirons to the large optical depth and/or the large pressure broaden-
0 ingoftheUCHIIregionat3.6cm. Basedontheobservations
andactivitiesoftheindividualsub-coresaswellastheirimpact
1 of H66α, H53α, and H30α, Ketoetal. (2008) arguedthat the
ontheoverallprocessofmassivestarformation.
: pressurebroadeningisresponsibleforthebroad-linewidthsob-
v W51 is a well-known complex of HII regions with about
i 1◦ area in the Galactic plane (Westerhout 1958). W51A, also servedintheRRLs.
X Observationsofmolecularlines(e.g.Ho&Young1996;Zhang&Ho
calledG49.5-0.4,isthemostluminousregioninW51(Kundu&Velusamy
r 1967).SeveraldiscretecomponentshavebeenfoundfromW51A 1997;Zhangetal.1998;Youngetal.1998;Sollinsetal.2004)
a andreferredalphabeticallyasa- iintherightascension(R.A.) showedevidenceforinfall(oraccretion)gaswithin5′′(<0.2pc)
aroundW51e2.ApossiblerotationwassuggestedbyZhang&Ho
order (Martin 1972; Mehringer 1994). W51e is one of the
(1997) on the basis of fitting the position- velocity (PV) dia-
brightest regions in W51A. High-resolution centimeter obser-
gramoftheNH (3,3)absorptionlineat13mm. Aspin-upro-
vations have revealedthat W51e consists of severalultracom- 3
tation with an axis of position angle (P.A.) ≈20◦ was further
pact(UC)HIIregions(Scott1978;Gaumeetal.1993),among
suggestedbased on the velocitygradientof CH CN(8 - 7 ) at
which the UC HII region of W51e2 appears to be the bright- 3 3 3
2mm(Zhangetal.1998). However,basedonthevelocitygra-
est. This region is severely obscured by the dust, and no IR
dientobservedfromtheH53αline,Keto&Klaassen(2008)in-
detection has been made at wavelengths shorter than 20µm
terpretedtheir resultsasevidenceforrotationalionizedaccre-
(Genzeletal.1982).TheinferreddistancetoW51e2is∼5.1kpc
(Xuetal.2009);thus,1′′correspondstoalinearscaleofabout tionflowaroundtheUCHIIregionandderivedarotationalaxis
of P.A.≈- 30◦. In addition, a possible bipolar outflow along
5100AU.
thenorthwestandsoutheast(NW- SE)directionswassuggested
Based onthe radiocontinuumobservations,spectralenergy
basedontheobservationsoftheCO(2-1)lineemission(Keto&Klaassen
distribution(SED)analysesofW51e2suggestthattheemission
2008),whichappearstobeperpendiculartotheaxisofthero-
sourcefromthisregionconsistsoftwocomponents,namely,(1)
thermalemissionfromacolddustcomponentand(2)free- free tationalionizeddisk. However,thesouthwest(SW)elongation
oftheUCHIIregioninGaumeetal.’s(1993)observationswas
emissionfromahotHIIregion(Rudolphetal.1990;Sollinsetal.
explainedas a collimatedionizedoutflow. On the otherhand,
1 NationalAstronomicalObservatories, ChineseAcademyofSciences, Jia20 proper-motion measurements of H2O/OH masers appeared to
DatunRoad,ChaoyangDistrict,Beijing100012,China. Email: shihui,hjl@ favor a hypothetical model with an outflow or expanding gas
nao.cas.cn
bubbleinW51e2(Imaietal.2002;Fish&Reid2007).
2Harvard-SmithsonianCenterforAstrophysics,60GardenStreet,Cambridge,
Thus,improvedimagesandcomprehensiveanalysisofhigh-
MA02138,USA.Email:[email protected]
1
2 SHI,ZHAO,&HAN
TABLE1
OBSERVATIONPARAMETERSFORW51E2
Parameters SMA:0.85mm SMA:1.3mm VLA:7mm VLA:13mm
Observationdate 2007Jun.18 2005Sep.1 2004Feb.14 2003Sep.29
Arrayconfiguration Veryextended(8ants) Extended(6ants) AB BC
PointingcenterR.A.(J2000) 19:23:43.888 19:23:43.895 19:23:43.918 19:23:43.918
Pointingcenterdecl.(J2000) +14:30:34.798 +14:30:34.798 +14:30:28.164 +14:30:28.164
Frequency(GHz) 343,353 221,231 43 22
Bandwidth 2.0+2.0(GHz) 2.0+2.0(GHz) 12.5(MHz) 12.5(MHz)
On-sourcetime(hr) 2.10 4.75 5.90 3.93
Systemtemperature(K) 150∼500 100∼250
Bandpasscalibrators J1229+020,J1751+096 3C454.4 3C84 3C84
Phasecalibrators J1733- 130,J1743- 038 J1751+096,J2025+337 J1923+210 J1923+210
J2015+371
Fluxcalibrators Callisto Ceres 3C286 3C286
resolution observations at the wavelengths from radio to sub-
millimeter are needed to reconcile the differences in the in-
terpretation of the results from the previous observations of
W51e2. In this paper, we show the high-resolution contin-
e2-NW
uum imageswith the SubmillimeterArray(SMA) at0.85and
1.3mmandwiththeVeryLargeArray(VLA)at7and13mm,
aswellaslineimagesandprofilesofthehydrogenrecombina-
tionlines(H26α,H53α,andH66α)andthemolecularlinesof
HCN(4-3) and CO(2-1). In Section 2, we describe the details e2-W
e2-E
ofthedataprocessingandresults. InSection3, wemodelthe
individualcomponentson the basis of the high-resolutionob-
servationsofW51e2with theSMA andVLA.The natureand
astrophysical processes of star formation cores in W51e2 are
discussed. AsummaryandconclusionsaregiveninSection4.
0.85 mm
2. DATA REDUCTION AND RESULTS
WeprocessedthearchivaldatafromtheSMAandVLAob-
servations of W51e2 and constructed both the continuum and
spectrallineimages.Detailsincalibrationandimagingaredis- FIG. 1.— High-resolution 0.85 mm continuum image of W51e2 ob-
served with the SMA. Contours are ±5σ×2n (n= 0, 1, 2, 3, ... and
cussedbelowalongwiththerelevantparametersthataresum- σ=7.11 mJybeam- 1). TheFWHMbeam (0.31′′×0.22′′, P.A.=60.6◦)is
marizedinTable1. shownatbottomright. Threecontinuumemissioncomponentse2-NW,e2-E,
ande2-Waremarkedwith“+”.Theλ3.6cmpositionoftheUCHIIregionin
2.1. TheSMAcontinuumimagesat0.85and1.3mm W51e2Gaumeetal.(1993)ismarkedwith“×”.
TheobservationsofW51e2at0.85mmwerecarriedoutwith
the SMA in the very extended array (eight antennas) and at
Figure 1, the symbol “+” marks the positions of the compo-
1.3mmintheextendedarray(sixantennas).Thedatawerecal-
nents). Thebrightestsource,W51e2-E,islocated∼0.9′′ east
ibratedinMiriad(Saultetal.1995)followingthereductionin-
oftheUCHIIregionwhichismarkedwith“×”forthepeakpo-
structionsforSMAdata3. Thesystemtemperaturecorrections
sitionat3.6cmfromGaumeetal.(1993).Thesecondarybright
were applied to the visibility data. The bandpass calibration
source, W51e2-W, coincides with the 3.6cm peak of the UC
wasmadewithstrongcalibratorsforeachofthetwodatasets
HIIregion.Thethirdcomponent,W51e2-NW,islocatedabout
(Table1). ThefluxdensityscalewasdeterminedusingCallisto
2′′northwestofW51e2-E.Inaddition,thesecompactemission
at0.85mmandCeresat1.3mm.
