Feedback from Winds and Supernovae in Massive Stellar Clusters - II. X-Ray Emission
aa r X i v : . [ a s t r o - ph . S R ] M a y Mon. Not. R. Astron. Soc. , 1–21 (2014) Printed 7 May 2018 (MN LaTEX style file v2.2)
Feedback from Winds and Supernovae in Massive StellarClusters. II: X-Ray Emission
H.Rogers and J.M.Pittard
School of Physics and Astronomy, The University of Leeds, Leeds, LS2 9JT
Received 17 February 2014. Accepted 31 March 2014
ABSTRACT
The X-ray emission from a simulated massive stellar cluster is investigated. The emis-sion is calculated from a 3D hydrodynamical model which incorporates the mechani-cal feedback from the stellar winds of 3 O-stars embedded in a giant molecular cloud(GMC) clump containing 3240 M ⊙ of molecular material within a 4 pc radius. Asimple prescription for the evolution of the stars is used, with the first supernovaexplosion at t = 4.4 Myrs. We find that the presence of the GMC clump causes short-lived attenuation effects on the X-ray emission of the cluster. However, once mostof the material has been ablated away by the winds the remaining dense clumps donot have a noticable effect on the attenuation compared with the assumed interstellarmedium (ISM) column. We determine the evolution of the cluster X-ray luminos-ity, L X , and spectra, and generate synthetic images. The intrinsic X-ray luminositydrops from nearly 10 ergs s − while the winds are ‘bottled up’, to a near constantvalue of 1.7 × ergs s − between t = 1–4 Myrs. L X reduces slightly during each star’sred supergiant (RSG) stage due to the depressurization of the hot gas. However, L X increases to ≈ ergs s − during each star’s Wolf-Rayet (WR) stage. The X-rayluminosity is enhanced by 2-3 orders of magnitude to ∼ ergs s − for at least 4600yrs after each supernova (SN) explosion, at which time the blast wave leaves the gridand the X-ray luminosity drops. The X-ray luminosity of our simulation is generallyconsiderably fainter than predicted from spherically-symmetric bubble models, due tothe leakage of hot gas material through gaps in the outer shell. This process reducesthe pressure within our simulation and thus the X-ray emission. However, the X-rayluminosities and temperatures which we obtain are comparable to similarly powerfulmassive young clusters. Key words: feedback – hydrodynamics – X-rays
Massive stars have a profound affect on their natal envi-ronment. They have strong ionizing radiation fields whichcreate HII regions, and their powerful winds sweep upsurrounding material creating wind-blown shells and cavi-ties. Their supernovae (SNe) chemically enrich the interstel-lar medium (ISM) and help to sustain turbulence withinit. Thus the presence of massive stars in stellar clustershas implications for future generations of star formation.The dispersal and destruction of molecular material bythe winds and ionzing radiation may inhibit further starformation within that region. Conversely, compression ofmaterial by winds and shocks may also trigger new starformation (Koenig et al. 2012) and new cluster formation(Beuther et al. 2008; Gray & Scannapieco 2011).Many high-mass star-forming regions are observed tocontain diffuse thermal X-ray emission, which requires high temperature plasma. It has long been recognized that thefast winds of individual massive stars create high pres-sure and high temperature bubbles (e.g. Dyson & de Vries1972; Castor et al. 1975; Weaver et al. 1977). In large clus-ters containing many early-type stars the individual stellarwinds may combine, collectively creating a so-called clus-ter wind (e.g. Chevalier & Clegg 1985; Cant´o et al. 2000;Stevens & Hartwell 2003).In many cases the observed diffuse emission frommassive-star forming regions (MSFRs) is relatively soft. Forinstance, in M17 and the Rosette nebula the characteris-tic temperature kT < c (cid:13) H.Rogers and J.M.Pittard cluster (Wang et al. 2006), and the Quintuplet cluster(Law & Yusef-Zadeh 2004) has kT > N H > × cm − , while clusters with cooler plasma have lessabsorption. Wolk et al. (2008) suggest that while the stel-lar winds are bottled up the shocked gas remains maximallyheated, but subsequent leakage and the resulting adiabaticcooling of the gas causes the gas temperature to drop.Unfortunately, past comparisons of X-ray observationswith theory have had mixed success. Many works on MSFRssimply compare the observed X-ray luminosity against themechanical wind power of the stars, or the thermal energy ofthe plasma against an estimate of the time-integrated energyinput of the winds (e.g. Townsley et al. 2003; Ezoe et al.2006b; G¨udel et al. 2008). The efficiency of the conversionof mechanical energy to radiation is then found to rangefrom 10 − to 0.1. This, and the estimated mass of the X-ray emitting gas, indicates that in many cases the winds arenot completely confined and that hot plasma must flow intothe wider environment. This conclusion is reinforced by thefact that the application of completely confined wind-blown-bubble models often leads to a significant overprediction ofthe X-ray luminosity (e.g. Rauw et al. 2002; Dunne et al.2003; Harper-Clark & Murray 2009).Other works have compared the X-ray luminosity andthe surface brightness profile of the diffuse emission to thepredictions of cluster-wind models. In their analysis of theArches and Quintuplet clusters, Wang et al. (2006) foundthat the radial intensity profiles of the diffuse emission weremore extended than theoretical predictions.Harper-Clark & Murray (2009) recently determinedthat the observed diffuse X-ray emission from the CarinaNebula was 60 times too faint compared to predictions fromthe Castor et al. (1975) model, and 10 times too luminous compared to the Chevalier & Clegg (1985) model. This ledHarper-Clark & Murray to develop a third model wherebydensity variations in the ISM surrounding the cluster causesgaps in the swept-up shell, through which some of the highpressure gas in the bubble interior can leak. Their newmodel predicts a lower pressure within the bubble than theCastor et al. model, as the wind material is not completelyconfined, and also a lower X-ray luminosity. Consequently,it is more consistent with observations. However, since thecovering fraction of the shell is a free parameter this modelsuffers from a lack of predictive power.In young MSFRs in which there has not yet been timefor any massive star to explode as a supernova, the diffuseX-ray emission must result from the action of stellar winds.However, in older clusters where some of the massive starshave exploded one still might not detect any signature ofa SN explosion because the effect of a SNR on the ther-mal properties of the hot cluster gas is likely to be rela-tively short-lived. This time scale is generally believed to be ∼ yr (e.g. Kavanagh et al. 2011). For this reason, moststudies of stellar clusters prefer a wind based explanationfor the diffuse X-ray emission, though Ezoe et al. (2009)favour a recent SN explosion in their study of the East-ern Tip of the Carina nebula. A distinction exists betweenindividual stellar clusters, and larger scale regions of starformation which create superbubbles where multiple cav-ity supernovae are believed to be responsible for the diffuseemission (such as those of 30 Doradus, e.g. Chu & Mac Low1990; Townsley et al. 2011).Given the challenges of interpreting such complex envi-ronments as MSFRs, and the highly idealized models of mosttheoretical and modelling work, in this paper the hydrody-namical models of stellar wind and supernova feedback in aninhomogeneous environment outlined in Rogers & Pittard(2013) (henceforth referred to as Paper I) are used as a basisto simulate the resulting X-ray emission from such regions.Of great interest are the X-ray luminosity and spectrum,and their temporal variation as the stars in the simulationcycle through various evolutionary stages, including mainsequence, red supergiant, Wolf-Rayet and supernova. In Sec-tion 2 the details of the model and the method of calculatingthe X-ray emission and absorption are discussed. The resultsare presented in Section 3. Comparisons to numerical modelsand observations are made in Sections 4 and 5 respectively.Section 6 summarises and concludes this work. The X-ray calculations in this paper are based on the 3Dhydrodynamical model described in Paper I. The simula-tions were performed using the hydrodynamical code AR-WEN, which uses a piecewise parabolic interpolation andcharacteristic tracing to obtain the time-averaged fluid vari-ables at each zone interface. An iterative Riemann solveris used to determine the time-averaged fluxes and solve theequations of hydrodynamics (see Paper I for more details).The simulations consist of three massive O stars which rep-resent the main sources of mechanical feedback in a mas-sive star forming region contained within an inhomogeneous c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission GMC clump of radius 4 pc and mass 3240 M ⊙ . The mediumsurrounding this clump is homogeneous, with a density of3.33 × − g cm − ( n H ≈ n e ≈ . − ) and a tempera-ture of 8000 K. The simulations were performed on a 512 grid with free outflow boundary conditions. The total sim-ulation volume covers a cubic region of ±
16 pc centered onthe GMC clump. The cluster wind is injected as thermalenergy within a radius of 6 cells (0.375 pc).The evolution of the three stars is treated simplisticallyas three distinct phases - the Main Sequence (MS), Red Su-pergiant (RSG) and Wolf-Rayet (WR) phases. The detailsof the stellar cluster are summarized in Table 1. At the endof the Wolf-Rayet phase the stars explode imparting 10 M ⊙ of material and 10 ergs of thermal energy into the environ-ment. The lifespans of the stars are designed in such a wayso that there are three distinct supernova explosions overthe course of the simulation.The simulation utilizes a temperature dependent aver-age particle mass, and molecular, atomic and ionized phasesare tracked separately. The net heating/cooling rate per unitvolume is parameterized as ˙ e = n Γ − n Λ, where n = ρ/ m H .Γ is the heating coefficient and is set at a constant valueof Γ = 10 − ergs s − . Λ is the cooling coefficient which isassumed to depend only on temperature. Cooling at lowtemperatures (T . K) is then adjusted to provide threethermally stable phases which correspond to the molecular(T ∼
10 K), atomic (T ∼
150 K) and ionized (T ∼ To calculate the X-ray emission the results from the hy-drodynamical model are read into a radiative transfer ray-tracing code, and the appropriate emission and absorptioncoefficients are calculated for each cell using the tempera-ture and density values. A synthetic image on the plane ofthe sky is then generated by solving the radiative transferequation along suitable lines of sight through the grid. So-lar abundances and collisional ionization equilibrium are as-sumed throughout this work. The X-ray emissivity is calcu-lated using the mekal emission code (Mewe et al. 1995, andreferences therein). The emissivity is stored in look-up ta-bles containing 200 logarithmic energy bins between 0.1 and10 keV, and 91 logarithmic temperature bins between 10 and 10 K. Line emission dominates the emissivity at tem-peratures below 10 K, with thermal bremsstrahlung domi-nating at higher temperatures. The present calculations alsohave an interstellar absorption column (N H = 10 cm − )added to them, and each model is assumed to be at a dis-tance of 1 kpc from an observer. Table 1.
