当前位置：文档库 › A ULX associated with a cloud collision in M99

A ULX associated with a cloud collision in M99

a r X i v :a s t r o -p h /0608648v 1 30 A u g 2006

Mon.Not.R.Astron.Soc.000,1–??(2006)Printed 5February 2008

(MN L A T E X style ?le v2.2)

A ULX associated with a cloud collision in M 99

Roberto Soria 1,2?and Diane Sonya Wong 3

Harvard-Smithsonian Center for Astrophysics,60Garden st,Cambridge,MA 02138,USA

2Mullard

Space Science Laboratory (UCL),Holmbury St Mary,Dorking,Surrey,RH56NT,UK

3Astronomy Department,601Campbell Hall,University of California at Berkeley,CA 94720-3411,USA

Received 11August 2006;accepted 24August 2006

ABSTRACT

The Sc galaxy M 99in the Virgo cluster has been strongly a?ected by tidal interactions and recent close encounters,responsible for an asymmetric spiral pattern and a high star formation rate.Our XMM-Newton study shows that the inner disk is dominated by hot plasma at kT ≈0.30keV,with a total X-ray luminosity ≈1041erg s ?1in the 0.3–12keV band.At the outskirts of the galaxy,away from the main star-forming regions,there is an ultraluminous X-ray source (ULX)with an X-ray luminosity ≈2×1040erg s ?1and a hard spectrum well ?tted by a power law of photon index Γ≈1.7.This source is close to the location where a massive H I cloud appears to be falling onto the M 99disk at a relative speed >100km s ?1.We suggest that there may be a direct physical link between fast cloud collisions and the formation of bright ULXs,which may be powered by accreting black holes with masses ～100M ⊙.External collisions may trigger large-scale dynamical collapses of protoclusters,leading to the formation of very massive ( 200M ⊙)stellar progenitors;we argue that such stars may later collapse into massive black holes if their metal abundance is su?ciently low.Key words:X-rays:galaxies —radio lines:galaxies —galaxies:individual (NGC 4254)—X-rays:binaries —black hole physics

1INTRODUCTION

The Sc galaxy M 99(NGC 4254),located a distance of about 17Mpc (Tully 1988),is the brightest spiral in the Virgo Cluster (M B =?20.8),with a gas mass ≈5×109M ⊙and a kinematic mass ≈1011M ⊙(Vollmer,Huchtmeier &van Driel 2005;Phookun,Vogel &Mundi 1993).It shows a pe-culiar spiral structure 1,with one arm less tightly wound and much brighter than the other two.Another unusual feature of M 99is the presence of a string of H I “blobs”to the south and west of the stellar disk,and a low-surface-density H I gas tail to the north-west;the total gas mass in those extra-disc structures is ≈2×108M ⊙(Phookun et al.1993).It is likely that there is a physical connection be-tween the unusual spiral structure and the disturbed gas distribution.In one scenario (Vollmer et al.2005),the spi-ral structure was a?ected by a close encounter with another Virgo Cluster galaxy ≈280Myr ago;the surrounding H I clouds are due to ongoing,face-on ram-pressure stripping,

E-mail:rsoria@https://www.wendangku.net/doc/c213557158.html,

?E-mail:dianew@https://www.wendangku.net/doc/c213557158.html,

1As a matter of historical curiosity,it was the second galaxy ever in which a spiral pattern was identi?ed (Rosse 1850),with the “Leviathan of Parsonstown”,a few months after a similar discovery for M 51.and are mostly moving away from the galaxy.An alterna-tive scenario (Phookun et al.1993)suggests that the H I clouds and tail are tidal debris of an entity that was at least partly disrupted in a close encounter with M 99;that debris is now infalling and merging with the galactic disk.Simu-lations show (Bekki,Koribalski &Kilborn 2005)that tidal debris could also have been stripped from the outer disk of M 99itself,by the Virgo Cluster potential as the galaxy crossed the central region of the cluster;part of the stripped gas would later fall back onto the same galaxy.In this sce-nario,heavy gas infall onto the M 99disk would be respon-sible for the lopsided spiral structure (Phookun et al.1993;Bournaud et al.2005).

Intriguingly,a large H I cloud (VIRGOHI 21),with a gas mass ≈2×108M ⊙but without any associated stars,located ≈120kpc to the north-west of M 99,has recently been dis-covered (Davies et al.2004).One interpretation (Minchin et al.2005,2006)is that it is an old,bound “dark galaxy”that has never formed stars because of its low hydrogen surface density.In this scenario,VIRGOHI 21would be dark-matter dominated,with a kinematic mass ≈1011M ⊙;if so,it would likely be the same galaxy responsible for the suspected close encounter with M 99and possible tidal gas stripping.Alter-natively,VIRGOHI 21could itself be simply another,larger piece of tidal debris (without a dark matter halo),stripped from the outer H I disk of M 99by the Virgo Cluster poten-

2006RAS

2R.Soria&D.S.Wong

tial or during a close encounter with another galaxy(Bekki et al.2005)