componentsappeartobesurroundedbyanamorphoushaloin
Corrections to the complex gains at 1.3mm were made by
the continuum emission at 0.85mm. The three bright contin-
applyingthesolutionsinterpolatedfromthetwonearbyQSOs
uum sources were also detected in the low-resolution contin-
J1751+096 and J2025+337. For the high-resolution data at
uumimageat1.3mm(Figure2(a)). Moreover,aweakcontin-
0.85mm,thevisibilitydataareinitiallycalibratedwithamodel
uumcomponent,W51e2-N,ispresent∼2′′northofW51e2-E.
ofW51e2derivedfromobservationsinalowangularresolution
at 0.87mm. The residual phase errors were further corrected
2.2. TheVLAcontinuumimagesat7and13mm
using the self-calibration technique. The final continuum im-
ages wereconstructedbycombiningall the line-freechannels The observations of W51e2 at 7 and 13mm were carried
inbothuppersidebandandlowersidebanddata. out with the VLA (see the summarized observing parameters
Figure1showsthehigh-resolution(0.3′′×0.2′′)continuum inTable1).Wemadethecalibrationsbyfollowingthestandard
imageofW51e2at0.85mm. ThecomplexofW51e2hasbeen data-reductionprocedureforVLA datawithAIPS4. Then,the
resolved into at least three bright, compact components (see calibratedvisibilitieswereloadedintotheMiriadenvironment.
3http://www.cfa.harvard.edu/sma/miriad 4http://www.aips.nrao.edu
NATUREOFW51e2:MASSIVECORESATDIFFERENTPHASESOFSTARFORMATION 3
(a) (b) (c)
e2-W e2-W e2-W
H26a H53a H66a
(d) (e) (f)
FIG. 2.— (a): 1.3mmcontinuumimageofW51e2observedwiththeSMA.Contoursare±5σ×2n(n=0,1,2,3,... andσ=14.0mJybeam- 1). TheFWHM
beam(1.35′′×0.69′′,P.A.=- 87.4◦)isshownatbottomright.Thethreecomponentsdetectedat0.85mmtogetherwiththeadditionalcomponente2-Ndetectedat
1.3mmaremarkedwith“+”. Theposition“+”ofthesubmillimetercomponentsandthepeaksofthe3.6cmcontinuum(“×”)andH26α(“⋆”)arealsomarked
intherestoftheimages. (b): 7mmcontinuumimageofW51e2observedwiththeVLA.Contoursare±5σ×2n/2(n=0,1,2,3,... andσ=1.12mJybeam- 1).
TheFWHMbeam(0.40′′×0.15′′,P.A.=80.3◦)isshownatbottom-right. (c): 13mmcontinuumimageofW51e2observedwithVLA.Contoursare±5σ×2n
(n=0,1,2,3,...,andσ=0.50mJybeam- 1). TheFWHMbeam(0.23′′×0.10′′,P.A.=82.6◦)isshownatbottomright. (d): ImageoftheH26αlineemission
integratedfrom25kms- 1to84kms- 1fromW51e2. Contoursare±5σ×2n(n=0,1,2,3,... andσ=0.146Jybeam- 1 kms- 1). TheFWHMbeam(0.25′′)is
shownatbottomright. AlltheRRLimagesareconvolvedtoacommoncircularbeam(0.25′′).ThestardenotesthepeakpositionoftheH26αlineandismarked
intherestoftheRRLimages. (e): ImageoftheH53αlineemissionintegratedfrom22kms- 1 to101kms- 1. Contoursare±5σ×2n (n=0,1,2,3,... and
σ=9.68×10- 3Jybeam- 1kms- 1).(f):ImageoftheH66αlineemissionintegratedfrom24kms- 1to120kms- 1.Contoursare±5σ×2n/2(n=0,1,2,3,...and
σ=6.14×10- 3Jybeam- 1kms- 1).
TheimagingandfurtheranalysiswerecarriedoutwithMiriad. with a 3σ cutoff in the velocity range from 25 to 84km s- 1.
ThedirtyimagesweremadeusingINVERTwithrobustweight- The H26α line emission from the UC HII region(W51e2-W)
ing(robust=0)andwerecleanedwiththehybriddeconvolution appearstobeverycompactandhasnotbeenresolvedwiththe
algorithm.Figures2(b)and2(c)showthecontinuumimagesat beam of 0.25′′, suggesting thatthe intrinsic source size of the
7 and 13mm, respectively. Strong continuumemission is de- hyper-compactHII core is <0.06′′. The peak position of the
tectedfromW51e2-Watboth7and13mmingoodagreement integrated H26α line emission (stars in Figure 2) appears to
withpreviousobservationsatcentimeters(Gaumeetal.1993). haveasignificantoffsetof∼0.15′′fromthecontinuumpeakat
In addition,a weakbutsignificant(>10σ)emissionfeatureat 13mm(alsosee theGaume’spositionin Figure2(d)). Unlike
W51e2-Ehasbeendetectedatonly7mm. AsshowninFigure the other RRLs and continuum emission at centimeter wave-
2(b), the weak 7mm continuum source at W51e2-E has been lengths, the line emission at the frequency of the H30α line
clearlyseparatedfromtheUCHIIregion,W51e2-W. (theimageisnotshowninthispaper)showsanextendeddistri-
butioncoveringtheentire3′′ region,suggestingthattheH30α
2.3. Hydrogenrecombinationlines line might have been severely contaminated by one or more
molecularlinesinW51e2(e.g.,C H CNwitharestfrequency
For the line cubes, the continuum emission was subtracted 3 7
of231.9009GHz).
using the linear interpolationfrom the line-free channelswith
ThelineofH66αisbroadandweak,whichisbarelycovered
UVLIN.TheH26αline(353.623GHz)wasincludedintheline
by the VLA correlator band with which we had difficulty de-
cubes made from the SMA data at 0.85mm. We also made a
terminingtheline-freechannels. Instead,thecontinuumlevels
lineimagecubetocovertheH30αline(231.901GHz)fromthe
weredeterminedbyaveragingallchannelsexcludingafewrel-
SMAdataat1.3mm. BoththeH26αandH30αlinedatawere
resampled to 1km s- 1. The rmsnoises of 49mJy beam- 1 and ativelystronglineemissionchannelsandweresubtractedfrom
81mJy beam- 1 in each of the channelimages are inferredfor the visibility data. The uncertaintyis <5% in the continuum
levels due to the line contamination. The image cubes of the
theH26αandH30αlineimagecubes,respectively.Figure2(d)
H53αandH66αlineswerecleanedandconvolvedtoacircular
showstheintegratedlineintensityimageoftheH26αlinemade
4 SHI,ZHAO,&HAN
corecorrespondstothecoldgasinfrontofthedustcoremov-
ing toward it, i.e., the infall. The CO(2-1) spectrum shows a
broaderredshiftedfeatureinabsorptionthanthatofHCN.Due
tothepoorangularresolutionintheCO(2-1)observations,the
observedbroadprofileinabsorptionis causednotonlybythe
infallbutalsobytheoutflow. Inparticular,thehighredshifted
absorptionarisesverylikelyfromtheoutflowgas. Incompari-
sonwiththatofW51e2-E,thespectrumtowardW51e2-Wonly
showsaweakspectralfeature(Figure4(d))whichcanbechar-
acterized as an inverse P Cygni profile, suggesting that only
littlemoleculargas(ifpresent)fallsontotheUCHIIregion.
