The Formation and Early Evolution of Young Massive Clusters

0.54 MB·

by  Steven N. Longmore

The Formation and Early Evolution of Young Massive Clusters
Steven N. Longmore
LiverpoolJohnMooresUniversity
J.M. Diederik Kruijssen
Max-PlanckInstitutfu¨rAstrophysik
NateBastian
4
LiverpoolJohnMooresUniversity
1
0
2 JohnBally

n UniversityofBoulder,Colorado
a
J
JillRathborne
6
1 CSIROAstronomyandSpaceScience

]
Leonardo Testi
A
EuropeanSouthernObservatory
G
.
h Andrea Stolte
p
ArgelanderInstitutfu¨rAstronomie,Bonn
-
o
r
t JamesDale
s
a ExcellenceClusterUniverse
[

1 Eli Bressert
v CSIROAstronomyandSpaceScience
5
7
1 JoaoAlves
4 UniversityofVienna
.
1
0 We review the formation and early evolution of the most massive (> few 104M⊙) and
4 dense (radius of a few pc) young stellar clusters, focusing on the role that studies of these
1 objectsin our Galaxy can play inour understanding of star and planet formation as awhole.
:
v Comparing the demographics of young massive cluster (YMC) progenitor clouds and YMCs
i acrosstheGalaxyshowsthatgasintheGalacticCentercanaccumulatetoahighenoughdensity
X
thatmolecularcloudsalreadysatisfythecriteriausedtodefineYMCs,withoutformingstars.In
r thiscaseformationcanproceed“insitu”–i.e.thestarsformatprotostellardensitiesclosetothe
a
finalstellardensity. Conversely,inthedisk,thegaseitherbeginsformingstarswhileitisbeing
accumulatedtohighdensity,ina“conveyorbelt”mode,orthetimescaletoaccumulatethegas
tosuchhighdensitiesmustbemuchshorterthanthestarformationtimescale. Thedistinction
between the formation regimes in the two environments is consistent with the predictions of
environmentally-dependent density thresholds for star formation. This implies that stars in
YMCsofsimilartotalmassandradiuscanhaveformedatwidelydifferentinitialprotostellar
densities. Thefactthatnostrong,systematicvariationsinfundamentalproperties(suchasthe
IMF) are observed between YMCsin the disk and GalacticCenter suggests that, statistically
speaking,stellarmassassemblyisnotaffectedbytheinitialprotostellardensity.Wethenreview
recenttheoreticaladvances andsummarizethedebateonthreekeyopen questions: theinitial
(proto)stellar distribution, infant (im)mortality and age spreads within YMCs. We conclude
that:(i)theinitialprotostellardistributionislikelyhierarchical;(ii)YMCslikelyexperienceda
formationhistorythatwasdominatedbygasexhaustionratherthangasexpulsion; (iii)YMCs
aredynamicallystablefromayoungage; and(iv)YMCshaveagespreadsmuchsmallerthan
theirmeanage. Finally,weshowthatitisplausiblethatmetal-richglobularclustersmayhave
formedinasimilarwaytoYMCsinnearbygalaxies.Insummary,thestudyofYMCformation
bridges star/planet formation in the solar neighborhood to the oldest structures in the local
Universe. 1
1. OverviewofYMCsandtheirroleinaglobalunder- populations of galaxies (LadaandLada 2003). Only a
standingofstarandplanetformation small fraction are simultaneously both massive and dense
enough to remain gravitationally bound long after their
1.1. Motivation
formation and subsequent removal of the remaining natal
Asexemplifiedbythereviewsinthisvolume,thebasic moleculargascloud. Beingabletostudyanensemblesin-
theoreticalframeworkdescribingtheformationofisolated, glestellar populationlongafterformationoffersmanyad-
low-massstarsisnowwellestablished. Thisframeworkis vantages,notleastofwhichistheabilitytoretrospectively
underpinnedbydetailedobservationalstudiesoftheclosest investigate the conditions under which the stars may have
star-formingregions. Buthowtypicalisthestarandplanet formed.
formationinTaurus,Perseus,orevenOrioncomparedtothe The mostextremeexamplesof such stellar systems are
formation environment of most stars across cosmological globularclusters,whichformedattheearliestepochsofthe
timescales? Universeandsurvivetothepresentday(BrodieandStrader
The fact that half of the star formation in the Galaxy 2006). Oneoftheground-breakingdiscoveriesoftheHub-
is currently taking place in the 24 most massive giant bleSpaceTelescopewasthatmassivestellarclusters,with
molecular clouds (Leeetal. 2012), suggests that even in propertiesthatrivalthosefoundinglobularclustersinterms
the Milky Way at the present day, star formation regions ofmassandstellardensity,arestillformingintheUniverse
in the local neighborhood are not typical. Environmen- today(Holtzmanetal.1992).Theseyoungmassiveclusters
tal conditions were likely even more different at earlier (YMCs) have stellar masses and densities orders of mag-
epochs of the Universe. The epoch of peak star forma- nitude largerthan typicalopen clusters and comparableto
tion rate density is thought to lie between redshifts of those in globularclusters. Crucially, theyarealso gravita-
2 and 3 (Madauetal. 1998; HopkinsandBeacom 2006; tionally bound and likely to be long-lived. As such, these
Bouwensetal.2011;Mosteretal.2013),whentheaverage stellarsystemsarepotentiallylocal-universe-analogsofthe
gas surface density (and hence inferred protostellar densi- progenitorsofglobularclusters.
ties) were significantly higher (see e.g. CarilliandWalter At the same time, the apparent continuum of young
2013). Giventhatmoststarsinthesolarneighborhoodare cluster properties (e.g. Bressertetal. 2010), suggests that
atasimilaragetotheSun(∼5Gyr: Nordstro¨metal.2004), YMCs merely represent extreme examples of their less
does it then make sense to compare the planetary popula- massive and dense counterparts– open clusters. As such,
tionsobservedaroundthesestarstotheprotoplanetarydisks characterizingandunderstandinghowYMCsformiscriti-
innearbystarformingregions? caltohelpmaketheconnectionbetweentherangeofphysi-
The fundamental question underlying this line of rea- calconditionsforstarandplanetformationbetweenGalac-
soning is, “Does the process of stellar and planetary mass ticandextra-galacticclusterformationenvironments.
assembly care about the environment in which the stars
form?”Iftheansweris“No”,thenstudyingtheneareststar 1.2.1. YMCs: definitionandgeneralproperties
formationregionswilltellusallthereistoknowaboutstar In their recentreview, PortegiesZwartetal. (2010) de-
andplanetformation. Iftheansweris“Yes”,itiscrucialto fined YMCs as stellar systems with mass & 104M⊙ and
understandhowandwhytheenvironmentmatters. Thepo- with ages less than 100Myr but substantially exceeding
tential implicationsof these answers providea strongmo- the current dynamical time (the orbital time of a typical
tivation for comparing the process of star and planet for- star). While the ultimate longevity of a stellar system
mationinextremeenvironments,withthatinnearby,well- will depend on the environment it experiences over time
studied,lessextremestarformationregions. (Spitzer 1958), this last criterion effectively distinguishes
betweenpresentlyboundclustersandunboundassociations
1.2. YoungMassiveClusters: idealprobesofstarand
(see§3.2.1forfurtherdetails).