Wind properties of the three stars in the cluster as theyevolve.Stellar MS stageMass ˙ M v ∞ Duration Mtm Energy( M ⊙ ) (M ⊙ yr − ) (km s − ) (Myr) (kg m s − ) (ergs)35 5.0 × − × ×
32 2.5 × − × ×
28 1.5 × − × × Stellar RSG stageMass ˙ M v ∞ Duration Mtm Energy( M ⊙ ) (M ⊙ yr − ) (km s − ) (Myr) (kg m s − ) (ergs)35,32,28 1.0 × −
50 0.1 1.0 × × Stellar WR stageMass ˙ M v ∞ Duration Mtm Energy( M ⊙ ) (M ⊙ yr − ) (km s − ) (Myr) (kg m s − ) (ergs)35,32,28 2.0 × − × × The energy bins are split into three energy bands whichrepresent the soft, medium and hard X-ray components ofthe spectra. The soft X-ray regime runs from 0.1–0.5 keV,the medium runs from 0.5–2.5 keV and the hard X-rays runfrom 2.5–10.0 keV. These bands will be referred to as BB1,BB2 and BB3 respectively throughout this paper.It should be noted that the individual stars are notresolved in the hydrodynamic simulations in Paper I, andtherefore there is no contribution to the X-ray emission fromthe cluster wind interacting with any natal material closeto the cluster (Parkin & Pittard 2010) or from intraclusterwind-wind interactions (Cant´o et al. 2000; Pittard & Parkin2010).
The X-ray lightcurve for the cluster throughout the simula-tion is shown in the top panel of Fig. 1. The initial observ-able luminosity of the cluster is L X ∼ × ergs s − . Overthe next 0.7 Myrs this luminosity decreases by a factor of10 to approximately L X ∼ × ergs s − , at which pointit remains fairly constant for the duration of the MS of thecluster. Initially the X-ray luminosity is high as the clus-ter wind blown bubble is confined within the GMC clump.However, as the wind blows out of the low density regions ofthe clump, hot gas escapes from the centre, as described bythe “leaky bubble” model of Harper-Clark & Murray (2009)and Paper I. The reduced pressure within the bubble causedby this leakage results in a lower X-ray luminosity.Fig. 2 shows simulated X-ray images of the cluster attime t = 0.13 Myrs, where extended bubbles to either sideindicate that some of the hot wind material is leaking fromthe GMC clump. However, it is clear that there is still par-tial confinement by the inhomogeneous GMC clump sincethe images are not spherically symmetric. At this time allthree stars are on the MS (see Table 1 for the stellar prop-erties). The left and middle panels show images of the soft c (cid:13) , 1–21 H.Rogers and J.M.Pittard
Figure 1.
The X-ray lightcurve for the cluster over the course ofthe simulation. a) Shows the total intrinsic luminosity producedby the cluster (solid red line) compared with the observable lu-minosity after attenuation (dotted green line). b) Shows the at-tenuated luminosity in all three energy bands defined in Sec 2.2.The solid red line shows the soft X-rays in BB1, the green dashedline shows the medium X-rays in BB2 and the blue dotted lineshows the hard X-rays in BB3. and medium energy X-rays produced at this time, whilst theright panel shows the hard X-rays. The emission is brightestat the centre in all three images, but particularly so in themedium and hard images. At this early time there is strongabsorption of the soft X-rays within the GMC clump, asrevealed by the low surface brightness of regions which aremore clearly emitting at higher energies (compare the leftand middle panels). This behaviour is not so prominent inthe medium energy X-ray image, although there is some ab-sorption occuring.The most striking feature in the images is the extendedemission to the top right of the cluster, which results fromthe hot gas that has already broken out of the clump inthis direction. It is also interesting to observe that the hotfluid adiabatically cools as it accelerates to supersonic speedsthrough the ‘nozzles’ from which it leaves the confiningclump. This is visible as a reduction in the X-ray surfacebrightness in the hard band. The surface brightness of thisgas increases at larger distances from the clump as it passesthrough a termination shock. At this point it runs up againstpreviously shocked gas which is inflating the bubble andsweeping up a shell of the ambient medium which surroundsthe GMC clump. The hardest X-rays are produced by thehot gas in the cluster centre, where the cluster wind is par-tially confined. The bottom panel of Fig. 1 shows that theluminosities in the BB1 and BB2 energy bands are almostequal during the period when all three stars are on the MS.However, at t < .
25 Myrs the luminositiy in the BB2 en-ergy band is dominant due to the greater attenuation of the lower energy X-rays by the remnant GMC clump. The hardX-ray luminosity in the BB3 energy band is about an or-der of magnitude lower than the luminosities in the soft andmedium energy bands throughout the MS-dominated phaseof the cluster evolution.At t = 0.13 Myrs, approximately 90% of the total (0.1-10 keV) intrinsic X-ray luminosity originates from the inner4 pc radius of the simulation, which is the original radius ofthe GMC clump containing the cluster. This is not unex-pected as the cluster wind is young and hot plasma ventsout of only a few open channels at this time. Since the GMCclump is mostly intact, significant attenuation of low energyX-rays occurs within the clump radius. This is reflected bythe fact that only ∼ ∼
100 (see Section 3.4 in Paper I). This entrainmentincreases the density of the flow and its emissivity, while alsocreating slow moving obstacles which faster moving parts ofthe flow shock against. The flow outside of the GMC clumpthus contains a multitude of shocks, and a wide range ofdensities and temperatures.X-ray images of the cluster at t = 2.53 Myrs are shownin Fig. 3. There is considerable diffuse emission in the softand medium X-ray bands (left and middle panels of Fig. 3).In contrast, the spatial extent of the hard X-rays is muchsmaller, and these instead primarily trace the stellar clus-ter and the hot, shocked gas immediately downstream ofthe reverse shock of the cluster wind. The gas responsiblefor this emission reduces in temperature as colder materialfrom the remains of the GMC clump mixes in with it, which c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Figure 2.
X-ray emission for the cluster at time t = 0.13 Myrs. Each panel has sides of 500 pixels and length 55.4 pc. [Left] shows softX-rays 0.1–0.5 keV, [Middle] shows medium X-rays at 0.5–2.5 keV and [Right] shows hard X-rays at 2.5–10.0 keV. The stellar cluster isat the centre of each panel.
Figure 3.
X-ray emission for the cluster at time t = 2.53 Myrs. Each panel has sides of 500 pixels and length 55.4 pc. [Left] shows softX-rays 0.1–0.5 keV, [Middle] shows medium X-rays at 0.5–2.5 keV and [Right] shows hard X-rays at 2.5–10.0 keV. limits the extent of the emission in this image. The extentof the diffuse emission reaches well beyond the original clus-ter radius, with more than 88% of the overall luminosityoriginating from outside that radius. Although the imagesfor the soft and medium regimes are similar in structure,with strong emission in the centre and a filamentary diffusestructure towards the edges, there is a higher intensity inthe BB2 image, and the stellar cluster is clearly discernableat the centre of the clump.
The attenuated X-ray luminosity is dependent on both theISM column density and the density and size of the GMCclump in which the cluster forms. However, as the molec-ular material in the clump is ablated by the winds it willhave less of an effect on the observable luminosity of thecluster. Fig. 4 gives an indication of the degree of attenua-tion caused by the ISM and by dense clump material. Thered solid line shows the intrinsic X-ray spectrum from thecluster, with the green dashed line showing the total atten-uated spectrum taking all absorption effects into account.As discussed previously, the vast majority of the absorptionoccurs at soft X-ray energies, with very little occuring aboveE = 1.0 keV. The blue dotted line shows the attenuation ef-fects caused only by absorption from the ISM. Close inspec- tion of Fig. 4 reveals that the ISM absorption has little effectabove 0.5 keV, whereas the circumcluster absorption affectsthe spectrum up to energies around 1 keV. Therefore it isclear that these two distinct absorption components affectthe spectrum in slightly different ways. It is consistent withthe hottest gas (and therefore the hardest emission) beingburied more deeply within the GMC clump. We note thatthe average column density from the centre of the clusterthrough the GMC clump to an observer at t = 0.06 Myrs is,at ≈ × cm − (see Fig.12 in Paper I), about 3 times theassumed ISM column.At t = 2.53 Myrs (bottom panel of Fig. 4) the two at-tenuated spectra are practically identical. This is becausethe X-ray emitting gas is no longer confined by the denseabsorbing material of the GMC clump as it is in the earlystages of the bubble’s expansion, but now suffuses throughthe entire volume of the simulation. This is again consistentwith the average column density from the centre of the clus-ter through the GMC clump to an observer at this time,which Fig.12 from Paper I shows to be about 10 . cm − ,or only about 4% of the assumed ISM column.Changing the viewing angle to the cluster at late timesleads to only very small differences ( ≈ c (cid:13) , 1–21 H.Rogers and J.M.Pittard
Figure 4.
X-ray spectra of the cluster at two times during theMS dominated phase. The red solid line shows the total intrinsicemission produced by the cluster. The green dashed line shows thetotal observable emission after all attenuation effects are consid-ered, whilst the blue dotted line shows the effect of just the ISMabsorption on the emission. a) Shows the spectra at t = 0.06 Myrs,when there is a notable difference between the ISM only and totalattenuated emission. b) Shows the spectra at t = 2.53 Myrs whenabsorption by dense material from the GMC clump has little ef-fect on the overall attenuated emission.. earlier times when the hot wind gas is still breaking out ofthe clump the difference in the attenuated luminosity as theviewing angle to the cluster changes is only around 15–20 %.This likely reflects the relatively low initial column densityof the clump. Larger variations can be expected from modelswith higher initial column densities. ⊙ Star
At t = 4.0 Myrs the most massive star evolves into a RSG.Its mass loss rate increases, and its wind velocity decreases.The averaged mass-loss rate and speed of the cluster windthen changes from 9 × − M ⊙ yr − and 2000 km s − to ≈ − M ⊙ yr − and 136 km s − . The cluster wind there-fore becomes slow and dense. The central cluster is nolonger a source of hard X-rays, and there is no replen-ishment of the highest temperature gas in the surround-ing environment as it flows away from the cluster throughthe remains of the porous GMC clump (see Fig.8 in Pa-per I). Together these changes lead to a substantial reduc-tion in the amount of hard X-rays being produced. In fact,the BB3 luminosity decreases 4 orders of magnitude fromL X ∼ × ergs s − just before the evolutionary transi-tion to L X ∼ × ergs s − by the end of the RSG phase(see Fig. 1). In contrast, the intrinisic luminosity briefly in- creases following this transition, due to an increase in lumi-nosity in the BB1 band.Whilst the reason for this is not completely understood,it is possible that the sudden drop in pressure during thetransition causes material stripped from the dense cloudsto mix more rapidly with the hotter gas. Overall however,because the soft X-rays suffer from attenuation from thedense RSG-enhanced wind material close to the centre ofthe cluster, the attenuated emission actually drops.The dense material deposited during the cluster wind’sfirst RSG phase is subsequently cleared from the simula-tion volume once the most massive star further evolves toits WR phase at t = 4.1 Myrs. The combined average speedof the cluster wind increases back to 2000 km s − while thecombined mass-loss rate becomes 2.04 × − M ⊙ yr − . Thevery high momentum that the cluster wind now has effi-ciently clears out the RSG dominated cluster wind fillingthe lower density channels and dramatically increases theablation rate of the remaining dense clouds. This causes in-creased emission in all three X-ray bands. The intrinsic lu-minosity increases by over 2 dex above that reached in theMS phase, with an initial peak that then declines quickly toa steady value. This phase is relatively short-lived, lastingonly 0.3 Myrs. The most massive star explodes as a supernova att = 4.4 Myrs, imparting 10 M ⊙ of ejecta and 10 ergs of ther-mal energy into the centre of the GMC clump. At this pointboth of the other stars remain in their MS phases.The X-ray lightcurve immediately following the super-nova explosion is shown in Fig. 5. The SN ejecta is highlyoverpressured and rapidly expands into the surroundingmedium. Although this approach leads to the desired re-sponse on the surrounding medium, in actual SN explosionsthe ejecta rapidly cools through adiabatic expansion, and isconsiderably cooler than the simulated ejecta at compara-ble times. Therefore the bright peak in the X-rays seen inFig. 5 immediately after the explosion should be ignored asit is an artifact of the utilized approach. The X-ray lumi-nosity of the hot ejecta drops rapidly from its peak as theejecta starts to expand and its density decreases. However,Fig. 5 shows that the rate of decline of the X-ray luminos-ity decreases, and a minimum is reached after which theX-ray luminosity increases again. This behaviour is causedby ejecta running into the remaining dense clouds near thecluster. The kinetic energy that this ejecta has acquired atthis time is then re-thermalized and subsequently radiatesmore strongly. The dashed line 900 yrs after the explosionindicates when the luminosity is dominated by the interac-tions of the ejectra with surroundng gas, and thus no longeraffected by the explosion setup.Synthetic X-ray images in all three energy bands duringthe explosion are shown in Figs. 6-8. When the star firstexplodes there is high intensity emission at the centre of thecluster, as seen in all three figures, which is a consequenceof the implementation of the explosion with thermal ratherthan kinetic energy. As the hot ejecta expands outwards theintensity at the centre decreases, although it is still muchhigher than the pre-explosion levels.Once the shockwave has expanded out far enough it be- c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Figure 6.