Toshowthespatialdistributionoftheabsorptiongas,wecar-
riedoutamomentanalysis. Thecalculationofthezerothmo-
mentcorrespondsto the integrated intensity over the velocity.
Inordertoavoidcancellationbetweenemissionandabsorption
FIG. 3.— Profiles of hydrogen recombination lines, H26α, H53α, and spectralfeaturesinthesamebeamarea,weseparatetheabsorp-
H66α, integrated from W51e2-W in an area of 0.5′′×0.5′′ centered at the tion and emission in the moment calculation. We found that
H26αlinepeak. Allthelineimageshavebeenconvolvedwiththesamecir- both the blueshifted and redshifted emission of the HCN(4-3)
cular beam of0.25′′. Thespectra aremultiplied with theratios ofthe rest line(theimagesarenotshowninthispaper)isextendedfrom
fvreerqtiuceanlcdyasohfetdhelinHe2d6eαnoltiensethtoetshyestRemRaLticrevstelforceiqtyueonfc5ie3s.9(±νH12.61αk/mνRsR- L1)w. hTichhe northwest to southeast across W51e2-E, which is, in general,
wedeterminedfromthepreviousmeasurementsofsevendifferenthotmolec- consistent with the molecular outflow direction as suggested
ularlines(seeSection2.4). basedontheCO(2-1)lineobservation(Keto&Klaassen2008).
TheHCN(4-3)absorption,integratedfrom44to62kms- 1with
a5σcutoffineachchannel(Figure4(a)),isfoundtobemainly
beamof0.25′′,givingthermsnoisesof2.4and1.5mJybeam- 1
concentratedonW51e2-E.Fromthelower-resolutionobserva-
per channel, respectively. The integrated intensities of H53α
tion, thedistributionofthe CO(2-1)absorptionline integrated
(witha3σcutoffineachchannel)andH66α(witha2σcutoffin from43 to 71km s- 1 (a 5σ cutoffin each channel, see Figure
eachchannel)areshowninFigures2(e)and2(f),respectively.
4(b)), also shows the absorptionpeaks at W51e2-Einstead of
BothH53α,andH66αlinesaredetectedfromonlyW51e2-W,
W51e2-W. We also note that the CO(2-1) absorption shows a
theUCHIIregion. FortheH26α,H53α,andH66αlinesfrom
relatively extendedfeature north of W51e2-E, further indicat-
W51e2-W,weintegratedthelineemissionregionwithasizeof
ingthattheabsorptionofCO(2-1)mightindeedbesignificantly
0.5′′aroundtheH26αpeakposition.TheprofilesoftheH26α,
contaminatedbytheoutflowgas.
H53α and H66α lines are multiplied by the frequency ratios
Alongwiththeabsorptionspectra,thedistributionoftheab-
of ν /ν , ν /ν and ν /ν , respectively,
H26α H26α H26α H53α H26α H66α sorbinggasinW51e2evidentlydemonstratesthatthe submil-
asshowninFigure3. Themeasurementsofthepeakintensity
limeter(dust)coreW51e2-E,insteadofW51e2-W(theUCHII
(I ), radialvelocity(V ), andFWHM linewidth(∆V)for
peak LSR region),isthecenterofaccretionforthemajorityofthehigh-
theH26α,H53α,andH66αlinesaregiveninTable3.
densitygasfromthismolecularcore.
2.4. AbsorptionofHCN(4-3)andCO(2-1)
2.5. SpectralEnergyDistributions
The systematic velocity of W51e2 varies when it is mea-
In lack of high angular resolution observations at millime-
sured using different hot molecular lines (Sollinsetal. 2004;
ter/submillimeter wavelengths, the flux densities from the en-
Remijanetal. 2004; Zhangetal. 1998; Rudolphetal. 1990).
tireregionofW51e2wereconsideredinthepreviousanalyses
Inordertominimizethepossibleeffectsofthedifferentdistri-
oftheSED(Rudolphetal.1990;Sollinsetal.2004;Ketoetal.
butionswithdifferentmolecularlines,weaveragedthesemea-
2008).Withthehigh-resolutionobservationsatthewavelengths
surementsofmolecularlinesH13CO+,SO ,SiO,CH CN,and
2 3 fromradiotosubmillimeter,wearenowabletospatiallysepa-
HCO+ and obtained a mean value of 53.9±1.1km s- 1. We
rateindividualemissioncomponentsinourSEDanalysis.
adoptthis value as the systematic velocityfor W51e2-Ehere-
Addingthe data at3.6cm (Gaumeetal. 1993) and at2 and
after.
6cm (Rudolphetal. 1990) along with the continuum data at
The lines of HCN(4-3) at ν = 354.505GHz and CO(2-1)
0 thefourwavelengths(13,7,1.3,and0.8mm)discussedinthis
at ν = 230.538GHz were covered in the SMA observations
0 paper,we havesevenmeasurementsin fluxdensitiescovering
at 0.85mm and 1.3mm, respectively. The dirty images were
nearly 2 order of magnitude in frequency. For 13, 7, and 0.8
made with natural weighting and deconvolved with the Hog-
mm data, the flux density measurementswere made by fitting
bomClarkSteerhybridalgorithm. Thefinalrmsnoisesofσ=
a Gaussian to the individual emission components in the im-
65 and 94mJy beam- 1 per channel (1km s- 1 in the channel ageswiththesamesizeoftheconvolvedFWHMbeam(0.4′′).
width)wereachievedforHCN(4-3)andCO(2-1),respectively.