planetformationinextremeenvironments
Given previous confusion in the literature caused by
Stars are observed to form in a continuous range of loose and varied definitions of what constitutes a stellar
environments and densities, from isolated molecular gas cluster (see § 3.2.1), it is important to point out the im-
clouds expected to form single low-mass stellar systems plications of the above criteria. Firstly, there are many,
(e.g. B68: Alvesetal. 2001), through to giant molecular well-known,massiveassociationsofstarswhichdonotpass
cloud complexes expected to form hundreds of thousands thesecriteria(e.gtheCygnusOBassociationandtheOrion
of stars across the full stellar mass range (see Molinari Nebular Cluster). Secondly, YMCs that do pass these cri-
et al. this volume). Similarly, young stellar systems are teria(e.g. Westerlund1,NGC3603,Trumpler14)maylie
found at a continuous range of mass and stellar densities within a muchlargerstellar association, which as a whole
(Bressertetal.2010).Giventhisapparentlycontinuousdis- doesnotpassthesecriteria. Thisisadirectconsequenceof
tribution in mass and density of both gas and stars, what the ‘continuum’of stellar propertiesdiscussed above. We
motivatesadefinitionforadistinctclassofstellarsystems emphasizethatthe focushereis onthe YMCsandnotthe
beyondmerephenomenology? more distributed stellar populations. This will impact the
Theanswerliesinthefactthatmoststellarsystemsdis- discussion in §3 on gas expulsion, longevityand the pres-
solve shortly after forming, thereby feeding the field star
2
enceofagespreadswithinclusters. sivestars. Whiletheprogenitorsofmanylate-Oandearly-
To date, nearly one hundred YMCs have been discov- Bstarshavebeenidentified,precursorstostarsdestinedto
ered in the local group and out to distances of a few be hundreds of solar masses still prove elusive. Identify-
Mpc. The properties of many of these are catalogued in ingsuchprecursorswillhelpinthetheoreticalchallengeto
PortegiesZwartetal.(2010). Forconvenience,wesumma- understandhowthemostmassivestarsform(seeTanetal.
rize their characteristic propertiesbelow. YMCs typically thisvolumeforareviewonhigh-massstarformation).
haveradiiof∼1pcandcorestellardensities≥103M⊙pc−3. YMCsoccupyauniquepositioninunderstandingcluster
Theyaregenerallyspherical,centrally-concentratedandof- formation.Asabridgeintheapparentcontinuumofcluster
ten mass segregated(i.e. more massive stars are preferen- mass and density distributions between open and globular
tially found towards the center of the cluster). The initial clusters,studyingtheirglobalandenvironmentalproperties
clustermass distributionisnottrivialto measure,butover canprovideinsightintowhatconditionsarerequiredinor-
many orders of magnitude in mass appears to be reason- der for bound clusters to form. Is there a single, scalable
ablywellapproximatedbyapowerlaw,dN/dM ∝ M−2, formationmechanismapplicabletoallclusters? Oraread-
across all environments. YMCs are found predominantly ditional mechanisms required to accumulate such a large
in starburst galaxies and mergers – a couple of thousand gasmassin asmallvolumeforthemostmassiveclusters?
are known to exist in the Antennae and NGC 3256, for YMCsmaybeusedasadirectprobetounderstandthecon-
example. These YMCs are typically more massive than ditionsrequiredforglobularclusterformation.
those found in the Local Group and Milky Way. In the
1.3. Scopeofthereview
localuniverse(i.e. notstarbursts/mergers),YMCsaretyp-
ically found in the disks of galaxies. Globular clusters Several fundamental, unanswered questions about the
are predominantly found in galactic halos. Rotation has formationandearlyevolutionofYMCscurrentlylimittheir
been observed in one YMC (R136: He´nault-Brunetetal. utility as probes of star and planet formation in extreme
2012a) as well as two intermediate age massive, dense environments. For example, while the spatial distribution
clusters (GLIMPSE-C01, NGC 1846: Daviesetal. 2011; of stars in YMCs older than a few Myr is relatively well
Mackeyetal.2013). Giventhedifficultyinmeasuringrota- known (King 1966; Elsonetal. 1987), it is not clear how
tion,itiscurrentlyunknownhowcommonthispropertyis thisrelatestotheinitialprotostellarorgasdistribution(e.g.
amongYMCs. Rolffsetal. 2011). Any initial substructure that existed in
the gas and protostars is erased quickly (McMillanetal.
1.2.2. TheroleofYMCsinthebroadercontextofplanet,
2007). Therefore, if the stars actually formed at a much
starandclusterformation
lowerdensity–andhenceinamuchlessextremeenviron-
ThepropertiesofYMCsmakethemidealprobesofstar ment than assumed from the present-day stellar density –
andplanetformationinextremeenvironments. Starsform- andthengrewintoamassive,denseclusterovertime,there
ingatsuchhigh(proto)stellardensitiessufferthemaximal would be little evidence of this in the final stellar surface
effectsof feedbackfromsurroundingstars. Also, the very densitydistributionasthestructurewouldhavebeenerased
short dynamical time increases the likelihood of interac- byviolentrelaxation. Apotential,newmethodofderiving
tions with nearby stars at all stages of the formation pro- theinitialconditionsofclusterformationaposterioriwould
cess. Therefore, studying the formation of stars within a be to consider quantitiesthatare conservedduringviolent
YMCcomparedtolowstellardensitysystems,offersanop- relaxation, such as the degree of mass segregation, and to
portunitytoquantifyhowdynamicalencountersandstellar combine these with a measure of the remaining substruc-
feedbackaffecttheprocessofstellarmassassembly. ture. Collapsing, virialised and unboundstellar structures
YMCscontainaverylargenumberofstarsofasimilar may follow distinct evolutionaryhistories in the plane de-
age (age spreads .1Myr: see § 3.2.3). These stars likely finedbythesequantities(Parkeretal.2013).
formedfromthesamegascloud,sowereborninthesame Moregenerally,itisnotclearifallclustersofthesame
globalenvironmentalconditionsandhavethesamechemi- massandradiusformfromgaswithsimilarproperties.Are
calcomposition.ThismakesYMCprecursorgascloudsthe there different ways to form bound clusters of similar fi-
perfecttestbedstostudytheoriginofthestellarinitialmass nal stellar properties? If so, and if stellar mass assembly
function(IMF).Forexample,bystudyingYMCprogenitor dependson the protostellarenvironment,itis importantto
cloudsbeforetheonsetofstarformation,itshouldbepos- understandhow and when these differentmechanismsop-
sibletodetermineifthefinalstellarmassissetbytheinitial erate.
massdistributionofgasfragments,oralternatively,bythese Understandingthesequestionsrequiresmakingthelink
initialfragmentssubsequentlyaccretingunboundgasfrom between the evolution of the initial progenitor gas clouds
the surroundingenvironment(see the review by Offner et towardsthefinal,gas-freestellarpopulations.However,the
al.,thisvolumeforamoredetaileddiscussionontheorigin propertiesof YMCs have been derived almost exclusively
oftheIMF). fromoptical/infraredobservations.Thishasstronglybiased
YMC precursor clouds are also, statistically-speaking, YMCdetectiontowardsclusterswithrelativelylowextinc-
thebestplacetosearchfortheprogenitorsofthemostmas- tions(A . 30),preferentiallyselectingclusterswhichare
v
3
already gas-free – i.e. older than a (few) Myr. This bias tion is almost instantaneous once the gas is accumulated.