Synthetic X-ray image in the BB1 (0.1–0.5 keV) energy band during the first 4600 years after the most massive star explodes.The explosion occurs at t=4.4000 Myrs (top left panel). Absorption is visible from 100 years after the explosion. Bow shock emissiondominates from 900 years. gins to interact with the high density remains of the GMCclump. This is apparent from around t = 4.4005 Myrs on-wards. Prior to this, absorption by dense clumps projectedin front of the blast wave is visible (see for example the topmiddle and top right panels in Fig. 6). As the shockwavesweeps through the inhomogeneous environment successivebowshocks form around each dense cloud that it encoun-ters. Individual bowshocks can be identified during the first900 yrs after the explosion, but at later times these mergeto create a single, though highly structured, region of emis-sion with variable surface brightness. The simulated X-rayemission should be largely unaffected by the explosion setuponce the emission from the bow shocks becomes dominant,which as is apparent from the previous discussion of the lightcurve occurs approximately 900 yrs after the explosion.The X-ray image is broadly spherical overall, thoughthere is substantial curvature to the main shock front onlocal scales. Ejecta begins to leave the grid approximately4600 yrs after the explosion. By 2000 yrs after the explosionthe most intense emission is observed someway behind themain shock front and the remnant takes on a “shell-like”morphology. This is likely due to the fact that at the time ofthe SN explosion the densest clouds surrrounding the clus-ter tend to occur in a shell with inner and outer radii of ≈ c (cid:13) , 1–21 H.Rogers and J.M.Pittard
Figure 7.
Same as Fig. 6, but for the BB2 (0.5–2.5 keV) energy band.
Figure 8.
Same as Fig. 6, but for the BB3 (2.5–10.0 keV) energy band.c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Figure 5.
The X-ray light curve for the cluster at the time ofthe first SN explosion. The black dashed line indicates 900 yearsafter the explosion, at which point emission from interactions withthe surrounding clump material is dominant in all three energybands. a) Shows the total intrinsic luminosity produced by thecluster (solid red line) compared with the observable luminosityafter attenuation (green dashed line). b) Shows the attenuatedluminosity in all three of the broadband energy bands. The solidred line shows the soft X-rays in BB1, the green dashed line showsthe medium energy X-rays in BB2 and the blue dotted line showsthe hard X-rays in BB3.
Figs. 6– 8 further illuminate the filamentary emission andabsorption which occurs in the BB1 and BB2 energy bands,and the smoother emission which occurs in BB3. The vol-ume within the reverse shock surrounding the stellar clusteris devoid of any hot gas and is visible as a deficit of emissionin the central regions of Figs. 6 and 7. The highly structurednature of the reverse shock is directly visible in the top rowof panels in Fig. 8. It is the X-ray bright parts of the top leftpanel in Fig. 8 which first “light-up” as the ejecta expandsoutwards. Bowshocks around the closest dense clouds to thecluster are responsible in both instances.The time evolution of the attenuated spectra during theperiod of the first SN explosion is shown in Fig. 9. The solidred line is at a time just before the star explodes whilstthe light blue dot-dashed line is t = 4600 yrs after the explo-sion, which is the approximate time at which the shockwavebegins to leave the grid. The attenuated spectrum for thecluster at t = 4.4009 Myrs, when bowshock emission beginsto dominate, is shown as the purple dotted line in Fig. 9.The spectrum is roughly the same shape, albeit consider-ably more luminous, as that of the pre-SN cluster. However,whilst the intensity of the soft and medium X-ray energieschange little, by the time the ejecta reaches the edge of thegrid (light blue dot-dashed line) there is a considerable de-crease in the hard X-ray (E & F l u x ( e r g s s - k e V - ) Energy (keV)3.3999 Myrs4.4001 Myrs4.4009 Myrs4.4046 Myrs
Figure 9.
The time evolution of the attenuated X-ray spectra ofthe cluster as the most massive star undergoes a SN explosion.The solid red line is at a time just before the star explodes, thepurple dotted line is ∼
900 yrs after the explosion and the lightblue dot-dashed line is ∼ After the 35 M ⊙ star has exploded the remaining two starscontinue in their MS phases for a further 0.1 Myrs, at whichpoint the 32 M ⊙ star begins to follow the same evolution-ary path as its predecessor. At t = 4.5 Myrs it evolves to aRSG and at t = 4.6 Myrs it becomes a WR star. The X-raylightcurve shown in Fig. 1 shows a similar pattern to the pre-vious evolution, in that the X-ray luminosity decreases oncethe less powerful RSG wind contributes to the cluster wind,while it increases once the star becomes a WR. However,as the evolution of the 32 M ⊙ star occurs so shortly afterthe first supernova, the luminosity is still declining from theaftermath of that event, partially as a consequence of theblast wave leaving the grid, and so it is hard to distinguishthis from the natural decline during the RSG phase. In fact,the luminosity at this stage is comparable to that of the pre-vious RSG phase despite the loss of one wind source and thecommensurate reduction in the momentum and energy fluxof the cluster wind. The 32 M ⊙ star explodes at t = 4.9 Myrs,inparting a further 10 M ⊙ of ejecta and 10 ergs of energyinto the simulation. 0.1 Myrs after this explosion the final re-maining star begins the evolutionary sequence already per-formed by its brethren. The mass, volume and density of the X-ray emitting gas areshown in Table 2. The mass of gas with a temperature inexcess of 10 K increases until t ≈ . ⊙ . The mass then drops slightly and stays around3-4 M ⊙ during the remaining MS phase of the cluster wind.During this time the X-ray emitting volume is just over 50%of the total simulation volume, since the shocked clusterwind has spread throughout most of the grid. The averagetemperature of the X-ray emitting gas is 2.5 × K, whilstthe mass-weighted average is 2.3 × K, implying that mostof the X-ray emitting gas is closer to the 10 K mark. Afterthe first star evolves to the RSG branch the mass of mate-rial which is hot enough to produce X-rays decreases alongwith both the average and the mass-weighted average tem-perature of the gas. Once the star evolves further to the WRphase the mass of X-ray producing gas increases by a factorof 10 and the volume of the material at T > K almost c (cid:13) , 1–21 H.Rogers and J.M.Pittard M a ss pe r b i n ( M O • ) log T (K) 4.38 Myrs4.401 Myrs4.40 Myrs4.401 Myrs4.404 Myrs4.42 Myrs Figure 10.
Mass of X-ray emitting material above T = 10 Kat five times during the simulation. The red solid line is att = 4.38 Myrs, shortly before the most massive star explodes. Theblue short-dashed line is at t = 4.40 Myrs immediately after thesupernova explosion (this reflects the conditions used to simulatethe explosion) and the green long-dashed line is at t = 4.401 Myrs,1000 yrs after the explosion when the emission is dominated bybowshock interactions. The purple dotted line is at t = 4.404 Myrswhen the ejecta begins to leave the grid. The light blue dot-dashedline is at t = 4.42 Myrs, 20,000 yrs after the explosion. Each tem-perature bin is of width 0.1 dex. C u m u l a t i v e M a ss ( M O • ) log T (K) 4.38 Myrs4.40 Myrs4.401 Myrs4.404 Myrs4.42 Myrs Figure 11.
Shows the cumulative mass of X-ray emitting mate-rial above T = 10 K at five times throughout the simulation. Thered solid line is at t = 4.38 Myrs, shortly before the most massivestar explodes, the blue short-dashed line is at t = 4.40 Myrs im-mediately after the SN explosion. The green long-dashed line is att = 4.401 Myrs at which point bowshock emission becomes dom-inant, and the purple dotted line is at t = 4.404 Myrs, when SNejecta begins to leave the grid. The light blue dot-dashed line isat t = 4.42 Myrs, 20,000 yrs after the explosion. doubles. The average temperature of the X-ray emitting gasincreases to 2.8 × K and the mass-weighted average to3.2 × K.The mass distribution of the simulation before, duringand after the first supernova explosion is shown in Fig. 10,and the amount of material which is at each temperatureabove 10 K is shown in Fig. 11. The red line shows the tem-perature distribution of the 25 M ⊙ of material above 10 Kshortly before the explosion at t = 4.38 Myrs. There is vir-tually no gas at temperatures greater than 10 K (0.35 M ⊙ ,see the red line in Fig. 11), and the mass-weighted aver-age temperature is T av = 3 . × K. The SN occursat t = 4.40 Myrs, and its hot ejecta is visible as the short-dashed blue line in Figs. 10 and 11. As discussed previously,bowshock emission from the SN becomes dominant approx-imately 900 yrs after the explosion. The temperature distri- bution of the 51 M ⊙ of material above 10 K at this timeis shown by the green long-dashed line in Figs. 10 and 11.There is a small amount of material above 10 K ( ∼ ⊙ ),but the majority of the material is between 10 − . K,after which there is an obvious decrease in X-ray emit-ting material. The mass-weighted average temperature isT av = 5 . × K at this time.The SN ejecta begins to leave the grid at t = 4.404 Myrs,shown as the purple dotted line in Figs. 10 and 11. There isapproximately 218 M ⊙ of material above 10 K at this time(see Table 2), with ∼
10% (21 M ⊙ ) of that material above10 K. 20,000 yrs after the explosion there is again virtuallyno gas at T > K as the shock heated gas slowly cools(shown by the light blue dot-dashed line in Fig. 10). How-ever, there is approximately 5 times more material between10 − K than before the supernova, with a peak at aboutT = 10 K. Fig. 11 also reveals that the maximum temper-ature of gas at t = 4.42 Myrs is actually lower than that att = 4.38 Myrs, at T max = 10 . K and T max = 10 . K re-spectively.