For the 1.3mm data, we determined the flux density by fit-
Figure4showsthespectraofHCN(4-3)towardW51e2-Eand
tingmultipleGaussiancomponentstotheemissionfeaturesin
W51e2-WandthespectrumofCO(2-1)nearthepeakposition a relativelylower resolution(1.1′′), which results in relatively
of W51e2-E.A significantabsorptionof the HCN(4-3)line is
largeuncertainties. FortheemissioncoresW51e2-W,W51e2-
detectedagainstthesubmillimetercore,W51e2-E,andmostof
E,W51e2-NW,andW51e2-N,thefluxdensitiesdeterminedare
which is redshifted with respect to the systematic velocity of
givenin Table 2. The relevantcontinuumspectra for the four
53.9±1.1kms- 1asindicatedbytheverticaldashedlineinFig-
emissioncomponentsareshowninFigure5.
ure4(c). NotethattheredshiftedabsorptionofHCNobserved
withthehighangularresolutionagainstthecompactcontinuum
NATUREOFW51e2:MASSIVECORESATDIFFERENTPHASESOFSTARFORMATION 5
White Contours: CO(2-1)
Contours: HCN(4-3) absorption Black Contours: HCN(4-3)
Color: 0.85 mm Color: 0.85 mm
(a) (b)
(c) (d) (e)
FIG. 4.— (a): Image of the HCN(4-3) absorption line integrated from 44 to 62 km s- 1. Dashed contours are ±5σ×2n (n=0, 1, 2, 3, ... and
σ=0.113Jybeam- 1 kms- 1), overlaid onthegray-scaled continuum imageat0.85mm. TheFWHMbeam0.33′′×0.24′′ (sup=0)isshownatbottomright.
(b):ImageoftheCO(2-1)absorptionlineintegratedfrom43to71kms- 1.Whitedashedcontoursare±5σ×2n(n=0,1,2,3,...andσ=0.771Jybeam- 1kms- 1),
overlaidonthegray-scaledcontinuumimageat0.85mm. TheFWHMbeam1.35′′×0.73′′(sup=0)isshownatbottomright. (candd): Thespectralprofilesof
theHCN(4-3)absorptionlinetowardW51e2-EandW51e2-W,respectively. Theverticaldashedlinedenotesthesystematicvelocityof53.9±1.1kms- 1.(e):The
spectralprofileoftheCO(2-1)absorptionlinetowardW51e2-E,withaverticaldashedlineforthesystematicvelocity.
3. NATUREOF THEW51E2COMPLEX with the radio continuum images observed at 1.3cm (Figure
Four distinct componentshave been identified in the 3′′ re- 2(c) and Gaumeetal. 1993), 3.6cm (Gaumeetal. 1993), and
7mm(Figure2(b)),suggestingthattheNE- SWextensionob-
gionofW51e2fromthehigh-resolutionimagesatwavelengths
served in the H53α line corresponds to the expansion of ion-
fromcentimetertosubmillimeter(Figure1and2). Thenature
izedgas(oroutflow). TheH66αlineemissionalsohasasim-
ofthesecomponentsisdiscussedasfollows.
ilarelongation(Figure2(f)),withpeakintensitynearGaume’s
3.1. W51e2-W:UCHIIregion position in the NE and an extension structure in the SW. We
also found a significant velocity gradient of the H53α line in
W51e2-Wistheonlycoredetectedatcentimeterwavelengths,
theW51e2-Wregion,inagreementwithwhatwasobservedby
anditscontinuumemissionisdominatedbythefree- freeemis-
Keto&Klaassen(2008), redshiftedintheSWandblueshifted
sion from the ionized core and traces the thermal HII region
in the NE. Thebroadwingsofthe recombinationlines shown
around the central star. Figure 2(b)- (f) show the UC HII re-
in Figure 3 (also in Table 3) correspondingto extended, opti-
gion,W51e2-W,includinganunresolvedcore(apossibleion-
callythinemissionindicatethatitishardtoexplainthemasan
ized disk) and a northeastand southwest (NE- SW) extension
ionizeddisk. ThevelocitygradientobservedintheH53αline
as a possible outflow. The NE and SW extensions are best to
is more likely produced by an ionized outflow rather than an
beseenat7and13mminFigure2(b)and(c),whichisingood
ionizedaccretiondisk.
agreementwiththepreviousobservationsat3.6and1.3cmby
We fitted the SED of W51e2-W with a free- free emission
Gaumeetal.(1993).Gaumeetal.(1993)foundthespectralin-
model
dicesαofabout2(opticallythick)and0.4fortheNEandSW
extensionsofW51e2-W,respectively,andsuggestedthatthere Sν =ΩsBν(Te)(1- e- τc) (Jy), (1)
isaone-side(SW)collimatedionizedoutflowfromthecore. whereSν isthefluxdensityatfrequencyν,Ωsisthesolidangle
TheH26αlineisanexcellenttracerforhyper-compaction- ofthesource,Bν(Te)isthePlanckfunction,andTe istheelec-
izedcores(starinFigure2(d)),providingagooddiagnosisfor trontemperature. Thecontinuumopticaldepthisexpressedby
the presence of an ionized disk. On the basis of the SMA (Mezger&Henderson1967)as
observations with an angular resolution of 0.25′′, the hyper- T - 1.35 ν - 2.1 EM
compactionizedcore whichpeakedat 59.1kms- 1 in W51e2- τ =0.0824 e α(ν,T).(2)
c K GHz pccm- 6 e
Whasnotbeenresolved,givingalimitontheintrinsicsizeof (cid:18) (cid:19) (cid:16) (cid:17) (cid:18) (cid:19)
θ <0.06′′ and linear size <310AU from a Gaussian fitting. Theemissionmeasure(EM)canbewrittenas
s
TheelongationshownintheH53αimage(Figure2(e))agrees EM=N2Lf , (3)
e V
6 SHI,ZHAO,&HAN
TABLE2
COREFLUXESOFW51E2ATCENTIMETERTOSUBMILLIMETERBANDS
parameters W51e2-W W51e2-E W51e2-NW W51e2-N ReferencesandBeamsizeΘ
R.A.(J2000) 19:23:43.90 19:23:43.96 19:23:43.87 19:23:43.97 Thispaper
Decl.(J2000) +14:30:34.62 +14:30:34.56 +14:30:35.97 +14:30:36.57 Thispaper
Deconvolvedsize(′′)∗ 0.45 0.71 0.38 1.24 Thispaper
S (Jy) 0.35±0.07 3.30±0.20 0.81±0.34 Thispaper:Θ=0.4′′
0.85mm
S (Jy) 2.15±0.12 0.62±0.12 0.73±0.08 Thispaper:Θ=1.1′′
1.3mm
S (Jy) 0.62±0.09 0.04±0.01 <0.0052 <0.0052 Thispaper:Θ=0.4′′
7mm
S (Jy) 0.43±0.03 <0.004 <0.004 <0.004 Thispaper:Θ=0.4′′;Gaumeetal.(1993):upperlimit
1.3cm
S (Jy) 0.21±0.05 Rudolphetal.(1990):Θ=0.4′′
2cm
S (Jy) 0.067±0.004 <0.001 <0.001 <0.001 Gaumeetal.(1993):Θ=0.21′′;Gaumeetal.(1993):upperlimit
3.6cm
S (Jy) 0.025±0.003 – – – Rudolphetal.(1990):Θ=0.6′′
6cm
∗Thesizesofe2-W,e2-E,ande2-NWarederivedfromthe0.85mmimage,ande2-Nisfromthe1.3mmimage.