meansthatverylittle is knownaboutYMCsyoungerthan It is therefore very unlikely that a YMC progenitor cloud
this,ortheirprogenitorgasclouds. with massM andradiusR* with nosignsofactivestar
gas
In this review we focus on: (i) the initial conditions formation would be observed, but significant numbers of
of proto-YMCs, (ii) the gas-rich, first (few) ∼Myr in the suchcloudsexhibitingongoingstarformationshouldbeob-
lifeofYMCsastheyareformingstars,and(iii)theevolu- served. Inthelattercase, dubbed‘in-situslowformation’,
tionshortlythereafter. Thisisintendedtocomplementthe theaccumulationtimeislongandstarformationisdelayed
PortegiesZwartetal.(2010)review,whichfocussedonthe until most of the mass required to build the YMC has ac-
aspectsofYMCsolderthanafewMyr. cumulated inside R*. Significant numbers of clouds with
mass close to M and radius R* but with no active star
gas
2. Molecular cloud progenitorsof YMCs – the Initial formationwouldthereforebeobservedinthiscase.
Conditions
2.1.2. Rignaist >R∗ : “conveyorbeltformation”
Understanding the formation of YMCs requires first
In this scenario, the gas that eventually ends up in the
finding samples of YMC progenitor clouds that repre-
YMCis initiallymuchmore(factorsof severalorgreater)
sent the initial conditions(i.e. before star formationcom-
extendedthanthatofthefinalcluster. Theinitial,quiescent
mences), which can be directly compared to their more
gascloudsbeginformingstarsatamuchlowerglobalsur-
evolved stellar counterparts. However, very few pre-star-
face/volumedensitythanintheprevious“in-situ”scenario.
forming YMC progenitor clouds have been identified. In
Inorderfortheproto-clustertoreachtherequiredfinalstel-
an attempt to understand the plausible range of properties
lar densities, thegasandformingstarsmustconvergeinto
for the initial molecular cloud progenitors of YMCs, we
a boundstellar system. Themostlikelyagentsto enhance
considersomesimplifiedformationscenariosbelow.
gasdensityaretheconvergenceoftwoinitiallyindependent
2.1. SimplifiedYMCformationscenarios gasflows,orthegravitationalcollapseofasinglecloud. In
thisscenario,onewouldneverexpecttoseecloudsofMinit
The most basic initial condition for YMC formation is gas
andR withnosignsofactivestarformation.
a gas reservoir with enough mass, Minit, to form a stellar ∗
gas As outlined in section §3.1, the long-term survival of
cluster of mass, M∗ ≥ 104M⊙. These two quantities are
the cluster is strongly influenced by the mechanism and
triviallyrelatedviathestarformationefficiency,ǫ,through
timescale for gas removal. The time for gas dispersal,
Mignaist = M∗/ǫ. To span the expected range of molecular t ,thereforeplacesastrongupperlimitonthetimedur-
cloudprogenitorproperties,weinvestigatetwoextremesin disp
ing which it is possible to form a cluster through conver-
theinitialspatialdistributionofthegas(i.e.beforetheonset
gence/collapse. Given a convergencevelocity, V , this
ofanystarformation)relativetothatofthefinalstellarclus- conv
sets an upper limit to the initial radius of the gas to be
ter. Firstly,weconsiderthecasewheretheinitialextentof
thegas,Rignaist,equalsthatoftheresultingcluster,R∗. Then RignaTisth=e tRim∗+escVacleonfvotrdisspta.r formation and the observed age
weconsiderthecasewhereRinitissubstantially(factorsof
gas spreads are key diagnostics for distinguishing between
severalormore)largerthanR∗. thesescenarios. We lookattheobservationalevidencefor
2.1.1. Rignaist =R∗ : “in-situformation” variationinthesepropertiesinYMCsin§3.1.
Inthisscenario,alltherequiredgasisgatheredintothe 2.2. Comparing YMC and progenitor cloud demo-
finalstar cluster volumebefore star formationcommences graphics
(i.e. in-situ star formation). In principle, a direct obser-
Wenowdemonstratehowonecanusetheobservedde-
vationalpredictionof thismodelwouldbe thatonewould
mographicsofmolecularcloudpopulations,comparedwith
expecttofindgascloudswithmassM andradiusR with
gas ∗ those of the stellar cluster populationsin the same region,
nosignsofactivestarformation. However,theprobability
totestthesedifferentYMCformationscenarios.
offindingsuchacloudunderthisscenariodependsonthe
Firstly, we assume that in a regionwith a large enough
ratioofthetime takentoaccumulatethe gaswithinthefi-
volume to sample all stages of the star/cluster formation
nalclustervolumetothetimetakenforstarformationtoget
process, the present day molecular cloud population will
underwaythere. Theveryhighdensitiesrequiredtoforma
createsimilarclustersasthoseobservedatthepresentday.
YMC implies that the gas inside the final cluster volume
Inpractice thisimpliesthatthe star formationrate, cluster
willhaveacorrespondinglyshortfree-falltimeinthissce-
formationrate and the distributionof stars into clusters of
nario. Ifstarformationhappensonadynamicaltimescale,
a given mass and density should have been constant over
thisimpliesthateitherthetimetakentoaccumulatethegas
several star formation life cycles. This seems a reason-
reservoirthere mustalso be veryshort, or thatstar forma-
able assumption for disks in nearby galaxies, but may not
tioninsidethefinalclustervolumeissomehowdelayedor
hold in mergers, starburst systems or dwarf galaxies (see
suppressedwhilethegasaccumulates.
KruijssenandLongmore2014).
In the former case, which we term ‘in-situ fast forma-
Themostmassivegasclouds(Mmax)seemtheobvious
tion’, the accumulation time is very short and star forma- gas
4
birthsitesforthemostmassiveclusters(Mmax).Ifnoexist- than the dust emission in extragalactic observations. This
∗
inggascloudshaveenoughmasstoformtheobservedmost meansthatthemassestimatesaboveprovidealowermass
massiveclusters(i.e.Mmax ≪Mmax/ǫ),thesecloudsmust limittothedetectabilityofgascloudsinCO.
gas ∗
gainadditionalmassfromelsewhere(e.g. throughmerging However, the expectedhigh volume and columndensi-
gas flows or accreting lower density gas from outside the tiesofYMCprogenitorcloudsmeansthatCO maynotbe
present-dayboundary)–i.e. “conveyorbelt”formation. the ideal molecular line tracer for identification purposes.