Chu et al. (1995) derived an analytical expression for the X-ray emission from a Weaver et al. (1977) wind-blown bubble(WBB), in terms of various physical parameters which areobservable, such as the density and size of the bubble. Thepredicted X-ray luminosity in the 0.1–2.4 keV band is: L X = (cid:0) . × erg s − (cid:1) ξI ( τ ) L / n / t / (1)where ξ is the metallicity relative to the solar value,L is the mechanical luminosity of the stellar wind(s)in units of 10 ergs s − , n is the number density ofthe ambient medium in cm − and t is the age ofthe bubble in 10 yr. The above equation contains a di-mensionless temperature τ , where the dimensionless inte-gral I( τ ) = (125/33) - 5 τ / + (5/3) τ - (5/11) τ / and τ =0 . L − / n − / t / .At t = 0.3 Myrs the expected luminosity in the 0.1–2.4 keV energy band as predicted using Equation 1 is L X ≈ . × ergs s − using the average ambient density ofthe mostly intact GMC clump of n ≈
250 cm − . Thiscompares to the combined luminosity from our BB1 andBB2 energy bands, which at L X = 3 . × ergs s − isroughly 5000 times lower than the prediction from the stan-dard Weaver et al. (1977) bubble. Because the edge of thebubble expands off the grid at t ∼ . . Using anestimate of the density as n ≈ . − (which is just 50%greater than the low density medium which surrounds theGMC clump in the simulations), the predicted X-ray lumi-nosity from Equation 1 would be L X ≈ . × ergs s − .Although our simulation only “captures” a small proportion c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Table 2.
The mass, density and volume of the X-ray emitting gas, and the average and mass-weighted average temperature of that gasat various times throughout the simulation.Time Phase of Mass at Density at Volume at % of log[T av ] at Mass-weightedeach star T > K T > K T > K Volume at T > log[T av ] at(Myrs) (35,32,28 M ⊙ ) (M ⊙ ) ( × − g cm − ) ( × pc ) T > K (K) T > K (K)0.00 MS,MS,MS 0.00 0.00 0.00 0.00 % 0.00 0.000.13 MS,MS,MS 1.89 5.17 0.25 7.63 % 6.40 5.580.32 MS,MS,MS 6.75 3.22 1.43 43.53 % 6.40 5.440.63 MS,MS,MS 4.26 1.60 1.81 55.28 % 6.40 5.360.95 MS,MS,MS 3.25 1.20 1.83 55.92 % 6.40 5.361.95 MS,MS,MS 3.43 1.33 1.75 53.32 % 6.40 5.372.53 MS,MS,MS 3.84 1.41 1.85 56.46 % 6.40 5.373.61 MS,MS,MS 4.06 1.42 1.95 59.40 % 6.40 5.394.06 RSG,MS,MS 2.13 1.08 1.34 40.89 % 6.30 5.284.31 WR,MS,MS 24.87 6.64 2.55 77.82 % 6.45 5.514.38 WR,MS,MS 24.99 6.69 2.54 77.54 % 6.40 5.514.40 SN,MS,MS 37.24 9.87 2.57 78.43 % 6.80 6.364.4009 SN,MS,MS 51.29 13.64 2.56 78.09 % 6.95 5.774.404 SN,MS,MS 217.97 52.67 2.82 85.95 % 6.80 5.834.41 MS,MS 231.57 53.04 2.97 90.64 % 6.65 5.864.42 MS,MS 132.53 29.94 3.01 91.92 % 6.45 5.774.56 RSG,MS 1.69 2.51 0.46 14.04 % 6.20 5.294.78 WR,MS 28.53 9.07 2.14 65.31 % 6.40 5.524.90 SN,MS 37.61 10.66 2.40 73.24 % 7.00 5.864.94 MS 81.54 21.05 2.64 80.57 % 6.35 5.58 of the X-ray luminosity at this time since a lot of the hot gashas flowed through the grid boundaries, the estimate fromEquation 1 is approximately 2000 times larger than the lu-minosity from our simulations at this time. Since this factoris likely to be many times greater than the “true” X-ray lu-minosity from our simulation (i.e. the luminosity we wouldinfer if our grid were big enough to contain the expandingbubble), we conclude that Equation 1 consistently overesti-mates the X-ray luminosity produced from our simulationsby a large margin.Harper-Clark & Murray (2009) also provide an analyti-cal expression for the expected X-ray luminosity of a Weaverwind-blown-bubble: L X ∼ × ξ (cid:18) L w × erg s − (cid:19) (cid:18) pcr (cid:19) (cid:18) × KT (cid:19) · (cid:18) . × yr t (cid:19) (2)where they have assumed an X-ray cooling rate Λ X ≈ × − ξ erg s − cm and t is the age of the cluster/windsource. Applying this expression to the Carina nebulaoverestimates the observed luminosity by a factor of 10 (Harper-Clark & Murray 2009). In the following we applyEquation 2 to our simulated cluster at a time when thestars are on their MS phases and the mechanical luminosityof the cluster wind is L w = 1 . × ergs s − . The aver-age temperature of the X-ray emitting gas at t = 2.53 Myrsis 2.5 × K (see Table 2). We only capture a small partof the bubble volume at this time. However, we can beguided by what an observer may choose as the bubble ra-dius. If the ISM column to the cluster was substantiallyhigher than our assumed value of 10 cm − , the emissionbelow 2.5 keV may be almost completely absorbed, in which case Fig. 3 shows that only the harder emission might be de-tected. The radius that an observer might then infer for the“bubble” in the BB3 image in Fig. 3 could then be approx-imated to 6 pc. This leads to a predicted X-ray luminosityof L X ≈ ergs s − using Equation 2, and an overesti-mate of the intrinsic emission “captured” in our simulationby roughly 4 orders of magnitude. To bring the values fromEquation 2 in line with the simulated results would requirea bubble radius of 130 pc.Clearly some of the underlying assumptions made inthese equations are incompatible with the simulated re-sults. The two main assumptions in the Weaver et al. bubblemodel which differ from our simulations are that the energydeposited by the stellar winds is confined within the bubbleand that the surrounding ISM is homogeneous. As discussedin both Harper-Clark & Murray (2009) and Paper I, leak-age of hot gas from the bubble interior leads to a significantreduction in the pressure. This in turn reduces the X-ray lu-minosity so that it is well below that from a confined bubble(c.f. Harper-Clark & Murray 2009). It is interesting that thecalculated X-ray luminosity from our simulations is roughly3–4 dex lower than the predictions from the confined bub-ble model, which is of order of the same difference betweenobservations of real clusters and the confined bubble model. Chevalier & Clegg (1985) and Cant´o et al. (2000) derivedan analytical model describing the cluster wind flow thatresults from the multiple interactions of the stellar windsproduced by the stars of a dense cluster of massive stars.Rodr´ıguez-Gonz´alez et al. (2007) developed the work ofCant´o et al. (2000) to include a non-uniform stellar distri-bution. c (cid:13)000
The mass, density and volume of the X-ray emitting gas, and the average and mass-weighted average temperature of that gasat various times throughout the simulation.Time Phase of Mass at Density at Volume at % of log[T av ] at Mass-weightedeach star T > K T > K T > K Volume at T > log[T av ] at(Myrs) (35,32,28 M ⊙ ) (M ⊙ ) ( × − g cm − ) ( × pc ) T > K (K) T > K (K)0.00 MS,MS,MS 0.00 0.00 0.00 0.00 % 0.00 0.000.13 MS,MS,MS 1.89 5.17 0.25 7.63 % 6.40 5.580.32 MS,MS,MS 6.75 3.22 1.43 43.53 % 6.40 5.440.63 MS,MS,MS 4.26 1.60 1.81 55.28 % 6.40 5.360.95 MS,MS,MS 3.25 1.20 1.83 55.92 % 6.40 5.361.95 MS,MS,MS 3.43 1.33 1.75 53.32 % 6.40 5.372.53 MS,MS,MS 3.84 1.41 1.85 56.46 % 6.40 5.373.61 MS,MS,MS 4.06 1.42 1.95 59.40 % 6.40 5.394.06 RSG,MS,MS 2.13 1.08 1.34 40.89 % 6.30 5.284.31 WR,MS,MS 24.87 6.64 2.55 77.82 % 6.45 5.514.38 WR,MS,MS 24.99 6.69 2.54 77.54 % 6.40 5.514.40 SN,MS,MS 37.24 9.87 2.57 78.43 % 6.80 6.364.4009 SN,MS,MS 51.29 13.64 2.56 78.09 % 6.95 5.774.404 SN,MS,MS 217.97 52.67 2.82 85.95 % 6.80 5.834.41 MS,MS 231.57 53.04 2.97 90.64 % 6.65 5.864.42 MS,MS 132.53 29.94 3.01 91.92 % 6.45 5.774.56 RSG,MS 1.69 2.51 0.46 14.04 % 6.20 5.294.78 WR,MS 28.53 9.07 2.14 65.31 % 6.40 5.524.90 SN,MS 37.61 10.66 2.40 73.24 % 7.00 5.864.94 MS 81.54 21.05 2.64 80.57 % 6.35 5.58 of the X-ray luminosity at this time since a lot of the hot gashas flowed through the grid boundaries, the estimate fromEquation 1 is approximately 2000 times larger than the lu-minosity from our simulations at this time. Since this factoris likely to be many times greater than the “true” X-ray lu-minosity from our simulation (i.e. the luminosity we wouldinfer if our grid were big enough to contain the expandingbubble), we conclude that Equation 1 consistently overesti-mates the X-ray luminosity produced from our simulationsby a large margin.Harper-Clark & Murray (2009) also provide an analyti-cal expression for the expected X-ray luminosity of a Weaverwind-blown-bubble: L X ∼ × ξ (cid:18) L w × erg s − (cid:19) (cid:18) pcr (cid:19) (cid:18) × KT (cid:19) · (cid:18) . × yr t (cid:19) (2)where they have assumed an X-ray cooling rate Λ X ≈ × − ξ erg s − cm and t is the age of the cluster/windsource. Applying this expression to the Carina nebulaoverestimates the observed luminosity by a factor of 10 (Harper-Clark & Murray 2009). In the following we applyEquation 2 to our simulated cluster at a time when thestars are on their MS phases and the mechanical luminosityof the cluster wind is L w = 1 . × ergs s − . The aver-age temperature of the X-ray emitting gas at t = 2.53 Myrsis 2.5 × K (see Table 2). We only capture a small partof the bubble volume at this time. However, we can beguided by what an observer may choose as the bubble ra-dius. If the ISM column to the cluster was substantiallyhigher than our assumed value of 10 cm − , the emissionbelow 2.5 keV may be almost completely absorbed, in which case Fig. 3 shows that only the harder emission might be de-tected. The radius that an observer might then infer for the“bubble” in the BB3 image in Fig. 3 could then be approx-imated to 6 pc. This leads to a predicted X-ray luminosityof L X ≈ ergs s − using Equation 2, and an overesti-mate of the intrinsic emission “captured” in our simulationby roughly 4 orders of magnitude. To bring the values fromEquation 2 in line with the simulated results would requirea bubble radius of 130 pc.Clearly some of the underlying assumptions made inthese equations are incompatible with the simulated re-sults. The two main assumptions in the Weaver et al. bubblemodel which differ from our simulations are that the energydeposited by the stellar winds is confined within the bubbleand that the surrounding ISM is homogeneous. As discussedin both Harper-Clark & Murray (2009) and Paper I, leak-age of hot gas from the bubble interior leads to a significantreduction in the pressure. This in turn reduces the X-ray lu-minosity so that it is well below that from a confined bubble(c.f. Harper-Clark & Murray 2009). It is interesting that thecalculated X-ray luminosity from our simulations is roughly3–4 dex lower than the predictions from the confined bub-ble model, which is of order of the same difference betweenobservations of real clusters and the confined bubble model. Chevalier & Clegg (1985) and Cant´o et al. (2000) derivedan analytical model describing the cluster wind flow thatresults from the multiple interactions of the stellar windsproduced by the stars of a dense cluster of massive stars.Rodr´ıguez-Gonz´alez et al. (2007) developed the work ofCant´o et al. (2000) to include a non-uniform stellar distri-bution. c (cid:13)000 , 1–21 H.Rogers and J.M.Pittard F l u x ( e r g s s - k e V - ) Energy (keV)Canto IntrinsicCanto AttenuatedSimulated IntrinsicSimulated Attenuated
Figure 12.