where N is the mean electron density, L is the path length, AssumingN(He)/N(H)=0.096(Mehringer1994)andα(ν,T)=
e e
and f is the volume filling factor. The dimensionless factor 1,wehaveT∗=12,000±2,000Kwhichappearstobeconsid-
V e
α(ν,T)istheorderofunity(Mezger&Henderson1967). For erablyhigherthanthemeanelectrontemperatureoftheoverall
e
ahomogeneousionizedcore,themodelcanbeexpressedwith HIIregion. TheEMofthehyper-compactHIIcorecanbeas-
two freeparameters,namely,T andEM,forthefirstapproxi- sessedas
e
mationdescribedabove. T 3/2 θ - 2 λ S
ThesolidcurveinFigure5(a)showsthebestfittothedata, EM=7.1pccm- 6 e s L ×
suggestinga turnoverfrequencyofν0≈27GHz. Thederived (cid:18)K(cid:19) (cid:18)arcsec(cid:19) (cid:18)mm(cid:19)(cid:18)Jy(cid:19)
mean values for the physical parameters for the overall HII ∆V
region, Te=4900±320K and EM=(1.3±0.2)×109pc cm- 6, are kms- 1 , (7)
similar tothepreviousresultinZhangetal.(1998). Thus, the (cid:18) (cid:19)
mean electron density of N ∼3×105cm- 3 is inferred on the whereTe is the electrontemperature,θs is the intrinsicsize of
assumptionofLf ∼0.5′′×5e.1kpc. the HII region, SL is the peak line intensity in Jy, ∆V is the
V FWHM line width in km s- 1, and λ is the observing wave-
Both the Lyman continuumflux(N ) and the excitationpa-
L lengthinmillimeters. BasedonthemeasurementsoftheH26α
rameter(U)canbederived(Rubin1968;Panagia1973;Matsakisetal.
linefromthehyper-compactionizedcoretogetherwiththeas-
1976)asfollows:
sumption of T ≈T∗, EM >7×1010 pc cm- 6 is found. The
e e
S D 2 ν 0.1 corresponding lower limit of the volume electron density is
NL &7.5×1046s- 1 Jyν kpc GHz × Ne >7×106cm- 3 assuming LfV <0.06′′×5.1 kpc. The in-
Te - 0.45(cid:18), (cid:19)(cid:18) (cid:19) (cid:16) (cid:17) (4) fgeersrtedthhaitghtheeleHct2r6oαntleimnepearraistuesrefarnodmhaighhoetl,ecvterroyncdoemnspitayctsurge--
104K gion(<0.06′′) whichisprobablycloseto thecentralionizing
(cid:18) (cid:19)
1 4 star(s). With the nature of optically thin and sensitive to the
U = 3.155×10- 15pccm- 2 NL 3 Te 15, (5) high-density ionized gas, the H26α line is an excellent tracer
s- 1 104K of the hot, high-density ionized region surroundingthe ioniz-
(cid:18) (cid:19) (cid:18) (cid:19)
whereDisthedistancetothesource.Weassumedα(ν,T)∼1. ing source. In comparison with the mean electron tempera-
Basedonourmodelfitting,thecontinuumemissionfromee2-W ture(Te=4900K)inferredfromthefree- freeemissioninalarge
becomesopticallythin ata frequency>80GHz (τ <0.1),and area(0.5′′),thehightemperatureofTe=12,000Kderivedfrom
we obtained N =3.0×1048 s- 1 and U=37.6pc cm- 2. The in- theH26αinacompactarea(<0.06′′)suggeststhatatemper-
L
ferredLymancontinuumfluxrequiresamassivestarequivalent aturegradientispresentalongtheradiusoftheUCHIIregion
toazero-agemean-sequencestaroftypeO8locatedinsidethe W51e2-W.
UC HII regionW51e2-W(Panagia1973), whichis consistent Observationsof RRLs in a wide rangeatwavelengthsfrom
withtheresultofO7.5inRudolphetal.(1990).Alternatively,a centimeterstosubmillimetersalsoofferameanstofurtherex-
clusterofB-typestarscanalsoberesponsiblefortheionization. plorephysicalconditionsofW51e2-W.Table3summarizesthe
Thetotalionizedmasswithin0.5′′ ofW51e2-Wis∼0.02M , reliable measurements of three RRLs from W51e2, namely,
⊙
whichagreeswiththevaluederivedbyGaumeetal.(1993). H26α, H53α, and H66α. We plotted the spectral profiles of
Forthehyper-compactcore(θ <0.06′′)asobservedwiththe thethreeRRLstogether,andeachoftheprofileshasbeenmul-
s
H26αline,theelectrontemperaturecanbeestimatedfromthe tipliedby a factorof ν26α/νnα, the ratio ofthe H26αlinefre-
H26αlineandthecontinuumfluxdensityat354GHz(0.85mm) quency to that of RRLs (Figure 3). In the case of optically
ontheassumptionofopticallythin,LTEcondition: thinandno-pressurebroadening,thethreelineprofilesshould
match each other. ComparingH53α with H26α, we find that
T∗ = 6985 ν 1.1 ∆VH26α - 1 S0.85mm × thepeakintensitiesofthetwolinesagreewitheachother,which
e α(ν,T) GHz kms- 1 S suggeststhat,ingeneral,boththeH53αandH26αlinesareun-
"(cid:18) e (cid:19)(cid:16) (cid:17) (cid:18) (cid:19) (cid:18) H26α (cid:19) deranopticallythin,LTEcondition. Severalvelocitypeaksin
0.87
1 the H26α line indicate multiple kinematic components in the
. (6) hyper-compactHII core. The H53α line is characterizedbya
1+N(He)!#
N(H) relativelysmooth profile with large velocitywings. The large
NATUREOFW51e2:MASSIVECORESATDIFFERENTPHASESOFSTARFORMATION 7
(a) (b)
(c) (d)
FIG.5.— TheSEDforW51e2cores.(a):SEDfittingforW51e2-Wwithafree- freeemissionmodel.Blackpointsdenotethemeasurementsfromthiswork,and
theasterisksarefromRudolphetal.(1990)andGaumeetal.(1993).AllfluxvaluesarelistedinTable2.(b):TheSEDfittingfortheW51e2-Ecorewithathermal
dustemissionmodel. ThearrowsdenotetheupperlimitsfromRudolphetal.(1990)andGaumeetal.(1993). (c): TheSEDfittingfortheW51e2-NWcore. The
upperlimitofthefluxdensityat43GHz(7mm)isfromourdata(3σofthe7mmimage).(d):TheSEDfittingfortheW51e2-Ncore.
velocitywingsobservedintheH53αlinecanbeexplainedby
TABLE3
thelargevelocitygradientinthelower-densityelectrongasof
the ionized outflow (Gaumeetal. 1993). However, the line PARAMETERSOFRRLSDERIVEDFROMOURDATA∗
intensity of H66α appears to be significantly weaker than the H26α H53α H66α
othertwohigh-frequencyRRLs. Thepeakofthemodifiedline RestFrequency(GHz) 353.623 42.952 22.364
profile of H66α is a factor of ∼5 less than that of the other Measurements
two. Considering the fact that the frequency of H66α is be- Ipeak(Jy) 1.29±0.02 0.137±0.001 0.015±0.003
lowtheturnoverfrequencyofν0≈27GHz,wefoundthatthe VLSR(kms- 1) 59.1±0.2 60.4±0.2 62.1±1.3
H66αlinefromtheionizedgasisobscuredseverelyduetothe ∆V (kms- 1) 23.1±0.5 32.0±0.3 35.1±8.3
self-absorption process in the hyper-compactHII core. From Derivedlinebroadenings
Equation(2),ameanopticaldepthofτc=1.6atνH66α=22.36 ∆VP(kms- 1) 0.01 1.3 6.5
GHzisfound. Theexponentialattenuationofexp(- 1.6)≈0.2 ∆VT (kms- 1) 15.0 15.0 15.0
suggeststhattheH66αlineisattenuatedmainlyduetotheself- ∆Vt (kms- 1) 17.5 28.3 31.1
absorptionintheionizedcore. Inaddition,thepressurebroad- ∆VD(kms- 1) 23.1 32.0 34.5
eningeffectwhichweakensthelowerfrequencylinesneedsto ∗All line profiles are integrated from the 0.5′′×0.5′′ region around
beassessed. thepeakpositionofH26αfrommapsofthesamebeamsizeof0.25′′.