On the other hand, if there are gas clouds of sufficient Toillustratethispoint,wenotethatafiducialYMCprogen-
mass (i.e. Mmgaasx ≥ Mm∗ax/ǫ), then the spatial/kinematic itorcloudof105M⊙ withradius1pc(e.g. aswouldbeex-
sub-structure of this gas and the distribution of star for- pectedto forma3×104M⊙ clusterthroughin-situforma-
mation activity within these clouds can provide a key to tion,assuminga30%starformationefficiency)wouldhave
the formation mechanism. If concentrations of gas with anaveragevolumeandcolumndensityof2×104M⊙pc−3
M∗/ǫwithin∼R∗exist,thenfindingasizablefractionwith (4 × 105cm−3) and 3 × 104M⊙pc−2 (2 × 1024cm−2),
no star formationactivitywould indicate YMCs are form- respectively. This column density correspondsto a visual
ing “in-situ”. If the gas in the most massive clouds is extinction of ∼2000mag. At such high densities, even if
spatially distributed such that no sub-region of any cloud observations can resolve the gas emission down to parsec
contains a mass concentration of M∗/ǫ within ∼R∗, then scales, the CO emission will be optically-thick. Therefore
in-situ formation seems highly unlikely. In which case suchobservationscanonlyprobeaτ = 1 surface,notthe
the stars forming in the gas must converge and become bulk of the gas mass. Similar resolution observations of
gravitationally-boundbeforethestarformationcandisrupt molecular transitions with a higher critical density (com-
the cloud. Evidence for such convergence should be im- parable to thatof the averagevolume density in the YMC
printed in the gas kinematics, e.g. velocity dispersions of progenitorcloud)arerequiredtopinpointtheseclouds. As
orderVconv =(Rignaist−R∗)/tdisp. InverseP-Cygniprofiles aninterestingaside,suchhighcolumndensitiesrenderHα
and red/blue-shifted line profile asymmetries may also be –thetraditionalSFindicatorinextragalacticobservations–
observedbut care mustbe takeninterpretingsuch spectral completelyunusable. Probinggascloudswithandwithout
linediagnostics(Smithetal.2012,2013). prodigiousembeddedstarformationactivitywilltherefore
relyoncomplementaryobservationstomeasurestarforma-
2.2.1. Observationaltracersanddiagnostics tiontracerslessaffectedbyextinction(e.g. cm-continuum
We now investigate the feasibility of directly compar- emissiontogetthefree-freeluminosity,orfar-IRobserva-
ing YMC and progenitor cloud demographics given cur- tionstoderivethebolometricluminosity).
rent observational facilities. A fundamental limitation is The gas mass inferred from observations is a beam-
the distance to which it is possible to detect a precursor averaged quantity. In other words, if a gas cloud is much
cloud of a given mass. ALMA’s factor >10 improvement smaller than the observationalresolution and sits within a
insensitivitycomparedtoexisting(sub)mminterferometers lower density environment, the measured beam-averaged
makesittheoptimalfacilityfordetectinggascloudsoutto column/volume density will be lower than the true value,
largedistances. Atafrequencyof230GHz(wavelengthof leading to incomplete YMC progenitor samples. How-
1.26mm)ALMAwillachievea10σ continuumsensitivity ever,evenwhennotoperatingatitsbestresolution,ALMA
limit for a one hour integration of approximately 0.1mJy should routinely resolve the ∼pc-scale YMC progenitor
(assuming8GHzbandwidth).Assuminggasanddustprop- cloudsizesouttoseveraltensofMpc.
erties similar to those in massive star forming regions in To measure what influence the high protostellar den-
the Milky Way (gas temperature of 20K, gas:dust ratio of sity environmenthasonformingprotostarsandtheirplan-
100:1, OssenkopfandHenning (1994) dust opacities) this etary systems, it is necessary to resolve individual stel-
sensitivitylimitcorrespondsveryroughlytoamasslimitof lar systems. In practice the projected protostellar sep-
{105,107}M⊙atadistanceof{0.5,5}Mpc. Thissimplis- aration will vary, both from source to source, and as a
ticcalculationneglectsseveralsubtleties(e.g.theeffectsof function of radius within an individual region. How-
beam dilution, higher gas temperatures in vigorously star ever, relying on the fact that the average core mass is
forming systems and metallicity variations on the gas-to- ∼1M⊙, the characteristic projected separation of proto-
dust ratio and dust opacity). However, it illustrates that stars within a protocluster of mass M∗ and radius R∗ is
the gascloudpopulationspreviouslyaccessible within the proportional to R∗(M∗/M⊙)−1/2. The typical projected
LMC/SMCcannowbeprobedouttoM31/M33distances, angular separation of protostars within a protocluster as a
andsimilarstudiescurrentlybeingdoneonM31/M33GMC functionofdistancetotheprotocluster,D,is(veryroughly)
populations will be possible out to more extremely star 4(R∗/pc)(M∗/104M⊙)−1/2(D/kpc)−1arcsec. Even at the
forminggalaxieslikeM82andNGC253. maximumresolutionofALMAof0.01′′(i.e.usingthemost
Emission from the CO molecule is another standard extended 10km baselines at the highest frequency [Band
tracer of GMC populations. A combination of the exci- 9]),itwillnotbepossibletoresolveindividualstellarsys-
tation conditionsand abundance means for a gas cloud of temsinYMCprogenitorcloudsbeyondabout100kpc(i.e.
a given mass, low J transitions of CO are usually brighter LMCand SMCdistances). Themaximumangularresolu-
tion limit for ALMA is comparable to that expected from
5
future 30−40m aperture optical/infrared telescopes. For 2013).
atleast the nextseveraldecades, observationsprobingthe ThedifficultyingeneratingacompletecatalogofYMC
physics shaping the IMF in dense stellar systems must be progenitor clouds has been finding those before star for-
limitedtostarformingregionsintheLMC/SMCandcloser. mation has begun. At this early stage there is no ioniz-
Assuming it is possible to resolve individual protostel- ing radiation and the luminosity is low. Therefore, these
lar systems, the observational limit then becomes one of regions do not stand out in cm or far-IR wavelength sur-
masssensitivity. EvenwithALMAandchoosingtheclos- veys. However, as discussed above, in all three scenarios
estpossibletargetsintheGalaxy,deepintegrationswillbe they must have a large gas mass in a small volume. As
requiredtoprobethegasexpectedtoformstarsacrossthe such, they should be easy to pick out as bright, compact
fullstellarmassrange. objectsatmm andsub-mmwavelengths. However,obser-
Understandingthegaskinematicsacrossarangeofden- vationallimitationshavemeantthatGalacticplanesurveys
sitiesandspatialscalesisnecessarytodistinguishbetween at these wavelengthshaveonly beenpossible overthe last
thedifferentformationscenariosofYMCs. The‘conveyor few years. Previous targeted surveys for young massive
belt’ model, for example, suggests that large amounts of proto-clustershave not found any starless gas clouds with
low- or moderate-density gas should be rapidly infalling. > 105M⊙ at parsec size-scales (e.g. Fau´ndezetal. 2004;
Given the new frontier in sensitivity being opened up by Hilletal. 2005; Simonetal. 2006; Rathborneetal. 2006;
ALMA,itisnotclearatthisstagewhatthebesttransitions Purcelletal.2006;PerettoandFuller2009).