Gas-rich galaxies with recent tidal interactions gener-ally show particularly active star formation,and M99is no exception,with a rate≈10M⊙yr?1inferred from its Hαlu-minosity(Kennicutt et al.2003),or≈10–20M⊙yr?1from an optical/near-IR photometric study(Gonzales&Graham 1996).As a comparison,this is twice as high as in M82(Ken-nicutt et al.2003),and three times the current total rate in our Local Group(Hopkins,Irwin&Connolly2001).An-other phenomenon often associated with tidal interactions and high star formation is the presence of ultraluminous X-ray sources(ULXs)with apparent isotropic X-ray luminosi-ties～1039–1040erg s?1(for recent reviews,see King2006; Fabbiano&White2006;Colbert&Miller2004).They are interpreted as accreting binary systems,an order of mag-nitude more luminous than the Eddington limit for typi-cal stellar-mass black holes(BHs)in the Local Group.Two ULXs were detected by ROSAT/HRI within the D25region of M99,on1997June30(Colbert&Ptak2002;Liu&Breg-man2005).In this paper,we use XMM-Newton to have a better understanding of the X-ray properties of M99,and in particular of its brightest ULX.We compare radio and X-ray information to investigate the possible connection be-tween ULX formation,star formation and collisional events in this galaxy.Finally,we outline a possible general scenario for the nature of ULXs,to be tested by future studies.

2XMM-NEWTON STUDY

2.1Data analysis

M99was observed with all instruments on-board XMM-Newton on2003June26(Revolution651;observation ID 0147610101);the thin?lter,full-frame mode was used for the European Photon Imaging Camera(EPIC)detectors. We processed the Observation Data Files with standard tasks in the Science Analysis System(SAS),version6.5.0. Unfortunately,most of the X-ray observation was disrupted by strong solar?ares:after inspecting the background lightcurves,we retained a good live-time interval of17.0 ks for the pn and20.5ks for the MOSs,out of the total 51-ks exposure.In addition,Optical Monitor(OM)images were taken in the UV W1,UV M2and UV W2?lters,with exposure times of3.0,6.0and17.8ks,respectively2.We?l-tered the EPIC event?les,selecting only the best-calibrated events(pattern 12for the MOSs,pattern 4for pn),and rejecting?agged events.

We used the Chandra Interactive Analysis of Observa-tions(CIAO)task wavdetect to identify discrete sources in the combined EPIC image,in di?erent energy bands3.We selected a source extraction region of radius25′′around the brightest point-like source(the only one of the two ROSAT ULXs that is seen again by XMM-Newton,see Section2.2). We built response functions with the SAS task rmfgen,and auxiliary response functions with arfgen,for pn and the two 2See https://www.wendangku.net/doc/c213557158.html,/onlines/uhb/XMM

A ULX associated with a cloud collision in M993

01121845.58142643.11.71±0.352.46±0.500.85±0.20

02121851.13142630.42.09±0.383.01±0.551.0±0.2

03121851.36142424.53.02±0.514.35±0.731.5±0.3

04121852.61142547.82.44±0.383.52±0.551.2±0.2

05121856.15142418.039.5+2.8

?1.256.1+3.4

?1.7

19.4+1.2

?0.6

Parameter Value χ2ν0.96(81.0/84)

f obs

0.3?12(10?12CGS)0.39+0.02

?0.06

f em

0.3?12(10?12CGS)2.5+2.4

?1.5

L0.3?12(1040erg s?1)8.6+8.3

?5.2 (L po/L)0.3?12≈0.10

4R.Soria&D.S.

Wong

Figure1.XMM-Newton/EPIC X-ray maps of the inner part of M99.Left panel:combined EPIC image and contours for the soft band (0.2–1.5keV),dominated by emission from di?use hot plasma.Middle panel:EPIC image in the hard band(1.5–12keV),with soft-X-ray contours superimposed for comparison.Right panel:greyscale optical image in the Bessell-R band(archival data from the VLT/FORS1) with soft-X-ray contours overplotted.

4,left panel)is dominated by a bright nuclear star cluster

(a nuclear structure typical of late-type spiral galaxies),lo-

cated at the dynamical centre,as can be inferred from the

morphology of the dust?laments.The cluster is spatially

resolved in the HST image,with a full width half maxi-

mum of0.′′15±0.′′02corresponding to(12±2)pc;point

sources in the same?eld have a width of0.′′10±0.′′01.An

approximate conversion to standard colours gives an ab-

solute magnitude M V≈?13.5mag.Within200pc from

the nucleus,there are a few,fainter,unresolved stellar clus-

ters with typical absolute magnitudes M V≈?11mag.The

secondary optical peak described by Lampland(1921)and

Walker(1967)corresponds to the stellar complex labelled

as A in Figure4(unresolved from ground-based telescopes),

and sometimes(at least in plates with worse seeing)to the

unresolved blend of A and B.When we compare the HST

image with an XMM-Newton/OM image of the same?eld,

in the near-UV(UV W2?lter,with an e?ective wavelength

of2120?A),we note that the true nucleus is undetected:the

brightest UV sources in the nuclear region are the young

stellar complexes labelled as A,B and C(Figure4,right

panel).The optical nucleus remains undetected even in the

UV W1?lter(e?ective wavelength of2910?A).Unfortunately,

the XMM-Newton/EPIC image does not provide enough in-

formation to determine whether the X-ray emission peaks

at the true optical nucleus or at any of those three UV-

bright sources nearby.It is plausible that the true nucleus

and source A would appear of comparable brightness in the

standard U band.The nuclear cluster is the brighter source

in the B band,as evident from the archival images in the

NASA/IPAC Extragalactic Database5(NED).Of course,

this does not explain how the peak brightness could be per-

ceived to shift between the two sources several times over

a few decades,in plates with the same density,taken a few

days apart.Repeated supernovae or variability of individual

stars are clearly unsatisfactory explanations;variations in

the brightness of a nuclear BH would also be very peculiar.