ThetotallinewidthofRRLs(∆V)canbeexpressedbythe
Dopplerbroadening(∆V )andpressurebroadening(∆V ).Ac-
D P
cordingtoBrocklehurst&Leeman(1971)andGordon&Sorochenkowhere∆V istheturbulentbroadeningofthegasincludingbroad-
t
(2002),wehave ening due to outflows and disk rotations and n is the princi-
palquantumnumberofthetransition. Wecalculatedbroaden-
T ∆V 2
∆VD=kms- 1s0.0458× Ke + kmst- 1 , (8) imngeasnoenletchteromnedaennseitlyec(tNron≈t3e×m1p0e5ractmur-e3)(Toeft≈he49U0C0KH)IIarnedgiothne.
(cid:18) (cid:19) (cid:18) (cid:19) e
∆VP=3.74×10- 14kms- 1n4.4 mλm cmNe- 3 × Tprheessthuerermbarolabdroenadinegniinsga(s∆trVoTn)gisfu1n5cktimonso- 1fftohreTper=in4c9i0p0alKq.uTahne-
T - 0.1 (cid:18) (cid:19)(cid:18) (cid:19) tfuremquneunmcybleirn,eis..e.F,o∆rVthPeihnicgrhe-afsreesqudernacsyticlianlley(Hto2w6aαr)d,tthheeplorews--
Ke , (9) surebroadening(∆VP=0.01kms- 1)canbeignored,whilefor
(cid:18) (cid:19) the H66α line, the pressure broadening of ∆V =7km s- 1 be-
P
∆V = ∆V2+∆V2, (10) comessignificant. Thepressurebroadeningis∆V =76kms- 1
D P P
q
8 SHI,ZHAO,&HAN
fortheH92αline,whichcanactuallywashouttheprofileofthe surroundingmolecular gas as indicated by both the HCN and
lineemission. Thus,inadditiontotheopacityeffect,thepres- COabsorptionlines(seeSection2.4). Thisproposedscenario
surebroadeningmayalso makeaconsiderablecontributionto forW51e2-Eisalsoconsistentwiththehourglass-likemagnetic
diminishtheH92αlinewhichwasnotdetectedintheobserva- fields inferred from polarization measurements of Tangetal.
tionsofMehringer(1994). Theturbulentand/orthedynamical (2009) for the W51e2 region. The configuration center of B
motionsoftheionizedgascontributeatotalof∆V ≈18kms- 1 vectorscoincideswith the peak position of the W51e2-E dust
t
in the line broadening for the H26α line and ∆V ≈28 and core.
t
31km s- 1 for the H53α and H66α lines, respectively. The From a Gaussian fitting to the absorption spectrum of the
H26α line traces the high-density ionized gas in the hyper- HCN(4-3)towardthecenterofW51e2-E(Figure4(c)),wede-
compact core or the ionized disk (if it exists) and is less sen- terminedthepeaklineintensity∆I =- 1.3±0.1Jybeam- 1,the
L
sitive to the lower-density outflow which produces the broad linecentervelocityV =56.4±0.2kms- 1andthelinewidth
HCN
wingsinthelineprofiles.Theturbulentand/ordynamicalbroad- of ∆V =9.5±0.4km s- 1. The infall velocity of the gas is
HCN
ening becomes dominant in the H53α and H66α lines, and theoffsetbetweenthecentervelocityandthesystematicveloc-
theselowerfrequencyRRLs(H53αandH66α)bettertracethe ity, i.e., V =V - V ≈2.5 km s- 1. The peak continuum
in HCN sys
lower-densityelectrongasintheionizedoutflow. intensityof W51e2-Eat 0.85mm is I =1.2±0.01Jy beam- 1
C
Inshort,W51e2-Wconsistsofahyper-compactHIIcore(< (correspondingtoabrightnesstemperatureof144±1K).The
310AU)withanemissionmeasureofEM>7×1010pccm- 6 ratio of - ∆I /I ∼1 suggests thatthe absorptionline is satu-
L C
andanionizedoutflow. Anearly-typestarequivalenttoanO8 rated. Thus, a lower limit on the optical depth (see Qinetal.
star (or a cluster of B-type stars) is postulatedto have formed 2008) is τ >3. Assuming that the excitation temperature
HCN
withinthehyper-compactHIIcore. Thebroadeningoftheline ofHCN(4-3)equalsthedusttemperature,100K,weobtaineda
profilesisdominatedbytheDopplerbroadeningincludingboth lowerlimitofHCNcolumndensityof7.9×1015cm- 2. Taking
thermalandturbulence/dynamicalmotions. Pressurebroaden- [H ]/[HCN]∼0.5×108(Irvineetal.1987)andthesizeofthe
2
ing becomes significant only for the H66α line and the lines infallregionas0.75′′ (∼4000AU, see Figure4(a)),we found
withalargerprincipalquantumnumber. that the hydrogen volume density N ∼6.9×106cm- 3. The
H2
infallrateofthegascanbeestimatedby
3.2. W51e2-E:Amassiveproto-stellarcore θ 2 D 2 V
W51e2-Eislocated∼0.9′′eastoftheUCHIIregionW51e2- dM/dt =2.1×10- 5M⊙yr- 1 arcisnec kpc kmins- 1 ×
W. It is the brightest source in the 0.85mm continuumimage (cid:18) (cid:19) (cid:18) (cid:19) (cid:18) (cid:19)
N
(Figure1). Theflux densitydropsdrasticallyat7mm (Figure H2 , (12)
2(b)).Itcannotbedetectedatlongerwavelengths.TheSEDof 106cm- 3
the continuum emission from W51e2-E (Figure 5(b)) appears where 2θi(cid:18)n =0.75′′ i(cid:19)s the diameter of the infalling region, D
toarisefromadustcoreassociatedwithproto-stars. is the distance to the source,Vin is the infall velocity, and the
denWsiittyhfthroemRaayhleoimgho-gJeenaenosuasp,pisrootxhiemrmatiaolnd,uthsteccoorneticnaunumbeflduex- rmivoeldecpualraarmmeatessrsraatlioowmer/mlimH2it=o1f.i3n6failslraastseu1m.3ed×.1W0-i3thMth⊙eydr-e1-
scribedas(Launhardt&Henning1997) isinferred,suggestingthatW51e2-Eistheaccretioncenterof
theW51e2complex.