for this purpose might be. However, studies of (less ex- However,thankstoaconcertedeffortfromtheobserva-
treme) massive and dense high-mass star forming regions tionalGalacticstarformationcommunityoverthelastfew
are pavingthe way (e.g. Perettoetal. 2013). Derivingthe years (see the review by Molinari et al., this volume), the
spatial and kinematic distribution of mass as a function data will soon be available to compile a complete list of
of size scale will likely require simultaneously observing YMCprogenitorcloudsintheMilkyWayneededtomake
many different transitions to solve for opacity, excitation definitivestatementsabouttherelativepopulationsofYMC
and chemistry variations. Extreme environments, like the progenitorcloudswith and withoutprodigiousstar forma-
Galactic Center, will prove especially challenging in this tionactivity. Todate,systematic,blind,large-areasearches
regard. for YMC progenitor clouds at all stages of the cluster
formation process have been published for two regions
2.3. YMCsandprogenitorcloudsintheMilkyWay of the Galaxy: the first quadrant (Ginsburgetal. 2012)
Extragalactic observations will be crucial to probe the and the inner 200pc (Longmoreetal. 2013a). In the near
formationofthemostmassiveYMCsinawiderangeofen- future, results from continuum surveys like ATLASGAL
vironments(e.g. galaxymergers). However,thediscussion (Schulleretal.2009;Contrerasetal.2013)combinedwith
in § 2.2.1 shows that for the foreseeable future the Milky spectralline studies (e.g. MALT90, CHAMP, ThrUMMS,
Way,andtoalesserextenttheLMCandSMC,aretheonly Mopra Southern Galactic Place CO survey – Fosteretal.
places in the Universewhere it will be possible to resolve 2011;Jacksonetal.2013;Barnesetal.2011;Burtonetal.
sitesofindividualformingprotostellar/planetarysystemsin 2013), will extend the search to the fourth quadrant. For
regionswhich have protostellar densities > 104M⊙pc−3. example, Urquhartetal. (2013) have already identified a
This means they are also the only places where it will be sample of YMC candidates with signs of active star for-
possibletodirectlytesttheeffectofextremeenvironments mationandContrerasetal. (inprep)haveidentifiedYMC
on individualprotostellar systems. This providesa strong candidatesat all evolutionarystages throughthe MALT90
motivationtoidentifyacompletesampleofYMCsandtheir survey. In the longer-term, HiGAL (Molinarietal. 2010)
progenitorcloudsintheGalaxy.Suchacatalogdoesnotyet will provide a sensitive, uniform survey across the whole
existduetothedifficultyinfindingcloudsatcertainstages Galaxy. However, our analysis relies on having complete
oftheformationprocess. samplesatallstagesoftheclusterformationprocess,sowe
On the one hand, it is straightforward to find all the focus on the extant surveysof the first quadrantand inner
clouds in the Galaxy with embedded stellar populations 200pcoftheGalaxybelow.
> 104M⊙. Their high bolometricluminosity(> 106L⊙)
2.3.1. ThefirstquadrantoftheGalaxy
and ionizing flux (Q> 1051/s) make them very bright
at far-IR wavelengths (where the spectral energy distribu- Ginsburgetal. (2012) used BGPS data (Aguirreetal.
tionpeaks)andcmwavelengths(whichtracesthefree-free 2011) to carry out a systematic search for YMC progen-
emission from the ionized gas at wavelengths where the itor clouds in the first Galactic quadrant, l = 6◦ − 90◦
cloudsand the restof the Galaxyare optically-thin). As a |b| < 0.5◦. This region is equivalent to ∼30% of the to-
result,thesesourceswithprodigiousembeddedstarforma- talGalacticsurfacearea,assuminganouterGalacticradius
tionhavebeenknownsincetheearlyGalacticplanesurveys of 15kpc. In this region Ginsburgetal. (2012) identified
atthesewavelengths(e.g.Westerhout1958),andmanysuch 18cloudswithmassMgas >104M⊙ andradiusr≤2.5pc.
objectshavebeenstudiedindetail(e.g.Plumeetal.1997; All of these clouds have gravitationalescape speeds com-
Sridharanetal. 2002; Beutheretal. 2002; Lumsdenetal. parable to or largerthan the sound-speedin photo-ionized
6
gas, so pass the Bressertetal. (2012a) criteria for YMC 2003). For that reason, it is crucial to only compare the
progenitor clouds. Crucially, all 18 of these clouds are demographics of stellar clusters in the first quadrant and
prodigiously forming stars. None of them are starless. GalacticCenterthatareyoungerthanthisage.
Ginsburgetal. (2012) use this to place an upper limit of Weconcludethatareliablemetrictoinvestigatethedif-
0.5Myrtothestarlessphaseforthecloudsintheirsample. ferentformationmechanismsis to comparethe numberof
Thisissimilartotheupperlimitonthelifetimesofclouds YMCsyoungerthanafewMyr(N ),tothenumberof
YMC
forming high-mass stars by Tackenbergetal. (2012). As- YMCprogenitorcloudswithprodigiousstarformationac-
sumingastarformationefficiencyof30%,only3ofthe18 tivity (Nactive), to the number of YMC progenitor clouds
cloud
identifiedcloudsaremassiveenoughtoformboundstellar withnodiscerniblestarformationactivity(NnoSF).Inother
cloud
clustersof104M⊙. words,theratioofNYMC :Naclcotuivde :NncloouSdF containsinfor-
mationabouttherelativelifetimeofthesethreestages.
2.3.2. Theinner200pcoftheGalaxy The inner 200pc of the Galaxy contains two YMCs –
Longmoreetal. (2013a) conducted a systematic search theArchesandQuintupletclusters(weexcludethenuclear
for likely YMC progenitor clouds in the inner 200pc of cluster as this most likely has a differentformationmode:
theGalaxybycombiningdustcontinuummapswithspec- seeGenzeletal.(2010b)forareview)–andtwoSFactive
trallinemapstracingmoleculargasathighvolumedensity. clouds, Sgr B2 and Sgr C. Combined with the four quies-
Basedonmapsoftheprojectedenclosedmassasafunction centcloudsfrom§2.3.2,theNYMC:Naclcotuivde:NncloouSdF ratio
ofradius,theyidentified6cloudsaspotentialYMCprogen- intheinner200pcoftheGalaxyisthen2:2:4.
itors. Intriguingly, despite having extremely high column Turning to the first quadrant, there is presently one
densities (up to ∼1024cm−2; 2×104M⊙pc−2) and being knownYMCinW49(AlvesandHomeier2003). Giventhe
opaqueupto70µm,fourofthesixpotentialYMCprogen- observationaldifficulties in finding unembeddedYMCs at
itorcandidatesshowalmostnosignsofstarformation.The large distancesthroughthe Galactic disk, othersmay well
upperlimittothefree-freeemissionfromsensitivecmcon- exist.Completenessisnotanissueforthetwoearlierstages
tinuumobservations,showsthatthereare,atmost,asmall (see§2.2.1). CombinedwiththenumberofSFactiveand
number of early B stars in these four clouds (Immeretal. quiescentcloudsfrom§2.3.1,theNYMC:Naclcotuivde:NncloouSdF
2012; Rodr´ıguezandZapata 2013). This is in stark con- ratioisthen1:3:0.
trast to the clouds of similar mass and density seen in the ComparingtheNYMC :Naclcotuivde :NncloouSdF ratiosbetween
diskoftheGalaxy,whichareallprodigiouslyformingstars the inner 200pc and first quadrantshows both regionsare
(see§2.3.1). producingasimilarnumberofYMCswithageslessthana
few Myr. However, there is a large disparity between the
2.3.3. Comparisonofthe1stquadrantandinner200pc number of progenitor clouds with/without star formation
Followingtheargumentsoutlinedin§2.2,ifthemolec- in the two regions. NncloouSdF = 0 for the first quadrant but
ular cloud population in a given region can be expected NncloouSdF = 4fortheinner200pc. Comparingtothepredic-
to produce the stellar populations in the same region, the tionsofthescenariosin§2.2, theGalacticCenterappears
cloudandstellardemographicscanbeusedto infersome- tobeformingYMCsinan“in-situ,slowformation”mode,
thingabouttheunderlyingformationmechanism. Wenow whereasthediskappearstobeconsistentwitha“conveyor
attempt this for the first quadrant and inner 200pc of the belt”or“fastin-situ”modeofformation.