The true explanation of this e?ect,if real,remains a mystery

for now.

https://www.wendangku.net/doc/c213557158.html,/

Figure3.Coadded XMM-Newton/EPIC spectrum and best-?t

residuals of the unresolved emission in the inner disk(within1′

from the nucleus).The spectrum has been?tted with a two-

temperature thermal plasma component plus a power law.See

Table2for the best-?tting parameters.

2.4A bright,hard ULX

The bright ULX located≈8kpc southeast of the nucleus

was the main objective of our X-ray study.Above1.5keV,

the ULX is brighter than the rest of the galaxy(Figure1).Its

background-subtracted X-ray spectrum(≈2100net EPIC

counts)is well?tted by a simple absorbed power-law(Fig-

ure5and Table3)of indexΓ=1.7±0.1.There is no hint

of a thermal component(blackbody or disk-blackbody)that

could be attributed to an accretion disk.Its emitted lumi-

nosity in the0.3–12keV band is1.9+0.2

?0.1

×1040erg s?1,which

is the same,within the errors,as the extrapolated ROSAT

luminosity observed in1997(Liu&Bregman2005).Inter-

estingly,the X-ray emission detected by Einstein/IPC(Fab-

biano,Kim&Trinchieri1992;archival X-ray image avail-

able in NED)is centred on the galactic disc and does not

show an enhancement at the ULX position,even allowing

for the lower spatial resolution and low number of counts

(140±16).As a further comparison,there is another X-ray

source clearly visible in the Einstein image,about8′west

of the galaxy(a background quasar).In the XMM-Newton

observation,its?ux in the Einstein energy band is less than

c 2006RAS,MNRAS000,1–??

A ULX associated with a cloud collision in M99

Figure4.Morphology of the nuclear region in the optical and near-UV.Left panel:HST/WFPC2image in the F606W?lter;right panel:the same region,seen by XMM-Newton/OM in the UV W2?lter.At the dynamical center,an old nuclear star cluster dominates the optical image but is undetected in the near UV.The labels identify smaller,younger clusters of OB stars,and are given simply to facilitate a comparison between the two images.

half of the ULX?ux;however,that background source was clearly stronger than the ULX in the Einstein image.We estimate that at the time of the Einstein observations(1980 June25),the ULX must have been at least a factor of5 fainter than in1997and2003.

We estimate that the XMM-Newton/EPIC astrometry is accurate to≈1.′′5.There are no other point-like X-ray sources in the?eld that can allow a better registration with optical positions.We searched for optical counterparts within the X-ray error circle in the XMM-Newton/OM im-ages and in several optical images from public archives but found none.The bright optical source suggested as a pos-sible counterpart by Colbert&Ptak(2002),based on the ROSAT/HRI position,is now clearly ruled out(Figure1, right panel).Its brightness and colours are consistent with a G0foreground star(Kharchenko2001).We took optical spectra of this source with the LRIS spectrograph on the Keck telescope,con?rming the spectral identi?cation and verifying that its radial velocity is consistent with a fore-ground star,ruling out the possibility that it belongs to https://www.wendangku.net/doc/c213557158.html,ing the best high-resolution optical images available in public archives(in particular,a30-s VLT/FORS1obser-vation in the Bessell-R band;a720-s B-band image from the 2.1-m telescope at Kitt Peak National Observatory;a200-s B-band image from the2.5-m Isaac Newton Telescope at La Palma),we conclude that there are no optical counterparts brighter than M B≈?9.5mag and M R≈?9mag.This is not a very stringent limit,and does not even rule out an old globular cluster.None the less,we can con?dently say that the ULX is right at the outer edge of the stellar disk(the limit beyond which the gas density is too low to collapse and form stars spontaneously),and that there are no super star-clusters or massive OB associations at that position or within≈800pc.Table 3.Best-?tting parameters to the combined XMM-Newton/EPIC spectrum of the ULX.The XSPEC model is tbabs Gal×tbabs×po.The quoted errors are the90%con?-dence limit and N H,Gal=2.7×1020cm?2(Dickey&Lockman 1990).

N H(1020cm?2)23.3+3.8

?3.4

Γ1.67+0.10

?0.10

K po(10?5)7.3+0.9

?0.8

3RADIO STUDY:A COLLIDING CLOUD?