A comparison between the intrinsic and attenuatedX-ray spectrum of our simulated cluster whilst all three stars areon the MS with the Cant´o et al. (2000) analytical cluster windmodel. The red solid line shows the intrinsic and the green long-dashed line shows the attenuated X-ray luminosity as calculatedfrom the Cant´o et al. model.
In order to compare these models with the simulationspresented in this work, the cluster mass-loss rate and theaverage velocity were set to ˙M cl = 9 × − M ⊙ yr − andv cl = 2000 km s − respectively, equivalent to the simulatedcluster at a time when all three of the stars are still onthe MS (see Table 1 for the individual stellar properties).The cluster radius in both of the analytical models was setto R cl = 0 .
04 pc, which is consistent with stellar clustersof comparable mass. The X-ray spectrum of these solutionsare shown in Fig. 12, along with the X-ray spectrum of oursimulated cluster at t = 2.53 Myrs.It is clear that the Cant´o et al. (2000) model producesX-ray luminosities significantly lower than from our simu-lated cluster. This is expected considering the lesser degreeof confinement of the stellar winds inherent in the clusterwind model (c.f. Harper-Clark & Murray 2009).
It is widely observed that there is a deficit of X-ray emission from stellar clusters compared with predic-tions based on the WBB model of Weaver et al. (1977)(Dorland et al. 1986; Dorland & Montmerle 1987; Oey 1996;Rauw et al. 2002; Dunne et al. 2003; Smith et al. 2005;Harper-Clark & Murray 2009). Many explanations of thiseffect have been suggested, for example lower stellar lu-minosities, mass or energy loss from the bubble, orhighly efficient mass loading which reduces the temper-ature of the cluster below X-ray temperatures. However,mass-loading may also produce higher X-ray luminosities(Stevens & Hartwell 2003). Section 4 demonstrated that theX-ray luminosities produced by the model in Paper I also ex-hibit a lower luminosity than predicted by WBB models. Acomparison will now be made between our model results andobservations of M17 and the Rosette Nebula. For a full lit-erature review of young massive stellar clusters from whichdiffuse X-ray emission has been detected see Table 3 andAppendix A.
M17 is a young blister HII region located on the northeastedge of one of the largest GMCs in the Galaxy, at an ap-proximate distance of 1.55 kpc. It is estimated to be only ∼ X = 2 × ergs s − . At t = 0.44 Myrs,the simulated cluster has an intrinsic 0.1–10 keV luminosity(BB1+BB2+BB3) of L X = 7 . × ergs s − , approx-imately 25 times lower than in M17. However, given thenumber and type of O-stars in M17, the difference in emis-sion from our simulated cluster and the observed X-ray lu-minosity from M17 can be considered to be perfectly accept-able. Townsley et al. (2003) estimate the mass of plasma at T ∼ K to be 0.3 M ⊙ . Whilst the simulated cluster has4.26 M ⊙ at t = 0.63 Myrs (see Table 2), this includes gas attemperatures greater than T > K. The amount of mate-rial at temperatures above 10 K in our simulated cluster isactually 0.31 M ⊙ at this time, which is remarkably similarto the observations of M17. For more information on M17see Appendix A. The Rosette Nebula is a blister HII region at the tip ofthe giant Rosette molecular cloud. It is estimated to be ∼ > K to be M X ∼ .
05 M ⊙ , which is again muchlower than in our simulated cluster at this time (3.43 M ⊙ ,see Table 2). However, the amount of X-ray emitting ma-terial above T > K in our simulated cluster is actually0.3 M ⊙ , and above T > K it is 0.02 M ⊙ . These valuesare a much better match to the observations of the RosetteNebula.The intrinsic 0.5–2 keV luminosity is ≈ × ergs s − ,with no significant emission detected above 2 keV. Att = 1.96 Myrs, our simulated cluster has an intrinsic 0.5–2.5 keV luminosity (BB2) of L X = 6 . × ergs s − .Given the higher mass-loss rate of the Rosette cluster( ˙ M ∗ = 2 . × − M ⊙ yr − , Stevens & Hartwell 2003) andthe higher number of O-stars present, this is a reasonablyclose match to the simulations. For more information on theRosette Nebula see Appendix A. c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Table 3.
The properties of young massive stellar clusters from which diffuse X-ray emission has been detected. The clusters are orderedroughly by age. The abbreviation “CF” in column 4 is short for champagne flow. Further details and references for values in this tablecan be found in Appendix A.Cluster/Region Age Distance X-ray Thermal/NT L X kT N H Name (Myr) (kpc) Morphology (erg s − ) (keV) (cm − )RCW 38 . × × Omega (M 17) ∼ × × Westerlund 2 . ± × × Rosette 2 1.55 CF T 7 × × Hourglass 1–2.5 1.3 CF T . . × × Arches 2–2.5 8 CF T (+NT) 3.8 × × NGC 2024 (Flame) 0.3–3 0.415 CF T 2 ×
11 0.21,3.3 × Orion (M 42) 3 0.49 CF T 5.5 × < × Quintuplet 3.5–4 8 CF T 3 × +4 . − . × NGC 3603 1–4 7 ± × × Westerlund 1 4–5 4–5 CF T (+NT?) 1.7–30 × × NGC 3576N 2.8 ± × × NGC 3576S 2.8 ± × × When the stars in the simulation explode they input 10 M ⊙ of material and 10 ergs of energy into the surroundings.As seen in Fig. 5 the results of this explosion should onlybe trusted after 900 yrs when the dominant source of X-rayemission is from interactions between the blast wave andthe surrounding clumpy medium. The ejecta begins to leavethe grid after 4600 years, at which point hot gas and itscorresponding emission begins to be lost. A comparison willnow be made between the model and observations of youngSNRs which are of age 900 < t < ∼ ∼
60 kpc(Wada et al. 2013). The progenitor is thought to have beena Wolf-Rayet star with a zero-age MS (ZAMS) mass of ∼
32 M ⊙ that underwent significant mass loss prior to ex-ploding as a Type Ib/c or IIL/b supernova. Gaetz et al.(2000) used Chandra to image the SNR, finding it to bealmost “textbook”, with a hotter outer ring surrounding acooler, denser inner ring which is likely the reverse-shockedstellar ejecta. The diameter of the SNR was estimated to be40” by Hughes (1988) (approximately 12 pc at 60 kpc dis-tance). More recently, Hughes et al. (2000) estimated theradius of the blast wave to be 6.4 pc, in good agreementwith their earlier work. Hughes et al. (2000) also estimatedan expansion age of ∼ − based on a radius of 6 pc and an age of 1000 yrs, although Flanagan et al. (2004)find from Doppler shifts that the majority of the bulk mat-ter is moving at a lower 1000 km s − . Hughes et al. (2000)also estimate the temperature in the postshock region to be0.4–1.0 keV. Gaetz et al. (2000) estimated the upper limit onX-ray emission of the central source to be < × ergs s − ,whilst Wada et al. (2013) estimate the 0.5–10 keV luminos-ity of the Be/NS binary to be ∼ . × ergs s − us-ing Suzaku data. This is only a factor of 2 lower thanour intrinsic 0.5–10 keV X-ray emission from the simulatedcluster, which is L X = 1 . × ergs s − and L X =1 . × ergs s − for 1000 and 2000 yrs after the first SNexplosion respectively. As the simulation assumes collisionalionization equilibrium (CIE), which is unlikely to be the casein 1E0102, it is not surprising that higher luminosities areobtained from the simulation. The clumpy environment inthe simulated cluster may also be partially responsible, if theenvironment of 1E0102 has lower density and/or is more ho-mogeneous. The expansion velocity of the simulated blast-wave is between ∼ − (see Table 4), whichalthough high is similar to the estimate of the blast wave byHughes et al. (2000). Also known as G292.0+1.8, this is a core-collapse SN withan estimated age of 2700 - 3200 years (Chevalier 2005;Winkler et al. 2009; Tanaka & Takahara 2013), and a dis-tance of 6 kpc (Gaensler & Wallace 2003). It is one of onlya handfull of O-rich SNRs known today (Park et al. 2007;Ghavamian et al. 2012). The X-ray emission from such O-rich SNRs is thought to arise from faster, non-radiativeshocks in lower density ejecta and interstellar gas. The cen-tral source is thought to be a pulsar wind nebula. The SNRhas a radius of approximately 15 pc assuming a distance of6.2 kpc (Gaensler & Wallace 2003).Gonzalez & Safi-Harb (2003) derived an average tem-perature for the SNR using two components - a high tem-perature plasma associated with the supernova blast waveand a low temperature plasma from the reverse shock. Thesetwo components were estimated to be 1.05 ± c (cid:13) , 1–21 H.Rogers and J.M.Pittard
Table 4.