Galaxy. Insummary,studyingthecurrently-availabledatainthe
The first step is testing whether the assumption of the GalaxysuggeststhatYMCsindifferentregionsaccumulate
observed gas clouds producing the observed stellar pop- their mass differently. The two regions studied contain a
ulations holds for these regions. The region observed by sizeable fraction of the gas in the Milky Way, so it seems
Ginsburgetal.(2012)covers30%ofthesurfaceareaofthe reasonable to conclude that this is representative of YMC
Galaxy (assuming a Galactic radiusof 15kpc). The inner formationasawholeintheGalaxy. Ofcourse,whensim-
fewhundredpcoftheGalaxycontainsroughly10%ofthe ilar data becomes available for the rest of the Milky Way
molecular gasin the Galaxy (see Pierce-Priceetal. 2000; – in particular the fourth quadrant which contains a large
Ferrie`reetal.2007;KalberlaandKerp2009;Molinarietal. fractionof the gas in the Galactic disk – it is importantto
2011,formassestimates). Ifthestarformationrateinthese testthisresult.
regions has remained constant over several star formation However, these Galactic regions only representa small
cycles,itseemsreasonabletoassumesuchlargegasreser- fractionofalltheenvironmentsintheUniverseknowntobe
voirswillproducestatisticallysimilarstellarpopulationsas formingYMCs. Clearlyitwouldbefoolhardyatthisstage
observedatthepresentday.However,onceastellarsystem to draw any general claims about YMC formation from a
has formed, the environmentin the Galactic Center is po- dataset sampling such a small fraction of the total num-
tentiallyalotmoredisruptivethaninthedisk.Indeed,even berofregionsformingYMCs. Futureobservationalstudies
dense clusters like YMCs are not expected to live longer comparingthegasandstellardemographicsacrossthefull
than(orbedetectableafter)afewMyrintheGalacticCen- rangeofenvironmentsare requiredtomakeanysuchgen-
ter(e.g.PortegiesZwartetal.2001,2002;KimandMorris eral, empirically-based statements about YMC formation.
In the upcomingALMA, JWST and ELT era, the datasets
7
neededtosolvethisproblemshouldbecomeavailable. beit an order of magnitude lower than predicted given the
Inthemeantime,wecanstillmakeprogressinageneral amountofdensegas),and(ii)atleasttwoYMCsarefound
understanding of the YMC formation process from what in the Galactic Center, means that some mechanism must
we learn in the Galaxy if we can understand two key as- be able to overcomeany potential suppression in star for-
pects: (i) if/how the underlying physical mechanism for mationinasmallfractionofthegas. Asthedetailsofthis
YMC formation in the Galaxy depends on the environ- mechanism are of potential interest in understanding why
ment, and, (ii) how those environmental conditions com- the YMC formation mode in the Galactic Center may be
paretootherYMC-formingenvironmentsacrosscosmolog- differentfrom the disk, in § 2.4.2 we examinethis further
icaltimescales. beforeturningin§2.4.3toYMCformationinthedisk.
2.4. TheroleoftheenvironmentforYMCformation 2.4.2. YMCformationintheGalacticCenter
Wenowinvestigatehowdifferencesintheenvironmental AglobalunderstandingofstarformationintheGalactic
conditionsmay be playinga rolein YMC formation. Fol- Centerishamperedbythedifficultyindeterminingthe3D
lowingfromthepreviousdiscussion,westartbycomparing distributionofthegasandstars. Buildingonearlierefforts
thegeneralpropertiesofthegasintheGalacticCenterand (e.g.Binneyetal.1991),Molinarietal.(2011)putforward
the disk, before focussing on the properties of individual a model that the “twisted ring” of dense molecular gas of
YMCprogenitorcloudsinthetworegions. projectedradius∼100pcthattheyidentifiedasverybright
sub-mm continuum emission in the HiGAL data, was on
2.4.1. ComparisonofgaspropertiesacrosstheMilkyWay
elliptical X2 orbits (i.e. orbits perpendicular to the long
The general properties of the gas in the disk and the axisofthestellarbar). Inthisscenario,thetwoprominent
center of the Galaxy are both well characterized, and are sitesofstarformationinthering–SgrB2andSgrC–lie
known to vary substantially from each other (for reviews at the location where the X2 orbits intersect with the X1
see Molinari et al this volume; MorrisandSerabyn 1996; orbits(i.e. gasstreamsfunneledalongtheleadingedgeof
Ferrie`reetal. 2007). In summary, the gas in the Galac- thestellarbarfromthedisktotheGalacticCenter). Inthis
tic Center lies at much higher column and volume den- picture,thecollisionofdensegascloudsmayleadtoYMC
sity (Longmoreetal. 2013b), has a much larger veloc- formation(seee.g.Stolteetal.2008).
ity dispersion at a given physical size (Shettyetal. 2012) Based on the observed mass distribution and kinemat-
and has a higher gas kinetic temperature (Aoetal. 2013; ics, Longmoreetal. (2013a) postulatedthat the gasin this
MillsandMorris 2013). The offset between the gas and ringmaybeaffectedbythevaryinggravitationalpotentialit
dust temperature (Molinarietal. 2011) in the Galactic experiences.Theyhypothesizedthattheneteffectofthein-
Center is thought to be either due to the orders of mag- teractionisacompressionofthegas,aidedbythegasdissi-
nitude larger cosmic ray flux than in the disk, or the patingthetidally-injectedenergythroughshocks.Ifthegas
widespread shocks observed throughout the gas (Aoetal. waspreviouslysittingclosetogravitationalstability,thead-
2013;Yusef-Zadehetal.2013). Thediskhasawell-known ditionalnetcompressionofthegasmightbeenoughforitto
metallicity gradient with galactocentric radius of −0.03 begincollapsingtoformstars.Ifthishypothesisprovescor-
to −0.07 dex kpc−1 (Balseretal. 2011). The metallicity rect,onecanusetheknowntimesincepericentrepassageto
in the Galactic Center is measured to vary within a fac- effectivelyfollowthephysicsshapingtheformationofthe
tor of two of the solar value (ShieldsandFerland 1994; mostmassivestellarclustersintheGalaxy,andbyinference
Najarroetal.2009). thenextgenerationofthemostmassivestarsintheGalaxy,
There is evidence that a combination of the envi- as a function of absolute time. Numericalsimulations by
ronmental factors and the global properties of the gas severaldifferentgroupsshowthatthisscenarioisplausible
leads to differences in how the star formation proceeds and can accurately reproduce the observed gas properties
between the two regimes. Given the large reservoir (Lucas&Bonnellinprep,Kruijssen,Dale,Longmoreetal
of dense gas in the Galactic Center, the present-day inprep). Thefactthatthegasinthisregionhasalreadyas-
star formation rate is at least an order of magnitude sembleditselfintocloudsof∼105M⊙ andradiusofafew
lower than that predicted by star formation relations pcbeforeanystarformationhasbegun,suggeststhatonce
where the star formation scales with the gas density (e.g. thegasbecomesgravitationallybound,itwillformayoung
Ladaetal. 2012; Krumholzetal. 2012a; Beutheretal. massivecluster.