If the optical data o?er no clue on the nature of the ULX, the radio data may instead suggest a complex,intriguing story.H I21-cm line observations of M99were carried out by Phookun et al.(1993),using the C and D arrays(exposure times of8hr and4hr,respectively)of the Very Large Array6 (VLA).The size of the synthesized beam was25.′′03×23.′′56. See Phookun et al.(1993)for details of the observations and data analysis.As expected,the H I disk is≈40%larger than the stellar disk,and extends slightly beyond the location of the ULX(Figure6).The total H I?ux from the galaxy corresponds to a gas mass≈5×109M⊙(Vollmer et al.2005;

6The VLA is a facility of the National Radio Astronomy Obser-vatory,which is operated by Associated Universities,Inc.,under cooperative agreement with the National Science Foundation.

c 2006RAS,MNRAS000,1–??

6R.Soria&D.S.Wong

Figure5.Coadded XMM-Newton/EPIC spectrum and best-?t residuals of the ULX spectrum,?tted with a power law of index Γ=1.7±0.1.See Table3for the best-?tting parameters. Phookun et al.1993),using a standard conversion between H I?ux and mass(e.g.,Rohlfs&Wilson2006).

Interestingly,the ULX is located close to where a large H I spur or cloud apparently joins the gas disk(Figure6). The gas mass of this cloud is≈107M⊙,considering only the fraction that visibly“sticks out”from the disk.More likely, its mass could a factor of two higher,considering that part of it overlaps with the disk(see also Fig.5in Phookun et al.1993).This cloud represents～10%of the gas mass out-side the galactic H I disk:either infalling onto the disk,or in the process of being shredded from it,according to alterna-tive models(recall Section1).It may be possible that the association is simply a coincidence:there is no de?nitive way to tell from the data available.It is none the less remarkable that,while there are various other gas clouds around the galaxy(references in Section1),the cloud near the ULX is the only one that overlaps and probably impacts the stellar disk;all the others are at larger radial distances,at the edge of the galactic H I disk.We estimate that only≈3%of the projected area of the stellar disk overlaps with this(or any other)external cloud.

The H I velocity map(Phookum et al.1993)tells a more dramatic picture(Figure7).The gas cloud is strongly blue-shifted with respect to the galactic disk,with a di?erence in the projected radial velocity>100km s?1,and is clearly overlapping with the disk.From the inferred radial veloci-ties,the cloud could be either in the process of being ejected (ram-pressure stripped)from the disk towards us,or,more likely in our opinion,it is tidal debris infalling onto the disk from behind.In the latter case,we have no elements to de-termine whether the cloud is actually hitting and merging with the disk.But is is interesting to note that the possible impact point is very close to the projected ULX position. We shall discuss in Section4whether there may be a di-rect physical connection between collisional processes and the formation of ULXs.

1 arcmin

Nucleus

ULX

Figure 6.Top panel:H I21-cm radio?ux,represented in greyscale over the galactic disk,and as contours for the outer (less dense)regions.Contour levels are on a square-root scale from0.02to0.3Jy km s?1beam?1.Bottom panel:the same H I radio?ux contours overplotted over a VLT/FORS1R-band image.

4DISCUSSION

4.1Hard state of ULXs

With an X-ray luminosity≈2×1040erg s?1,the ULX in M99is among the brightest of this mysterious class of sources:the cut-o?in the luminosity function is at≈3×1040erg s?1(Gilfanov,Grimm&Sunyaev2004;Swartz et al.2004).If the emission is isotropic and Eddington-limited, the mass of the accreting BH is 100M⊙.Apart from its extreme brightness,there is a notable,unexplained feature associated with this source and other bright ULXs:the X-ray spectrum is well?tted by a simple power-law with a rather hard photon index(Γ=1.7±0.1).In stellar-mass BH binaries,such spectra are characteristic of the low/hard state(e.g.,McClintock&Remillard2006),with typical X-

c 2006RAS,MNRAS000,1–??

A ULX associated with a cloud collision in M99

Figure7.Velocity map of the H I21-cm radio emission.Char-acteristic radial velocities(in km s?1)are also marked on the plot,at various places on the disk and the cloud.The gas cloud is consistent with tidal debris infalling onto the galactic disk from behind(with respect to our viewpoint),at a projected relative speed>100km s?1;it appears as though it is impacting or merging with the disk at or near the ULX position.

ray luminosities in the standard XMM-Newton or Chandra bands less than a few per cent of the Eddington luminosity. At higher mass accretion rates and luminosities,stellar-mass BH binaries are dominated by either an accretion disk com-ponent(thermal,or high/soft state)or by a steep power-law component(Γ 2.5).If ULXs follow the same spectral-state classi?cation as stellar-mass systems,the inferred spectral slope and luminosity of the source in M99would suggest an accreting BH with a mass～104M⊙.This would require so-far untested formation processes,such as primordial rem-nants or nuclei of accreted dwarf galaxies that have recently captured a donor star.Alternatively,ULXs may simply have a di?erent spectral behaviour;for example,they might be found in a steady high/hard state which is not known in Galactic X-ray binaries.For other examples of bright ULXs with relatively hard power-law-like spectra,see Winter et al.(2005),and Stobbart,Roberts&Wilms(2006);see also the discussion in Gon?c alves&Soria(2006)and references therein.