Young SNRs from core-collapse SNe compared with the simulated results. The SNRs are ordered roughly by age. Referencescan be cound in the accompanying text in § x Diameter Age V exp
Temperature Distance Prog.Name Name (ergs s − ) (pc) (yr) (km s − ) (keV) (kpc) Mass (M ⊙ )1E0102.2-7219 ∼ . × ∼ ∼ > . ×
15 2700–3200 . < × ∼ × ∼
32 3700–4450 > > × ∼
18 1060 ∼ × ∼
27 2060 ∼ × ∼
29 2630 ∼ × ∼
30 3300 ∼ × ∼
31 3860 ∼ ± ⊙ , though Hughes & Singh(1994) estimated a lower mass of 20-25 M ⊙ . A more re-cent estimate of the temperature by Park et al. (2007) us-ing Chandra data found a highly non-uniform distributionof hot, X-ray emitting gas in the remnant ranging fromkT ∼ ∼ ⊙ star, whenthe average temperature of the X-ray emitting plasma is ≃ X = 7 . × ergs s − , but made noestimate for the entire remnant. Our simulated cluster ataround this time has a total 0.1–10 keV intrinsic luminosityof L X ∼ × ergs s − . N132D is one of the brightest SNRs in the Large Magel-lanic Cloud (LMC) and has an estimated age of 3150 years(Morse et al. 1995) and an inferred progenitor mass of 30-35 M ⊙ (Blair et al. 2000). With a diameter of 80”, the dis-tance to the SNR of approximately 55 kpc (Hughes 1987) im-plies a real diameter of ∼
21 pc. This is a similar estimate tothe extent of the X-ray shell, which has an estimated radiusof 12 pc (Morse et al. 1995). The expansion velocity of theSNR has been estimated by several authors (e.g. Morse et al.1995; Hwang et al. 1992), with values ranging from 2250–3700 km s − . In comparison, the radius of our simulated SNRat around 3300 yrs after the explosion is ∼
15 pc, with aninferred expansion velocity of ∼ − .The X-ray luminosity in the 0.2–4 keV energy band wasestimated by Hughes (1987) to be 4.5–7.5 × ergs s − ,based on thermal plasma temperatures of 10 . − . Kand a hydrogen column density of 10 − . cm − (Raymond & Smith 1977). The estimated luminosity is ac-tually higher than the simulated results (L X = 1 . × ergs s − at 3300 yrs), despite the fact that the simula-tions assume CIE. However, given the inherrent differences in the simulated and actual environments these results canbe considered a reasonable match. The plasma temperatureis very similar to that found in the other SNRs mentioned,at approximately 0.6-0.7 keV, compared with an average of0.54 keV from our simulated remnant (see Table 2). Thereis patchy X-ray absorption around the remnant thought tobe caused by gas just outside the molecular cloud towardsthe nothern tip of N132D (Kim et al. 2003). Puppis A is a nearby Galactic SNR and has age estimatesranging from 3700 yr (Winkler & Kirshner 1985) to 4450 yr(Becker et al. 2012), making it most comparable to thesimualation 3900 yrs after the explosion (see Table 4). Adistance of 2.2 kpc has been estimated based on HI and COstudies (Reynoso et al. 2003), although a closer distance of1.3 kpc has also been proposed by Woermann et al. (2000)based on OH line detections. This remnant is embedded ina complex region composed of large atomic and molecularclouds and an interstellar density gradient. The remnant isabout 50’ in diameter (approximately 32 pc at a distanceof 2.2 kpc). A progenitor mass of 25 M ⊙ was inferred byCanizares & Winkler (1981).Optical knots detected from Puppis A are evident onlyin the northeast, implying the ejection of the matter dur-ing the explosion was asymmetric (Katsuda et al. 2008).Oxygen-rich filaments are detected to have radial velocitieshigher than ∼ − . These filaments are interpretedas SN ejecta which have remained mostly uncontaminatedby the ISM (Winkler & Kirshner 1985).Recently, Dubner et al. (2013) studied Puppis A us-ing Chandra and XMM-Newton. They estimated the X-ray luminosity between 0.3 and 8.0 keV to be L X = 1 . × ergs s − assuming a distance of 2.2 kpc. The X-ray emis-sion from Puppis A appears to be dominated by the swept-up ISM due to very low metal abundances (Hwang et al.2008). The total intrinsic X-ray luminosity of our simu-lated remant 3860 yrs after the explosion is L X = 1 . × ergs s − , which is a reasonable match to Puppis A.The average temperature in the remnant is 0.6 keV, verysimilar to the average temperature of 0.54 keV seen in the c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission simulated remnant at 3860 years after the explosion (see Ta-ble 2). This paper investigates the X-ray emission from a massiveyoung stellar cluster embedded in an inhomogeneous GMCclump treating only the mechanical effects from winds andsupernovae. The hydrodynamical input model was previ-ously simulated in Paper I, and this work explores the emis-sion arising from that model. Initially the dense parts of theclump decrease the observed X-ray emission due to attenua-tion, but once the cluster wind has destroyed and ablated alarge portion of this material the attenuation from the ISMmaterial is dominant.At very early times, when the wind material is stillconfined by the inhomogeneous GMC material the X-rayluminosity is reasonably bright, at L X ≈ × ergs s − .However, as the cluster wind errodes and destroys the sur-rounding clump it is no longer completely confined andtherefore hot gas is able to leak through the gaps in theshell. This causes a reduction in the X-ray luminosity asthe pressure within the bubble decreases. Once the low den-sity gas from the clump has been ablated away the coveringfactor of the cluster remains more or less constant, lead-ing to an approximately constant intrinsic X-ray luminosityof 1.7 × ergs s − and an attenuated X-ray luminosity of7 × ergs s − .The most massive star becomes a RSG at t = 4.0 Myrs,resulting in a large drop in the X-ray luminosity in all threeof the X-ray broadbands studied. The most dramatic de-crease is seen in the BB3 (2.5–10.0 keV) emission, where theattenuated X-ray luminosity drops four orders of magnitude,from L X ∼ × ergs s − to L X ∼ × ergs s − bythe end of the RSG phase. The drop in X-ray luminosity inthe other two broadbands over this period is around a factorof 50 and 100 for BB1 (0.1–0.5 keV) and BB2 (0/5–2.5 keV)respectively. Although a lot of material is deposited in theRSG-enhanced cluster wind, the amount of material at X-ray emitting temperatures is very low, contributing to thelack of X-ray emission observed at this time.100,000 years later the most massive star further evolvesto become a WR star, causing a dramatic increase in X-rayemission in all three broadband regions studied. The amountof material at a temperature greater than 10 K increasesby an order of magnitude over that seen in the RSG stage,and a total of 78% of the computational volume containsX-ray emitting material. The high momentum cluster windsweeps up the slower moving material deposited in the pre-vious phase and heats it to high temperatures, with the av-erage temperature of hot gas (
T > K) at this time beingaround T= 2 . × K. The total attenuated X-ray emissionincreases to L X ∼ × ergs s − , which is about 30 timesgreater than that observed when all three stars were on theMS. At t = 4.4 Myrs the most massive star in our simulationexplodes as a SN, ejecting 10 M ⊙ of material and 10 ergsof energy into the simulation. Due to the way in which theexplosion is initialised, the emission from the SNR only be-comes independent of the initial conditions of the explosiononce the interaction of the blastwave with the surrounding material becomes dominant. In this work this occurs approx-imately 900 years after the explosion. The ejecta begins toleave the grid 4600 years after the explosion, and thereforethe emission from the SNR can be compared with obser-vations of SNRs only between the ages of 900 < t < ⊙ star was compared withfour young core-collapse SNe with ages ranging from ∼ ACKNOWLEDGEMENTS
HR acknowledges a Henry Ellison Scholarship from The Uni-versity of Leeds and JMP acknowledges funding from theRoyal Society for a University Research Fellowship and fromthe STFC. We would also like to thank the referee for theirtimely and useful comments.
APPENDIX A: NOTES ON INDIVIDUALSTELLAR CLUSTERSA1 RCW 38
RCW 38 is a very young ( < A V ∼ D = 1 . − . × .
75 pc) X-ray plasma which ispredominantly non-thermal (Wolk et al. 2002). The power-law index of the emission steepens toward the cluster core. c (cid:13) , 1–21 H.Rogers and J.M.Pittard
Contamination of the diffuse emission by unresolved point-sources is not significant at distances of more than 0.15 pc( ∼
15 arcseconds) from the cluster center, though may beresponsible for the more thermal nature of the diffuse emis-sion measured in the core (Wolk et al. 2006). The cause ofthe non-thermal emission remains unclear.The diffuse emission is strongest in the central core nearIRS 2, and is confined on the southeast along a ridge. Re-cent
Spitzer observations reveal that winds from IRS 2 havecaused outflows towards the northeast, northwest and south-west of the central cluster (see e.g. Fig. 4 in Winston et al.2012).An excellent review of this cluster is given in the Hand-book of Star Formation, where the luminosity of the diffuseX-ray emission is given as about 3 × ergs s − (Wolk et al.2008). A2 The Omega Nebula (M17)
M17 is a very young blister H II region located onthe northeast edge of one of the largest giant molecularclouds in the Galaxy, with an extent of 4 ◦ ( ∼
110 pc,Elmegreen et al. 1979). The geometry of M17 is thoughtto resemble the Orion Nebula HII region except that itis seen edge-on rather than face-on (Meixner et al. 1992;Pellegrini et al. 2007). M17 is photoionized by the mas-sive stellar cluster NGC 6618, which has 14 identified Ostars (Broos et al. 2007), and is estimated to be ∼ . A V = 3 −
15 with an average of 8to the OB stars, though some parts of the cluster have A V >
20, Hanson et al. 1997). The earliest O stars are anO4+O4 visual binary known as Kleinmann’s AnonymousStar (Kleinmann 1973), which may in fact be a pair of collid-ing wind binaries (Broos et al. 2007; Hoffmeister et al. 2008;Rodr´ıguez et al. 2012). Evidence for an older (2 − kT = 0 . kT = 0 .
29 and kT = 0 .
57 keV,with the highest temperature component providing 56% ofthe intrinsic luminosity (Townsley et al. 2011). The absorp-tion to each of these emission components increases withthe temperature of the component, so that the kT com-ponent suffers 6 times as much obscuration as kT . Thereare indications that the shocked gas is not in complete ion-ization equilibrium, which is suggestive of it recently beingshocked. Several gaussian lines are also needed - the cause is speculated to be charge exchange processes. The total X-rayluminosity is 2 . × erg s − .Townsley et al. (2003) previously determined that theX-ray plasma had a mass of 0 . M ⊙ , which when rescaledto a distance of 2.1 kpc becomes 0 . M ⊙ (for an assumeddistance D , V x ∝ D , L x ∝ D , n e , x ∝ ( L x /V x ) / ∝ D − / ,and M x ∝ n e , x V x ∝ D / ). Townsley et al. (2003) determinethat n e , x ∼ . − . The analysis by Hyodo et al. (2008),which does not quite include the most easterly extent ofthe plasma, is generally consistent with the earlier work ofTownsley et al. (2003), except for the determination of asignificantly lower plasma temperature of ≈ .