2012; Longmoreetal. 2013b; Kauffmannetal. 2013). Theextremeinfrared-darkcloud,G0.253+0.016(M0.25,
Kruijssenetal. (2013) find that the currently low SFR in the“LimaBean”,the“Brick: Lisetal.1994;LisandMenten
the Galactic Center is consistent with an elevated critical 1998; Ballyetal. 2010; Longmoreetal. 2012) is the best
density for star formation due to the high turbulent pres- studied example of such a cloud. Despite containing
sure. Theyproposeaself-consistentcycleofstarformation ∼105M⊙ of gas in a radius of ∼3pc, the only signs of
intheGalacticCenter,inwhichtheplausiblestarformation potentialstarformationactivityareone22GHzH2Omaser
inhibitorsarecombined. However,thefactthat(i)thereis (Lisetal. 1994) and several compact radio sources at its
a non-zero star formation rate in the Galactic Center (al- periphery (Rodr´ıguezandZapata 2013), indicating that at
8
most a few early B-stars have formed. As expected, the cloudsinthediskwhereoneexpectstofindYMCprogen-
average gas column density is very high (> 1023cm−3; itorsarealso cloudswhicharepossiblyundergoingglobal
Molinarietal. 2011). Kauffmannetal. (2013) showed gravitationalcollapse.
that at 0.1pc scales, there are very few sub-regions with The clouds of gas mass > 105M⊙ in the disk are typ-
high column density contrast (corresponding to densities ically many tens to hundreds of parsecs in size. The av-
2×105cm−3) comparedto the ambient cloud. In the sce- erage volume and column density is therefore low (e.g. a
nario proposedby Longmoreetal. (2013a), this cloud has few 102cm−3, a few 1021cm−2), especially compared to
recentlypassedpericentrewiththesupermassiveblackhole similarmasscloudsintheGalacticCenter. However,these
atthecenteroftheGalaxyandisbeingtidallycompressed globalsize-scalesaremuchlargerthantheparsecscalesof
perpendicular to the orbit and stretched along the orbit. interestforYMCformation. The immediateconclusionis
Preliminarynumericalmodelingresultssuggestthediffuse thattheYMCsembeddedwithinthesecloudscantherefore
outer layers of the cloud may be removed in the process, only make up a small volume filling factor of the whole
leading to the large observed velocity dispersions and ex- cloud.
plainingtheobservedcloudmorphology(Kruijssen,Dale& The fact that no 105M , pc-scale, starless sub regions
⊙
Longmoreinprep.).Thebulkofthecloudmasscanremain have been found within GMCs in the disk suggests that
bound,eventhoughstandardvirialanalysiswouldsuggest the GMCs do not begin life with such dense subregions.
thecloudisgloballyunbound(e.g.Kauffmannetal.2013). However, we know of at least 18 dense, parsec scale sub
Thepredictionisthatthetidally-injectedenergyispresently regions of ≥ 104M⊙ that have prodigious star formation
supportingthecloudagainstcollapsebutasthecloudcon- (Ginsburgetal.2012).Bylearninghowthesecoupletothe
tinues on its orbit this energy will be dissipated through larger(10−100pcscale)clouditmaybepossibletounder-
shocks and the cloud will eventually collapse to form a standhowYMCsassembletheirmassinthedisk. Asmen-
YMC. Given the large difference between the observed tionedearlier,eachoftheregionsinthefirstquadrantcon-
dust temperature (Molinarietal. 2011) and gas tempera- tainingacandidateprogenitorcloudhasbeenwell-studied
ture(Guestenetal.1981;MillsandMorris2013),compli- and much is known about the gas properties and (embed-
cated chemistry, extreme excitation conditions, evidence ded)starformationactivity. Therefore,suchastudyisfea-
forwidespreadshocksandahighcosmicrayrate(Aoetal. sible. However,anin-depthreviewofthesedetailedstudies
2013;Yusef-Zadehetal.2013),detailedobservationalstud- isbeyondthescopeofthiswork. Instead,wefocusontwo
ies (Kendrewetal. (2013), Johnstonetal. (2013), Rath- regions: W49 andW43. The formeris the mostluminous
borneetalsub.) andnumericalmodeling(e.g.Clarketal. star forming region in the Galaxy and contains the most
2013, Kruijssen,Dale, Longmoreetal. inprep.,Lucas& massive and dense progenitor cloud in the Ginsburgetal.
Bonnellin prep.) are requiredto test thishypothesis. Un- (2012)sample.Assuch,thisisthemostlikelysiteoffuture
derstandingtheoriginandstarformationpotentialofthese YMC formation in the first quadrant. In terms of large-
extremeGalacticCentercloudspromisestobeanexciting scaleGalacticstructure,W49andmostotherYMCprogen-
avenueforstudyinthenextfewyears. itorcloudcandidatesarenotfoundatany‘special’placein
theGalaxy(otherthanpotentiallylyingwithinspiralarms).
2.4.3. YMCformationintheGalacticdisk
W43isthepossibleexceptiontotherule,andispostulated
To date, no clouds of gas mass ∼105M⊙ and radius to lie at the interface between the Scutum-Centaurus (or
∼1pcwithnosignsofstarformationhavebeenfoundout- Scutum-Crux)armandthestellarbar.
side the Galactic Center. So what were the initial con-
ditions for the YMCs that are known to have formed in W49: Lying at a distance of 11.11+0.79kpc from Earth
−0.69
the disk of the Milky Way? Clues to their origin can be (Zhangetal.2013),W49isthemostluminousstarforming
gleaned from the properties of the present-day molecular regionin the Galaxy (107.2L : Sieversetal. 1991, scaled
⊙
cloud population in the disk. The mass distribution fol- to the more accurate distance of Zhangetal. (2013)), em-
lows a power law, dN/dM ∝ M−γ, with 1.5 < γ < 1.8 beddedwithinoneofthemostmassivemolecularclouds(∼
(ElmegreenandFalgarone1996;Rosolowsky2005). Inor- 1.1×106M⊙withinaradiusof60pc;Galva´n-Madridetal.
der to produce 104M⊙ of stars, mass conservation means 2013, and references therein). The spatial extent of the
that YMCs must have formed within progenitor clouds of entire cloud is 120pc, but the prominent star formation
gas mass at least 104M⊙/ǫ ∼105M⊙. Clouds more mas- region, W49A, is confined to an inner radius of ∼10pc.