The underlying problem is how to explain the lack of evidence or the relative weakness of the accretion disk com-ponent in such bright sources.If the disk is faint because it is truncated(as in the standard low/hard state),we expect a low radiative e?ciency,as well as a dimensionless accretion rate much below Eddington;therefore,a very high accre-tion rate(in physical units)and an even higher BH mass are needed to explain the observed luminosity.If the disk extends to the innermost stable circular orbit,the e?ciency may be higher(standard e?ciency～0.1);however,most of the detected photons must have been reprocessed in a comp-tonizing medium,because we do not see the thermal disk spectrum directly.This requires that most of the gravita-tional energy is released in or transferred to the upscattering medium,for example via magnetic coupling(Kuncic&Bick-nell2004).Finally,especially for a pure power-law source, we cannot rule out that the X-ray emission comes from a relativistic jet pointing towards our line of sight(microb-lazar)(K¨o rding,Falcke&Marko?2002).However,physical and statistical arguments make the strong-beaming scenario rather implausible as a general explanation for the whole population of ULXs(Davis&Mushotzky2004).Direct evi-dence against beamed emission is provided in some cases by the energy requirements of a photoionized nebula surround-ing a ULX(Pakull&Mirioni2003).Relativistic beaming has also been ruled out,in one case,by the presence of quasi-periodic oscillations(M82X-1:Strohmayer&Mushotzky 2003),and in another case by the radio/X-ray?ux ratio (Holmberg IX X-1:Miller,Fabian&Miller2004).

4.2ULXs and galaxy collisions

Bright ULXs are preferentially associated with tidally inter-acting or collisional systems,or actively star-forming galax-ies,or both(e.g.,Swartz et al.2004for a population study). For example,many bright ULXs have been found in the Antennae(Zezas&Fabbiano2002),the Cartwheel(Gao et al.2003),the Mice(Read2003),NGC7714/15(Smith, Struck&Nowak2005),NGC4485/90(Roberts et al.2002) the M81/M82group(in M82,Holmberg II and Holmberg IX).

The simplest explanation is that galaxy collisions trig-ger or enhance star formation,which,in turn,increases the birth rate of high-mass X-ray binaries(hence,the normaliza-tion of the point-source X-ray luminosity function).This also increases the probability of forming very luminous sources, near the upper cut-o?.For example,it explains the large number of X-ray sources with luminosities 1039erg s?1in the Antennae(Zezas&Fabbiano2002).

While this is probably a correct argument,it cannot be the whole story,especially for the small sample of ULXs with luminosities 1040erg s?1.Some of them(e.g.,Holm-berg II X-1:Dewangan et al.2004;Holmberg IX X-1:Miller et al.2004)are in dwarf galaxies with relatively small star-formation rates.Others,such as the ULX in M99or the brightest ULXs in NGC7714(Soria&Motch2004)and NGC4559(Soria et al.2005),are in strongly star-forming galaxies but far away from the main star-forming regions. There are no ULXs in the inner disk of M99despite an SFR～10M⊙yr?1;instead,the bright ULX object of our study is at the outer edge of the stellar disk,where the star formation rate is and probably has always been orders of magnitude lower.And yet,the ULX does seem to be as-sociated with a collisional event between the disk and an infalling gas cloud(Section3).How can we make sense of this apparently contrasting evidence?Here,we try to specu-late one possible scenario for ULX formation,consistent with the sketchy observational evidence available so far.The fol-lowing statements are conjectures to be tested with further observational and theoretical modelling.

i)Most ULXs are not strongly beamed.They are powered by BHs with masses～20–200M⊙;the upper limit can easily be reduced to 100M⊙if mild anisotropy or mild super-Eddington emission are allowed.The donor is likely to be a Roche-lobe-?lling OB donor star,which can supply the required amount of accreting gas over its nuclear timescale

c 2006RAS,MNRAS000,1–??

8R.Soria&D.S.Wong

(Rappaport,Podsiadlowski&Pfahl2005;Copperwheat et al.2006).If so,ULXs represent the upper end of high-mass X-ray binaries,consistent with their preferential location in young stellar environments;the accreting BHs are a factor of a few more massive than Galactic systems,and require stellar progenitors with masses～50–400M⊙.

ii)A starburst or high SFR may favour but is not a su?-cient condition for the formation of bright ULXs;other con-ditions may be required.For example,low metal abundance is essential to allow the evolution of a very massive star into a massive BH(Pakull&Mirioni2003;Heger et al.2003).

A high SFR is not a necessary condition,either,as proved by the occasional presence of ULXs in gas-rich but low-SFR environments.Super star-clusters are also not a necessary condition for ULX formation.

iii)Galaxy-galaxy or cloud-galaxy collisions may induce the formation of bright ULXs,as well as trigger intense star formation,as separate,parallel consequences.This could ex-plain why the two phenomena(ULXs and starbursts)are often,but not always,associated with each other.For exam-ple,local collisional events in NGC4559(Soria et al.2005) and M99may be responsible for their respective ULXs in metal-poor regions at the very edge of their stellar disks.