25 keV.
A3 Westerlund 2 (RCW 49)
Westerlund 2 (hereafter W2) is a compact young open clus-ter embedded in and responsible for the luminous HII re-gion RCW 49. W2 contains at least a dozen OB stars.Two WR stars, WR20a and (especially) WR20b, lie out-side the cluster core (see references in Churchwell et al.2004). W2 is also located in the direction of one of theGalaxys strongest sources of γ -rays (Aharonian et al. 2007;H.E.S.S. Collaboration et al. 2011). The distance to W2 hasbeen very difficult to pin down, with estimates ranging from2 to more than 8 kpc in the literature, but a new study putsit at 2 . ± .
43 kpc, and determines an age of no more than2 Myr (Carraro et al. 2013).Diffuse X-ray emission from W2 was identified in a
Chandra observation (Townsley et al. 2005). The emissionis brightest at the core of W2, and extends preferen-tially towards the west. The emission can be fitted witha 3-temperature thermal plasma model with kT = 0 . kT = 0 . kT = 3 . L x = 3 × ergs s − . At D = 2 .
85 kpc, this increases to L x =4 . × ergs s − . The absorbing column to kT is less thanthe identical columns to kT and kT . The hardest thermalcomponent is not well constrained, and replacing it with apower-law component (Γ = 2 .
3) also gives an acceptablefit. The diffuse flux will be slightly underestimated due tothe use of a 5 arcmin radius extraction region and a nearbyon-chip background region.More recently diffuse emission has also been analyzedfrom a
Suzaku observation (Fujita et al. 2009). The
Chandra pointing was used to determine the point source contamina-tion to the
Suzaku -derived diffuse emission, and the cen-tral region ( r< A4 The Orion Extended Nebula (M42)
The Orion Nebula Cluster (ONC), also known as the Trapez-ium cluster, contains the nearest rich and concentrated sam-ple of pre-MS stars. The OB members of the ONC pho-toionize the Orion Nebula (M 42), a blister HII region atthe near edge of Orion A, the nearest giant molecular cloud( D ≈
450 pc). G¨udel et al. (2008) recently detected dif-fuse, soft (0 . − c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Orion Nebula (EON). The characteristic temperature of theplasma is kT ≈ . . −
10 keV energy band is L x = 5 . × ergs s − .Two regions of diffuse emission, a northern and a south-ern, are identified with respective emission measures ofEM = n V = 1 . × and 1 . × cm − . The atten-uating column N H is very low, being 4 . × cm − for thenorthern, and < cm − for the southern.The total mass of the X-ray emitting gas is estimatedto be 0 . M ⊙ , which is roughly 10 yrs of mass-loss of thedominant O5.5 star θ Ori C. The radiative cooling time isestimated to be ≈ . − . n e = 0 . − . − . The X-ray and ionized gas are in approximate pressure equilibrium( n HII ≈
100 cm − ), and the hot gas is likely channeled by thecooler denser structures rather than disrupting them by ex-pansion. Leakage of the hot plasma via an X-ray champagneflow into the nearby Eridanus superbubble is suggested. A5 The Rosette Nebula
The Rosette Nebula is a blister HII region at the tip ofthe giant Rosette molecular cloud. It has a distinct ring-like appearance in both radio and optical images, and isphotoionized by the open cluster NGC 2244 whose stellarwinds have cleared a hole in the Nebula’s centre (Celnik1985; Townsley et al. 2003). NGC 2244 contains 7 O-typestars, all of which have MS luminosity classes, with the ear-liest spectral type being O4V((f)). A recent analysis of 6 ofthese stars by Martins et al. (2012) determined an upper agelimit of 2 Myr for the most massive stars, in excellent agree-ment with earlier determinations (e.g. Hensberge et al. 2000;Park & Sung 2002). Photometric distance estimates rangebetween 1.4 and 1.7 kpc, and 1.55 kpc is adopted in thiswork. Wang et al. (2008) find an absence of mass segregationand conclude that the cluster is not dynamically evolved.The two dominant O stars (HD 46223, O4V((f)); HD 46150,O5V((f))z) are widely separated (by at least 3 pc). In con-trast, the O stars in the Trapezium Group and M17 areconcentrated within the inner 0.5 pc.Townsley et al. (2003) find that soft diffuse X-rayplasma surrounds the OB association and fills the neb-ula cavity completely. It likely originates from the O-starwinds which are thermalized by wind-wind interactions or byshocking against surrounding molecular material. The X-rayemission is brightest in the central 3 pc radius, correspondingroughly to the central cavity. The diffuse emission can be fitby a two-temperature thermal plasma model, with compo-nents kT = 0 . ± .
02 and kT = 0 . ± . N H = 2 ± × cm − . The hotter com-ponent is dominant. The intrinsic 0 . − D = 1 .
55 kpc) is ≈ × ergs s − . There is no significantemission above 2 keV. Correcting Townsley et al.’s values fora slightly greater assumed distance, the diffuse plasma num-ber density and mass are estimated as n e , x ∼ . − and M x ∼ . M ⊙ . A6 The Quintuplet Cluster
The Quintuplet cluster is named after its five brightest stars(Nagata et al. 1990). It is located near the Galactic Centre, is unusually dense, and is host to at least 10 massive, windy,WR stars and more than a dozen luminous OB supergiants(Figer et al. 1999a,b). It is somewhat less massive and densethan the Arches cluster, however. Its age is estimated to beabout 3 . − kT = 2 . ± . L x ∼ ergs s − . The diffuse emission is much fainter thanthat in the Arches and has a very low surface brightness.It also has essentially the same spectral shape as the inte-grated spectrum from the detected sources. Considering thedistance to the cluster, contamination by unresolved point-sources may be an issue.Wang et al. (2006) analyze a deeper Chandra expo-sure. They report the same concerns as Law & Yusef-Zadeh(2004) and in addition note that the extent of the dif-fuse emission from the Quintuplet cluster is uncertain.With an extraction radius of 1 arcmin, Wang et al. (2006)find that a single-temperature thermal plasma model yields kT = 10 +4 . − . keV and N H = 3 . +0 . − . × cm − , givingan absorption-corrected 2 − L x ∼ × ergs s − . The radial diffuse X-ray intensity profile fallsoff more rapidly than SPH simulations (Rockefeller et al.2005) predict. A7 Westerlund 1
Westerlund 1 (hereafter W 1) is the most massive stellar clus-ter known in the Galaxy (Clark et al. 2005; Brandner et al.2008). It contains a rich population of massive stars whichinclude more than 20 WR stars (Crowther et al. 2006), morethan 80 OB stars, and short-lived transitional objects in-cluding luminous blue variables (LBVs), red supergiants(RSGs) and half the currently known population of yellowhypergiants (YHGs) in the Galaxy. Estimates for its agerange from 3 . ± . ± − ∼ − Chandra observationare analyzed and reported by Clark et al. (2008), while thediffuse emission is analyzed by Muno et al. (2006). The dif-fuse emission has an intrinsic (2 − L x ∼ ± × ergs s − , and a Lorentzian spatial dis-tribution with a HWHM along the major axis of 25 ± ∼ . . − kT ∼ . kT ∼ . . ∼
2. The absorbing column, N H ∼ × cm − .In the thermal model, kT increases with distance fromthe cluster, while in the non-thermal model, Γ is signifi- c (cid:13) , 1–21 H.Rogers and J.M.Pittard cantly steeper in the centre-most region considered. Thereis no evidence for a recent SN explosion. Less than 10 − of the mechanical luminosity is dissipated as 2 − − − ≈ × − ergs cm − s − arcmin − , which isnot consistent with a cluster wind where almost all of thediffuse X-ray emission is produced within the core radius R c (Stevens & Hartwell 2003). A thermal interpretation ofthe halo of diffuse emission is further challenged by the hightemperature and lack of line emission.More recently, Kavanagh et al. (2011) analyze anXMM-Newton pointing and determine that the hard com-ponent in an inner 2 arcmin radius region is actually ther-mal, with a clearly detected He-like Fe 6.4 keV line. No ev-idence of a non-thermal component was found. They re-port that the diffuse emission has a 2 − L x ∼ . × erg s − . A8 The Lagoon Nebula (M8, NGC 6530)
The Lagoon Nebula is an HII region associated with theyoung (13 Myr) open cluster NGC 6530, which contains sev-eral O-stars and about 60 B-stars. It is about 1.3 kpc away(see Henderson & Stassun 2012, and references therein). On-going star formation occurs in several places, notably theHourglass Nebula (the brightest part of M8) and M8 E.The Hourglass Nebula is illuminated by an O7V star (Her-schel 36). Henderson & Stassun (2012) argue that NGC 6530is slightly younger than the Orion Nebula Cluster (ONC),being . .
65 Myr assuming the ONC is 2 Myr old. If theONC is actually 3 Myr old, this would give the HourglassNebula an age of 2.5 Myr. The Lagoon Nebula is summa-rized in Tothill et al. (2008).Rauw et al. (2002) claim that soft diffuse emission was“probably” detected from the southern lobe of the Hour-glass nebula. The emission can be fitted with an absorbedMEKAL model with N H = 1 . +0 . − . × cm − , kT =0 . +0 . − . keV and an intrinsic (0 . − . . × ergs s − . However, there is unboutedly some con-tamination from unresolved point sources. No diffuse emis-sion is seen from a qualitative examination of a Chandra observation of M8 which did not cover the Hourglass Neb-ula (Townsley et al. 2003) - see also Damiani et al. (2004).
A9 The Arches Cluster
The Arches cluster, like the Quintuplet cluster, lies closeto the Galactic Centre, being only 26 pc away in projec-tion (see, e.g. Figer et al. 1999b). It is slightly younger(2 − . Keck -LGS adaptive optics.The deepest
Chandra observation to date is byWang et al. (2006). Diffuse thermal X-ray emission with aprominent Fe K α r < . kT = 2 .
56 keV, τ = 1 . × cm − s, N H = 1 . × cm − and an intrin-sic (2-8keV) luminosity of L x = 3 . × ergs s − .In contrast, the emission in the outer regions of thecluster shows a prominent line at 6.4 keV, a power-lawcontinuum emission of non-thermal origin, and a non-axissymmetric spatial distribution with a bowshock mor-phology. This may result from an ongoing collision betweenthe cluster and the adjacent molecular cloud, which hasa relative velocity of 120 km s − . The interpretation of the6.4 keV Fe K α flourescence emission from neutral Fe is stilldebated, with Wang et al. (2006) favouring a cosmic rayorigin but Capelli et al. (2011) favouring photoionizationof nearby molecular clouds by X-ray photons. Wang et al.(2006) find that the SE extension (which is where the 6.4 keVemission is) is best fitted with a PL+Gaussian spectralmodel with Γ = 1 . +1 . − . and N H = 6 . × cm − , and hasan intrinsic (2 − L x = 4 . × ergs s − .An even more extended “LSBXE” region is fitted witha MEKAL+PL+GAUSSIAN spectral model with kT =0 .