sive than this have virial ratios close to unity (albeit with W49A is comprised of three subregions – W49N, W49S
significantscatter)(Rosolowsky2007;Dobbsetal.2011b). and W49SW – each with radii of a few pc and separated
While there are many difficulties in using virial ratios to fromeachotherbylessthan10pc. Approximately20%of
unambiguously determine if an individual cloud is gravi- the mass (and practically all of the dense gas) lies within
tationallybound,it seemsreasonableto assumethatmany 0.1% of the volume (∼ 2 × 105M⊙ of gas within a ra-
of the most massive clouds are likely to be close to being diusof6pc:Nagyetal.2012;Galva´n-Madridetal.2013).
gravitationallybound. Thisplacesaninterestingconstraint W49Nisthemostactivelystarformingofthese,containing
– the fundamental gas reservoir limitation means that the bothaclusterofstars>4×104M⊙andawellknownring
9
ofH regions,allwithinaradiusofafewpc(Welchetal. 2.4.4. Summary:environmentmatters
II
1987;AlvesandHomeier 2003;HomeierandAlves2005).
Insummary,thediskandGalacticCenterareassembling
Ginsburgetal. (2012) identified both W49N and W49SW
gas into YMCs in different ways. In the Galactic Center,
aslikelyYMCprogenitorclouds. Historically,severalsce-
the mechanism is ‘in-situ slow formation’, where the gas
narios have been put forward to explain the interaction
is able to reach very high densities without forming stars.
of these dense clumps with the larger scale cloud, from
Something, possibly cloud-cloud collisions or tidal forces
globalgravitationalcollapsetocloudcollisionsandtrigger-
andthegasdissipatingenergythroughshocks,allowssome
ing (e.g. Welchetal. 1987; BuckleyandWard-Thompson
partsof thegasreservoirto collapseunderitsowngravity
1996). In the most recent, multi-scale dense gas sur-
toformaYMC.
vey,Galva´n-Madridetal.(2013)showthelargerscalegas Conversely,inthedisk,thelackofstarless105M⊙,r<
cloudis constructedof ahierarchicalnetworkof filaments
1pc gas clouds suggest YMCs either form in a ‘conveyor
that radially converge on to the densest YMC progenitor
belt’mode,wherestarsbeginformingasthemassisbeing
clouds,whichactasahubforthefilaments(reminiscentof
accumulatedtohighdensity,orthetimescaletoaccumulate
the “hub-filament” formation scenario described in Myers
thegastosuchhighdensitiesmustbemuchshorterthanthe
(2009) and observed towards the very luminous massive
star formationtimescale. In two of the mostfertileYMC-
star formation region G10.6 by Liuetal. (2012)). Based
formingregionsinthefirstquadrant(W49andW43),recent
onkinematicevidence,theyconcludetheregionasawhole
studies have shown evidence of large-scale gas flows and
isundergoinggravitationalcollapsewithlocalizedcollapse
gravitationalcollapsefeedingtheYMCprogenitorclouds.
ontotheYMCprogenitorclouds.
After YMCs have formed, the remains of their natal
clouds can also provide clues to the formation mecha-
W43 Located within the so-called ‘molecular ring’, the nism. Observations of the remaining gas associated with
region lying between Galactic longitudes of 29.5◦ and the formation of Westerlund 2 and NGC 3603 suggest
31.5◦ containsa particularlyhigh concentrationof molec- these YMCs formed at the interaction zones of cloud-
ular clouds, several well-known star formation complexes cloudcollisions(Furukawaetal.2009;Ohamaetal.2010;
such asthe mini-starburstW43-Main,and 4YMC precur- Fukuietal.2014).
sors (Motteetal. 2003; Ginsburgetal. 2012). The 13CO From the above evidence alone, it would be prema-
emissionintheregionshowscomplicatedvelocitystructure ture to claim large scale gas flows as a necessary con-
over60kms−1alongthelineofsight. NguyenLuongetal. dition to form YMCs in the disk. However, combined
(2011b)andCarlhoffetal.(2013)concludethatalmostall with the fact that the most massive gas clouds have virial
of this 13CO gas is associated with W43, and that the gas ratios closest to unity (Rosolowsky 2007; Dobbsetal.
within ∼ 20kms−1 of the 96kms−1 velocity component 2011b), and numerous numerical/observational studies of
is part of a single W43 molecular cloud complex. This other (generally less massive/dense) cluster forming re-
complexhasanequivalentdiameterof∼140pc,totalmass gions show evidence for large-scale gas flows feeding
of ∼ 7 ×106M⊙ and many subregions of high gas den- gas to proto-cluster scales (e.g. W3(OH), G34.3+0.2,
sity. Themeasureddistanceisconsistentwiththecomplex G10.6-0.4, SDC335.579-0.292, DR21, K3-50A, Serpens
lying at the meeting point of the Scutum-Crux arm at the South, GG035.39-00.33, G286.21+0.17, G20.08-0.14 N:
end of bar. NguyenLuongetal. (2011b) and Motte et al Ketoetal. 1987b,a; Liuetal. 2012; Perettoetal. 2013;
(sub.) arguethatthe largevelocitydispersionand compli- Csengerietal.2011;Hennemannetal.2012;Klaassenetal.
catedkinematicstructureindicatetheconvergencepointof 2013;Kirketal.2013;Henshawetal.2013;NguyenLuongetal.
high velocity gas streams. Three YMC progenitor clouds 2011a;Barnesetal.2010;Smithetal.2009;Galva´n-Madridetal.
lie within this longitude-velocityrange makingit a partic- 2009),oftenwith“hub-filament”morphology(Myers2009;
ularly fertile place for YMC formation. Two of the three Liuetal. 2012), suggests this is a fruitful area for further
progenitorcandidates–W43-MM1andW43-MM2–have investigation.
projected separations of less than a few pc, and are very If YMCs in the disk of the Milky Way form predomi-
likely to be associated. Nguyen-Lu’o’ngetal. (2013) con- nantly as a result of large-scale gas flows, one would not
cludethatthese haveformedvia collidingflows drivenby expect to see pc-scale regions of ∼ 105M⊙ with no star
gravity. Given that several red super giant (RSG) clusters formation. Rather, the initial conditionsof the next YMC
are found at a similar location and distance (Figeretal. generationmustbemassiveGMCswithlittle signsofcur-
2006;Daviesetal.2007),thisregionoftheGalaxyappears rentstarformation,andkinematicsignaturesofeitherlarge-
to have been forming dense clusters of > 104M⊙ for at scaleinfallorconvergingflows. Searchingfortheseclouds
least 20Myr. It also flags the other side of the bar as an isanotherinterestingavenueforfurtherinvestigation.
interestingplacetosearchforYMCsandYMCprogenitor SowhatcanwelearnaboutYMCformationmoregener-
clouds (indeed Daviesetal. 2012, have already identified allyfromthisanalysis? Onepotentialinterpretationofthe
oneYMCcandidatethere). differencebetweentheYMCprogenitorclouddemograph-
icsbetweenthefirstquadrantandinner200pcisevidence
for two ‘modes’ of YMC formation. However, there are
10