The key issue we need to explain is why a collisional event would directly(i.e.,not simply through enhanced star formation)favour the formation of a very massive star, which may then become a ULX progenitor if other condi-tions(e.g.,low metal abundance)are also satis?ed.A possi-ble explanation is suggested by studies of the star-formation process in the Galactic protocluster NGC2264-C,in the Cone Nebula(Peretto,Andr′e&Belloche2006).In that case, a molecular clump with a mass≈1700M⊙is undergoing a large-scale collapse onto its central region,on a dynam-ical timescale(essentially in free fall).This results in the formation of a few massive protostars,one of which has al-ready reached a mass≈40M⊙but is still accreting at a rate ～10?3M⊙yr?1(Peretto et al.2006).It is plausible that two or three protostars would accrete a few100M⊙of gas and co-alesce even before ending their protostellar phase(≈3×105 yr).Based on this scenario,we speculate that the formation of a star with an initial mass of a few100M⊙may occur in a protocluster such as NGC2264-C,via massive global gas infall and mergers already during the protostellar phase(So-ria2006);this is perhaps a more common process than the runaway merger of O stars in the core of super star-clusters (Portegies Zwart&McMillan2002;Freitag,G¨u rkan&Rasio 2006).

It can be shown that an external shock or pressure wave can trigger the global,dynamical collapse of a molecular clump such as NGC2264-C;instead,an isolated clump in hydrostatic equilibrium would have fragmented into much smaller Jeans-mass protostars(for recent reviews,see for example Struck2005,Elmegreen2002,2004).The critical momentum required to trigger the collapse of a molecular clump is of order of the clump mass times its sound speed; moreover,the accretion rate onto the protostellar cores is proportional to the momentum imparted to the collapsing molecular clump(Motoyama&Yoshida2003).Supernova shocks are a possible trigger of dynamical collapses in nearby molecular clumps;a fast,colliding gas cloud may carry even more energy and momentum than a supernova.Thus,we speculate that collisional events(such as the cloud-disk col-lision at the outer edge of M99,or the possible dwarf galaxy-disk collision in NGC4559:Soria et al.2005)may lead to the formation of extremely massive protostars,via dynam-ical collapse and protostellar mergers on a timescale～105 yr.

The second step of the process to form a ULX would then be to ensure that the massive stellar progenitor re-tains most of its gas until core collapse.We speculate that this is where low metal abundance comes into play,because it strongly reduces the mass loss rate through line-driven stellar winds(e.g.,Eldridge&Vink2006,and references therein)and therefore leaves behind a bigger BH.This is why the(metal-rich)Pistol star in the Quintuplet Cluster (initial mass≈200–250M⊙:Figer et al.1998),the Wolf-Rayet stars in the Arches cluster,or the massive protostars in the NGC2264-C protocluster will never form massive BHs,and perhaps why there are no ULXs in that part of our Galaxy.Conversely,we expect metal abundance to be much lower at the outer edge of the disk in M99,perhaps made even lower by the infalling gas clouds(see,e.g.,Chen, Hou&Wang2003,and Andrievsky et al.2003,for recent studies of metal abundances as a function of galactocentric distance in the Milky Way;and MacArthur et al.2004for similar trends in other disk galaxies).

In summary,our suggested scenario di?ers from models based on the runaway coalescence of O stars in super star-clusters,because it does not require such massive systems. We speculate that the formation of a massive stellar pro-genitor may occur in smaller(～103–104M⊙)protoclusters, during their embedded phase,through infall and merger pro-cesses similar to what is currently observed in NGC2264-C. Protoclusters of that size are massive enough to produce O stars,but would by no means be classi?ed as super clusters; in fact,they may not even be massive enough to survive their embedded phase as bound systems(Kroupa&Boily 2002).

Our scenario also di?ers from ULX formation models based on Population-III stars,because it only requires low but not primordial abundances,and therefore it can work at any redshift.At zero metallicity,stars with initial masses ≈140–300M⊙may be disrupted by pair-instability super-novae that leave no remnants(Heger et al.2003).Instead, at Z～0.1Z⊙,stars in that mass range may become the progenitors of accreting BHs in ULXs,via direct collapse, precisely in the mass range required by the X-ray observa-tions.Further discussion of each of those speculations,and detailed comparisons with the observations,is beyond the scope of this paper.

5CONCLUSIONS

We studied the X-ray properties of M99,the most massive spiral galaxy in the Virgo Cluster.As expected from its high SFR(≈10M⊙yr?1),the X-ray emission is dominated by a soft thermal-plasma component.The total unabsorbed lu-minosity(not including a bright ULX)inside the D25ellipse is≈(1.2±0.2)×1041erg s?1in the0.3–12keV band;about 15%of this is due to resolved or unresolved discrete sources. The emission appears almost uniformly di?used across the inner disk(at 5kpc from the nucleus),and there is no

c 2006RAS,MNRAS000,1–??

A ULX associated with a cloud collision in M999

starburst core.The temperature of the hot gas,kT 0.30 keV,is also rather low,typical of disk emission rather than of a starburst environment.