β
κ ν ν 2
Sd(ν)=6.41×10- 6Jy cm20g- 1 ν GHz × 3.3. W51e2-NWandW51e2-N
M T (cid:18) D -(cid:19)2 (cid:18) 0(cid:19) (cid:16) (cid:17) W51e2-NWwasclearlydetectedbyTangetal.(2009)inthe
d d , (11) continuumemissionat0.87mm. Theyalsofoundasignificant
M⊙ K kpc concentrationofBfieldsatthee2-NW.Genzeletal.(1981)and
(cid:18) (cid:19)(cid:18) (cid:19)(cid:18) (cid:19)
where κ is the dust opacity at frequency ν , β is the power- Imaietal.(2002)detectedbrightandcompactH Omasersnear
0 0 2
law indexof the dustemissivity, M is the mass of dust, T is thissource. Wealsodetectedlocalenhancedcontinuumemis-
d d
the dusttemperatureinK, andD isthedistancetothe source. sion at 0.85 and 1.3mm. No significant continuum emission
Weadoptthedustopacityvalueofκ(λ=1.3mm)=0.8cm2g- 1 has been detected at 7 and 13mm (Figure1 and 2), and there
(Ossenkopf&Henning1994;Launhardt&Henning1997)and arenosignificantabsorptionlinesofHCN(4-3)andCO(2-1)in
havethreefreeparametersforthedustcore(M ,T ,andβ). this region. We fitted the SED of continuum emission with a
d d
ThesolidlineinFigure5(b)showsthebestleast-squaresfit- thermaldustmodelas shownin Figure 5(c). A lowerlimit of
ting to the observed data. The index of β =0.26±0.08 de- β ≥0.35 is derived from the fitting, which leads to an upper
rivedfromourfittingindicatesthatemissionfromthee2-Ecore limit of M T ≤39M K. Assuming the dust temperature in
d d ⊙
mightbedominatedbylargegrainsofdust(Miyake&Nakagawa thisregiontobe∼100K,weinferredanupperlimitof40M
⊙
1993; Chenetal. 1995; Andrews&Williams 2007). In addi- in the total gas mass if the ratio of H gas to dust is equal to
2
tion, becauseofa lackoffluxdensitymeasurementsathigher 100. W51e2-NWappearsto be a massivecoreata veryearly
frequenciesorshorterwavelengths,thedustmassandtempera- phaseofstarformation.
turearedegenerateinourbestfitting,i.e.,M T =142±9M K. Located ∼ 2′′ north from W51e2-E, W51e2-N is detected
d d ⊙
Remijanetal.(2004)derivedthekinetictemperatureinW51e2 incontinuumemissionfromourlower-resolution(1.4′′×0.7′′)
tobe153±21Kfromthehigh-resolutionobservationsofCH CN, image at 1.3mm (Figure 2(a)). This source appears to be re-
3
whichisclosetothedusttemperatureT =100KusedinZhangetal. solvedoutwith a high angularresolution(0.3′′×0.2′′) obser-
d
(1998). AdoptingthatT =100Kandassumingthattheratioof vationat0.85mm. Nosignificantcontinuumemissionwasde-
d
H gastodustis100,wefoundthatthetotalmassof∼140M tected at the longer wavelengths. The SED consisting of the
2 ⊙
is in the W51e2-E core. Emission from the W51e2-E region fluxdensityat1.3mmandtheupperlimitsat7,13,and36mm
has been slightly resolved, so that the dust core of W51e2-E is also fitted with a thermal dust model. From the fitted val-
might host a number of proto-stellar cores which accrete the uesoftheparametersβ≥0.75andMdTd≤68M⊙K,wefound
NATUREOFW51e2:MASSIVECORESATDIFFERENTPHASESOFSTARFORMATION 9
anupperlimitofM .70M ifT ∼100KandtheH gasto HCNmolecularlineinabsorptionagainstthecompactHIIre-
d ⊙ d 2
dustratiois100. Nomaseractivitieshavebeendetectedfrom gion indicates that the significant accretion onto this core has
W51e2-N, suggesting that W51e2-N is probably a primordial been stopped. The line ratios between the H26α, H53α, and
molecularclumpinW51e2. H66αshowthatboththeH26αandH53αlinesfromW51e2-
W are in an optically thin, LTE condition. The emission of
3.4. ApropagatingscenarioofstarformationinW51e2 the H66α line from the hyper-compactcore appears to be at-
tenuated mainly due to the self-absorption in this HII region.
Apparently,inthemassivemolecularcoreW51e2,thegravi-
The line profile of the H26α appears to be dominated by the
tationalcollapseoccurredfirstatW51e2-WwhereanO8staror
Dopplerbroadeningduetothethermalmotionsofthehotelec-
aclusterofB-typestarswereformed. Basedontheabsorption
tronswhilethelinebroadeninginboththeH53αandH66αis
spectrumofthe HCN line, thereappearstobelittle molecular
dominatedbytheDopplereffectduetothedynamicalmotions
gas presentin the immediate environs. The accretion appears
of the ionized outflow. The pressure broadening in both the
to be pausedby the intensiveradiationpressure fromthe cen-
tral ionized star(s). Most of the surrounding gas (∼0.02M ) H26αandH53αlinesisnegligibleandbecomessignificantfor
⊙ theH66αlinecorrespondingto∆V ∼7kms- 1.
has been ionized. The thermal pressure in the UC HII region P
2. W51e2-Eisamassive(∼140M )dustcoreassuggested
drives an ionized outflow from the ionized core. The submil- ⊙
based onthe SED analysis. No hydrogenrecombinationlines
limeter observations of dust emission and molecular line ab-
andnoradiocontinuumemission(λ>1cm)havebeendetected
sorptionshowthatthemajoraccretionnowhasbeenre-directed
toW51e2-E,0.9′′ (∼4600AU)eastfromW51e2-W.W51e2-E from this region, which suggests that W51e2-E is the major
corehostingmassiveproto-stellarobjects.Boththeabsorptions
becomesthenewdominantgravitationalcenteraccretingmass
from the surroundingswith a rate of 10- 3 M yr- 1. Star for- of HCN(4-3) and CO(2-1) are redshifted with respect to the
⊙ systematic velocity, 53.9±1.1km s- 1, indicating that a large
mation activities take place in this massive core, as shown by
amountofmoleculargasmovestowardthemassivecore.From
maseractivities,outflow,andorganizedB-fieldstructure.Since
no free- free emission and no RRLs have been detected, the theratiosoftheabsorptionlinetocontinuum,alargeinfallrate
massivecoreW51e2-Elikelyhostsoneormoremassiveproto- of >1.3×10- 3M⊙yr- 1 is inferred from the HCN absorption
line,correspondingtotheaccretionradiusof2000AU.
stars. Theoffsetoftheradialvelocitiesbetweenthetwocores,
W51e2-E and W51e2-W, is δV =59.1- 53.9=5.2km s- 1. If 3. W51e2-NW appears to be a massive (. 40M⊙) core at
an earlier phase of star formation. Given its location in the
W51e2-WandW51e2-EaregravitationallyboundandW51e2-
projectedpathofthemolecularoutflowfromW51e2-E,thestar
W is circularlyorbitingaroundW51e2-E,W51e2-Ehasa dy-
formationactivitiesofW51e2-NWarelikelytobetriggeredby
namicalmassof>140M⊙,ingoodagreementwiththemass
theoutflowimpactonthemediumintheW51e2complex.
derivedfromtheSEDanalysis.