We brie?y discussed the morphology of the nucleus in the optical and UV bands,showing the presence of a(redder) massive nuclear star cluster and a few smaller,much bluer clusters of young stars around it.The spatial resolution of XMM-Newton does not reveal whether there are faint point-like X-ray sources in the nuclear region,and if so,whether they are located in the old nuclear cluster or associated to the young stars around it.

The main goal of this paper was a study of the prop-erties and origin of a bright ULX at the outer edge of the stellar disk.Its unabsorbed luminosity of≈2×1040erg s?1in the0.3–12keV band,together with a pure power-law spectrum with photon indexΓ≈1.7,are di?cult to interpret in the framework of classical spectral states for stellar-mass BHs.Such high/hard states are not uncommon in ULXs but are not generally seen in Galactic BHs.It sug-gests that we are not seeing the accretion disk directly,and that most of the gravitational power is e?ciently transferred to a comptonizing region.

An intriguing new discovery is that there is a massive gas cloud(H I mass～107M⊙)seen in projection very close to the ULX position.From its radial velocity,we suggested that this cloud is falling onto the galactic disk(from be-hind)and perhaps impacting it near the ULX location.The cloud occupies≈3%of the projected area of the stellar disk; therefore,we cannot entirely rule out the possibility that it is a chance association.However,we explored the specula-tive idea that there is a direct connection between collisional events and ULX formation,and not simply an indirect e?ect due to an enhancement in star formation.We suggested a possible scenario of ULX formation that would be consistent with this interpretation.We argued that cloud collisions may trigger a large-scale dynamical collapse of molecular clumps, rather than Jeans-mass fragmentation;in turn,this may lead to the formation of very massive stellar progenitors in the cluster core.If the metal abundance is low(Z～0.1Z⊙,not primordial abundances),we speculate that the star might retain enough of its gas to directly-collapse into a BH mas-sive enough(up to～100M⊙)to explain the luminosity of the brightest ULXs.

This scenario has to be tested with further individual and population studies of ULXs.At the individual level,the suggestion that ULXs have masses up to an order of magni-tude higher than Galactic BHs(but not more)needs to be tested with more advanced X-ray spectral modelling,and a better understanding of the thermal disk and comptonized emission components in such systems.At the population level,we need a more systematic study of the spatial as-sociation between ULXs,low-metallicity environments and collisional events.Further work on this issue is currenly un-der way(D.Swartz et al.,in preparation). ACKNOWLEDGMENTS

We thank Manfred Pakull for discussions on the nature of ULXs and particularly his intuition about the role of metal abundance.We thank Mark Cropper for discussions on the e?ect of cloud and satellite collisions.We are also grateful to Alex Filippenko,Anabela Gon?c alves,Zdenka Kuncic,Dave Pooley and Kinwah Wu.RS acknowledges support from an OIF Marie Curie Fellowship.DSW is grateful for?nancial support under Chandra grant G05-6092A to D.Pooley and A.V.Filippenko.

REFERENCES

Andrievsky S.M.,et al.,2002,A&A,381,32

Arnaud K.A.,1996,in Jacoby G.H.,Barnes J.,eds,As-tronomical Data Analysis Software and Systems V,ASP Conference Series,Vol.101,p.17

Bekki K.,Koribalski B.S.,Kilborn V.A.,2005,MNRAS, 363,L21

Bournaud F.,Combes F.,Jog C.J.,Puerari I.,2005,A&A, 438,507

Chen L.,Hou J.L.,Wang J.J.,2003,AJ,125,1397 Colbert E.J.M.,Miller M.C.,2005,Tenth Marcel Gross-mann Meeting on General Relativity,Rio de Janeiro,July 20-26,2003.M.Novello,S.Perez-Berglia?a and R.Ru?ni, Eds.World Scienti?c,Singapore(astro-ph/0402677) Colbert E.J.M.,Ptak A.F.,2002,ApJS,143,25 Copperwheat C.M.,Cropper M.S.,Soria R.,Wu K.,2006, MNRAS,submitted

Davies J.,et al.,2004,MNRAS,349,922

Davis D.S.,Mushotzky R.F.,2004,ApJ,604,653 Dewangan G.C.,Miyaji T.,Gri?ths R.E.,Lehmann I., 2004,ApJ,608,L57

Dickey J.M.,Lockman F.J.,1990,ARA&A,28,215 Eldridge J.J.,Vink J.S.,2006,A&A,452,295 Elmegreen B.G.,2002,ApJ,577,206

Elmegreen B.G.,2004,in Lamers H.J.G.L.M., Smith L.J.,Nota A.,eds,Formation and Evolution of Massive Young Star Clusters,Astronomical Society of the Paci?c(PASP Conference series),Vol.322,277 (astro-ph/0405552)

Fabbiano G.,Kim D.-W.,Trinchieri G.,1992,ApJS,80, 531

Fabbiano G.,White N.E.,2006,in Lewin W.H.G.,van der Klis M.,eds,Compact Stellar X-ray Sources.Cambridge Univ.Press,Cambridge,p.475

Figer D.F.,Najarro F.,Morris M.,McLean I.S.,Geballe T.R.,Ghez A.M.,Langer N.,1998,ApJ,506,384 Freitag M.,G¨u rkan M.A.,Rasio F.A.,2006,MNRAS,368, 141