4. W51e2-Nappearstobearelativelylessdensebutmassive
A bipolar outflow from W51e2-E has been suggested from
clump (. 70M ) from which no obvious evidence has been
theobservationsofmolecularlines. Theimpactoftheoutflow ⊙
foundforstarformationactivities.
onthemediuminits pathmaytriggerthecollapseofthesub-
core W51e2-NW and induce further star formation activities
there. W51e2-N represents a gas clump with a considerable We thank the referee for helpful comments, Dr. S.-L. Qin
amount of mass for the star formation. This speculative sce- for providing help with the data reduction and analysis, and
narioappearsto reasonablyexplainwhatwehaveobservedin Professor Y.-F. Wu for helpful discussions. H. Shi and J.-L.
theW51e2complex. HanaresupportedbytheNationalNaturalScienceFoundation
(NNSF) of China (10773016, 10821061, and 10833003) and
4. SUMMARY AND CONCLUSIONS theNationalKeyBasicResearchScienceFoundationofChina
(2007CB815403).
Wehavepresentedhigh-resolutionimagesoftheW51e2re-
gionat0.85,1.3,7,and13mmforcontinuumandhydrogenre-
combinationlines(H26α,H53α,andH66α)andthemolecular
lines(HCN(4-3)andCO(2-1)). TheW51e2complexhasbeen
resolved into four distinct components, W51e2-W, W51e2-E, REFERENCES
W51e2-NW,andW51e2-N.Wehavecarriedoutacomprehen- Andrews,S.M.&Williams,J.P.2007,ApJ,659,705
siveanalysisofthecontinuumSED,theRRLs,andthemolecu- Brocklehurst,M.&Leeman,S.1971,Astrophys.Lett.,9,35
Chen,H.,Zhao,J.-H.,&Ohashi,N.1995,ApJ,450,L71+
larabsorptionlinesandfoundthatthefourcoresareatdifferent Fish,V.L.&Reid,M.J.2007,ApJ,670,1159
phasesofmassivestarformation. Gaume,R.A.,Johnston,K.J.,&Wilson,T.L.1993,ApJ,417,645
Genzel,R.,Becklin,E.E.,Moran,J.M.,Reid,M.J.,Jaffe,D.T.,Downes,D.,
1. W51e2-W,associatedwiththeUCHIIregion,istheonly
&Wynn-Williams,C.G.1982,ApJ,255,527
sourcefromwhichtheH26α,H53α,andH66αlineshavebeen Genzel, R.,Downes, D.,Schneps, M.H.,Reid, M.J.,Moran, J.M.,Kogan,
detected. The Lyman continuum flux inferred from the SED L.R., Kostenko, V.I., Matveenko, L.I., &Ronnang, B.1981, ApJ,247,
1039
analysissuggeststhata massiveO8starora clusterofB-type
Gordon, M. A. & Sorochenko, R. L., eds. 2002, Astrophysics and Space
starshavebeenformedwithintheHIIregion. Theunresolved Science Library, Vol.282, RadioRecombination Lines.TheirPhysicsand
H26αlineemissionregionsuggeststhepresenceofahot(T∗= AstronomicalApplications
e Ho,P.T.P.&Young,L.M.1996,ApJ,472,742
12,000±2,000K), hyper-compactionizedcorewith a veloc- Imai,H.,Watanabe,T.,Omodaka,T.,Nishio,M.,Kameya,O.,Miyaji,T.,&
ityof59.1±0.2kms- 1,alinearsizeof<310AU,andalarge Nakajima,J.2002,PASJ,54,741
emission measure EM> 7×1010pc cm- 6 (or a high density Irvine,W.M.,Goldsmith,P.F.,&Hjalmarson,A.1987,inAstrophysicsand
SpaceScienceLibrary,Vol.134,InterstellarProcesses,ed.D.J.Hollenbach
Ne >7×106 cm- 3). The H53α and H66α line images show &H.A.Thronson,Jr.,561–609
an SW elongation from the hyper-compactionized core, sug- Keto,E.&Klaassen,P.2008,ApJ,678,L109
Keto,E.,Zhang,Q.,&Kurtz,S.2008,ApJ,672,423
gesting either an expansion or an outflow of the ionized gas
Kundu,M.R.&Velusamy,T.1967,Annalesd’Astrophysique,30,59
with relatively lower density. No significant detection of the Launhardt,R.&Henning,T.1997,A&A,326,329
10 SHI,ZHAO,&HAN
Martin,A.H.M.1972,MNRAS,157,31
Matsakis,D.N.,Evans,II,N.J.,Sato,T.,&Zuckerman,B.1976,AJ,81,172
Mehringer,D.M.1994,ApJS,91,713
Mezger,P.G.&Henderson,A.P.1967,ApJ,147,471
Miyake,K.&Nakagawa,Y.1993,Icarus,106,20
Ossenkopf,V.&Henning,T.1994,A&A,291,943
Panagia,N.1973,AJ,78,929
Qin,S.-L.,Zhao,J.-H.,Moran,J.M.,Marrone,D.P.,Patel,N.A.,Wang,J.-J.,
Liu,S.-Y.,&Kuan,Y.-J.2008,ApJ,677,353
Remijan, A., Sutton, E.C., Snyder, L.E.,Friedel, D.N.,Liu, S.-Y.,&Pei,
C.-C.2004,ApJ,606,917
Rubin,R.H.1968,ApJ,154,391
Rudolph,A.,Welch,W.J.,Palmer,P.,&Dubrulle,B.1990,ApJ,363,528
Sault,R.J.,Teuben,P.J.,&Wright,M.C.H.1995,inAstronomicalSocietyof
thePacificConferenceSeries,Vol.77,AstronomicalDataAnalysisSoftware
andSystemsIV,ed.R.A.Shaw,H.E.Payne,&J.J.E.Hayes,433–+
Scott,P.F.1978,MNRAS,183,435
Sollins,P.K.,Zhang,Q.,&Ho,P.T.P.2004,ApJ,606,943
Tang,Y.,Ho,P.T.P.,Koch,P.M.,Girart,J.M.,Lai,S.,&Rao,R.2009,ApJ,
700,251
Westerhout,G.1958,Bull.Astron.Inst.Netherlands,14,215
Xu, Y., Reid, M. J., Menten, K. M., Brunthaler, A., Zheng, X. W., &
Moscadelli,L.2009,ApJ,693,413
Young,L.M.,Keto,E.,&Ho,P.T.P.1998,ApJ,507,270
Zhang,Q.&Ho,P.T.P.1997,ApJ,488,241
Zhang,Q.,Ho,P.T.P.,&Ohashi,N.1998,ApJ,494,636