Gao Y.,Wang Q.D.,Appleton P.N.,Lucas R.A.,2003, ApJ,596,L171

Gilfanov M.,Grimm H.-J.,Sunyaev R.,2004,NuPhS,132, 369

Gon?c alves A.,Soria R.,2006,MNRAS,in press (astro-ph/0606273)

Gonzalez R.A.,Graham J.R.,1996,ApJ,460,651 Heger A.,Fryer C.L.,Woosley S.E.,Langer N.,Hartmann D.H.,2003,ApJ,591,288

Hopkins A.M.,Irwin M.J.,Connolly A.J.,2001,ApJ, 558,L31

Jenkins L.P.,Roberts T.P.,Ward M.J.,Zezas A.2005, MNRAS,357,109

Kennicutt R.C.,Jr,et al.,2003,PASP,115,928 Kharchenko N.V.,2001,Kinematics and Physics of Celes-tial Bodies,17,409

c 2006RAS,MNRAS000,1–??

10R.Soria&D.S.Wong

King A.R.,2006,in Lewin W.H.G.,van der Klis M.,eds, Compact Stellar X-ray Sources.Cambridge Univ.Press, Cambridge,p.507

K¨o rding E.,Falcke H.,Marko?S.,2002,A&A,382,L13 Kroupa P.,Boily C.M.,2002,MNRAS,336,1188 Kuncic Z.,Bicknell G.V.,2004,ApJ,616,669 Lampland C.O.,1921,PASP,33,L167

Liu J.-F.,Bregman J.N.,2005,ApJS,157,L59 MacArthur L.A.,Courteau S.,Bell E.,Holtzman J.A., 2004,ApJS,152,175

McClintock J.E.,Remillard R.A.,2006,in Lewin W.H.

G.,van der Klis M.,eds,Compact Stellar X-ray Sources. Cambridge Univ.Press,Cambridge,p.157

Miller J.M.,Fabian A.C.,Miller M.C.,2004,ApJ,607, 931

Minchin R.F.,et al.,2005,ApJ,622,L21

Minchin R. F.,et al.,2006,MNRAS,submitted (astro-ph/0508153)

Motoyama K.,Yoshida T.,2003,MNRAS,344,461

Ott J.,Walter F.,Brinks E.,2005,MNRAS,358,1453 Page M.J.,Davis S.W.,Salvi N.J.2003,MNRAS,343, 1241

Pakull M.W.,Mirioni L.,2003,proceedings of the sym-posium New Visions of the X-ray Universe in the XMM-Newton and Chandra Era,26-30November2001,ESTEC, The Netherlands(astro-ph/0202488)

Peretto N.,Andr′e P.,Belloche A.,2006,A&A,445,979 Phookun B.,Vogel S.N.,Mundi L.G.,1993,ApJ,418,113 Portegies Zwart S.F.,McMillan S.L.W.,2002,ApJ,576, 899

Rappaport S.A.,Podsiadlowski Ph.,Pfahl E.2005,MN-RAS,356,401

Read A.M.,2003,MNRAS,342,715

Roberts T.P.,Warwick R.S.,Ward M.J.,Murray S.S. 2002,MNRAS,337,677

Rohlfs K.,Wilson T.L.,2006,Tools of Radio Astronomy, Heidelberg(Springer-Verlag)

Rosse W.P.,1850,Phil.Trans.Royal Soc.London,140, 499

Smith B.J.,Struck C.,Nowak M.A.,2005,AJ,129,1350 Soria R.,2006,in Meurs E.J.A.,Fabbiano G.,eds,Proc. IAU Symp.230,Populations of High Energy Sources in Galaxies.Cambridge Univ.Press,Cambridge,p.473 Soria R.,Cropper M.,Pakull M.,Mushotzky R.,Wu K., 2005,MNRAS,356,12

Soria R.,Motch C.,2004,A&A,422,915

Stobbart A.-M.,Roberts T.P.,Wilms J.2006,MNRAS, 368,397

Strohmayer T.E.,Mushotzky R.F.,2003,ApJ,586,L61 Struck C.,2005,Astroph.Update Vol.2,in press (astro-ph/0511335)

Summers L.K.,Stevens I.R.,Strickland D.K.,Heckman T.M.,2003,MNRAS,342,690

Summers L.K.,Stevens I.R.,Strickland D.K.,Heckman T.M.,2004,MNRAS,351,1

Swartz D.A.,Ghosh K.K.,Tennant A.F.,Wu K.,2004, ApJS,154,519

Tully R. B.,1988,Nearby Galaxy Catalog,Cambridge (Camb.University Press)

Valtchanov I.,Pierre M.,Gastaud R.,2001,A&A,370,689 Vollmer B.,Huchtmeier W.,van Driel W.,2005,A&A,439, 921Walker M.F.,1967,PASP,79,593.

Winter L.M.,Mushotzky R.F.,Reynolds C.S.,2006,ApJ, submitted(astro-ph/0512480)

Zezas A.,Fabbiano G.,2002,ApJ,577,726

c 2006RAS,MNRAS000,1–??