45 keV, Γ = 1 . N H = 9 . × cm − , withan intrinsic (2 − L x = 1 . × ergs s − .Fig. 15 in Wang et al. (2006) shows the radial diffuse X-ray intensity profiles around the Arches cluster. It falls offmuch less rapidly than simulations (Rockefeller et al. 2005). A10 NGC 3576 (RCW 57)
NGC 3576 is a giant HII region located at a distance of 2 . ± . Chandra pointings.A southern pointing was centered on NGC 3576, while anorthern pointing was designed to search for a young stel-lar cluster associated with the O8V+O8V eclipsing binaryHD 97484 (EM Car) and the O9.5Ib star HD 97319. Diffuseemission is seen to the SE of NGC 3576 (hereafter identifiedas NGC 3576S), while hard X-rays were seen in the north-ern pointing. Townsley et al. (2011) identified these sourcesas NGC 3576S and NGC 3576N, respectively. The northernpointing revealed a young cluster (termed NGC 3576OB)which appears older than NGC 3576 to the south.NGC 3576S requires a 2-temperature spectral fit, with kT = 0 . +0 . − . and kT = 0 .
53. The absorbing columnsare N H = 1 . × cm − and N H = 2 . × cm − ,respectively. The softer component dominates the totalemission which has an intrinsic L x = 1 . × ergs s − .Townsley et al. (2011) suggests that the hot plasma respon-sible for this emission has forced itself out through a low-density pathway, analogous to the outflow seen from M17,but seen more face-on and at a slightly earlier phase. A gaus-sian at 0.72 keV (which accounts for 16% of the total emis- c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission sion) is required for a good fit. This may represent chargeexchange processes.In contrast NGC 3576N requires the presence of apower-law component in spectral models. The intrinsic lumi-nosity is L x = 1 . × ergs s − , 24% of which is contributedby a power-law continuum. A 3-temperature model is alsorequired for a good fit, with the hardest NEI component( kT = 0 . A11 NGC 3603
The luminous giant HII region NGC 3603 contains thecompact star cluster HD 97950, which is one of the mostmassive young star clusters in the Milky Way. It con-tains 3 core H-burning WN-stars and up to 50 O-stars(Drissen et al. 1995). The most massive stars in the coreappear to be coeval with an age of about 1 Myr, while lessmassive stars and stars in the cluster outskirts appear to beolder (Melena et al. 2008; Pang et al. 2013, and referencestherein). It shows clear mass segregation, despite its youngage. Pang et al. (2013) suggest that dynamical processesmay have been dominant for the high mass stars. Star forma-tion appears to have occurred almost instantaneously, withKudryavtseva et al. (2012) deriving an upper limit to theage spread of 0.1 Myr. The distance to NGC 3603 is thoughtto be 7 ± Chandra cycle 1 observation was presented byMoffat et al. (2002), who noted diffuse X-ray emissionwithin a central region of 2 arcmin radius with an intrin-sic luminosity L x = 2 × ergs s − . However, this is 20%of the integrated point source emission within this regionand may be completely due to undetected point sources.Townsley et al. (2011) recently re-analyzed this obser-vation, finding 1328 point sources compared to the 348sources found by Moffat et al. (2002). The diffuse X-rayemission is anti-coincident with the mid-IR emission whichtraces the surrounding heated dust. This is consistent withthe hot plasma from the shocked stellar winds filling the cav-ities that they have carved. Excluding an area around thecore of NGC 3603 (which is likely dominated by unresolvedpoint sources) and a region to the west (which may con-tain foreground emission related to the NGC 3576 cluster),the diffuse X-ray emission is dominated by an NEI thermalplasma component with kT = 0 .
53 keV, τ = 2 × cm − s, N H = 2 × cm − , and which contributes 86% of the to-tal intrinsic L x = 2 . × ergs s − . No evidence for chargeexchange processes was found though the exposure is quiteshort. A12 NGC 2024 (The Flame Nebula)
The Flame Nebula, NGC 2024, is one of the nearest sitesof massive star formation ( D = 415 pc, Anthony-Twarog1982). It is part of the Orion B giant molecular cloud (e.g.Mitchell et al. 2001) and is near the Horsehead Nebula. A3D structure of the region was proposed by Barnes et al. (1989) (see also Emprechtinger et al. 2009). Bik et al. (2003)suggested that the O8V-B2V star IRS 2b is the ionizingsource of the HII region, but Burgh et al. (2012) note that itcould be a supergiant. The age of NGC 2024 is unclear, withestimates ranging from 0.3 Myr (Meyer 1996) to several Myr(Comeron et al. 1996).Diffuse X-ray emission with a radius of 0.5 pc from thecentre of NGC 2024 was reported by Ezoe et al. (2006b).The emission has a very hard continuum ( kT > α line. Fitting the data with a “leakyabsorber” model (where emission from a single temperatureplasma reaches the observer via two paths with differentabsorption) returns kT ≈
11 keV with N H = 0 . × and 3 . × cm − . The intrinsic X-ray luminosity in the0 . − L x = 2 × ergs s − . Ezoe et al. (2006b)note that a single massive star with a wind comparable to,or stronger than, that of a typical B0.5V star has enoughenergetics to power the observed X-ray emission. This workshows that diffuse emission is present in a MSFR in whichonly late O to early B stars exist. REFERENCES
Aharonian F., Akhperjanian A. G., Bazer-Bachi A. R.,Beilicke M., Benbow et al., 2007, A&A, 467, 1075Anthony-Twarog B. J., 1982, AJ, 87, 1213Banerjee S., Kroupa P., 2013, ApJ, 764, 29Barbosa C. L., Damineli A., Blum R. D., Conti P. S., 2003,AJ, 126, 2411Barnes P. J., Crutcher R. M., Bieging J. H., StoreyJ. W. V., Willner S. P., 1989, ApJ, 342, 883Becker W., Prinz T., Winkler P. F., Petre R., 2012, ApJ,755, 141Beuther H., Linz H., Henning T., eds, 2008, Massive StarFormation: Observations Confront Theory Vol. 387 of As-tronomical Society of the Pacific Conference SeriesBik A., Lenorzer A., Kaper L., Comer´on F., WatersL. B. F. M., de Koter A., Hanson M. M., 2003, A&A,404, 249Blair W. P., Morse J. A., Raymond J. C., Kirshner R. P.,Hughes J. P., Dopita M. A., Sutherland R. S., Long K. S.,Winkler P. F., 2000, ApJ, 537, 667Brandner W., Clark J. S., Stolte A., Waters R., NegueruelaI., Goodwin S. P., 2008, A&A, 478, 137Broos P. S., Feigelson E. D., Townsley L. K., Getman K. V.,Wang J., Garmire G. P., Jiang Z., Tsuboi Y., 2007, ApJS,169, 353Burgh E. B., France K., Snow T. P., 2012, ApJL, 756, L6Canizares C. R., Winkler P. F., 1981, ApJL, 246, L33Cant´o J., Raga A. C., Rodr´ıguez L. F., 2000, ApJ, 536, 896Capelli R., Warwick R. S., Porquet D., Gillessen S., PredehlP., 2011, A&A, 530, A38Carraro G., Turner D., Majaess D., Baume G., 2013, A&A,555, A50Castor J., McCray R., Weaver R., 1975, ApJL, 200, L107Celnik W. E., 1985, A&A, 144, 171Chevalier R. A., 2005, ApJ, 619, 839Chevalier R. A., Clegg A. W., 1985, Nature, 317, 44Chini R., Hoffmeister V., 2008, Star Formation in M17.p. 625 c (cid:13) , 1–21 H.Rogers and J.M.Pittard
Chu Y.-H., Chang H.-W., Su Y.-L., Mac Low M.-M., 1995,ApJ, 450, 157Chu Y.-H., Mac Low M.-M., 1990, ApJ, 365, 510Churchwell E., Whitney B. A., Babler B. L., IndebetouwR., Meade M. R., et al., 2004, ApJS, 154, 322Clark J. S., Negueruela I., Crowther P. A., Goodwin S. P.,2005, A&A, 434, 949Clark P. C., Bonnell I. A., Klessen R. S., 2008, MNRAS,386, 3Clarkson W. I., Ghez A. M., Morris M. R., Lu J. R., StolteA., McCrady N., Do T., Yelda S., 2012, ApJ, 751, 132Comeron F., Rieke G. H., Rieke M. J., 1996, ApJ, 473, 294Crowther P. A., Hadfield L. J., Clark J. S., Negueruela I.,Vacca W. D., 2006, MNRAS, 372, 1407Damiani F., 2010, in de Grijs R., L´epine J. R. D., eds,IAU Symposium Vol. 266 of IAU Symposium, X-rays fromyoung star clusters: a complement to optical and infraredviews. pp 190–202Damiani F., Flaccomio E., Micela G., Sciortino S., HarndenJr. F. R., Murray S. S., 2004, ApJ, 608, 781Dorland H., Montmerle T., 1987, A&A, 177, 243Dorland H., Montmerle T., Doom C., 1986, A&A, 160, 1Drissen L., Moffat A. F. J., Walborn N. R., Shara M. M.,1995, AJ, 110, 2235Dubner G., Loiseau N., Rodr´ıguez-Pascual P., SmithM. J. S., Giacani E., Castelletti G., 2013, A&A, 555, A9Dunne B. C., Chu Y.-H., Chen C.-H. R., Lowry J. D.,Townsley L., Gruendl R. A., Guerrero M. A., Rosado M.,2003, ApJ, 590, 306Dyson J. E., de Vries J., 1972, A&A, 20, 223Elmegreen B. G., Lada C. J., Dickinson D. F., 1979, ApJ,230, 415Emprechtinger M., Wiedner M. C., Simon R., Wieching G.,Volgenau N. H., Bielau F., Graf U. U., G¨usten R., Hon-ingh C. E., Jacobs K., Rabanus D., Stutzki J., WyrowskiF., 2009, A&A, 496, 731Eriksen K. A., Morse J. A., Kirshner R. P., Winkler P. F.,2001, in Holt S. S., Hwang U., eds, Young Supernova Rem-nants Vol. 565 of American Institute of Physics Confer-ence Series, Fabry-Perot [O III] λ c (cid:13) , 1–21 eedback from Winds and Supernovae in Massive Stellar Clusters. II: X-Ray Emission Meixner M., Haas M. R., Tielens A. G. G. M., EricksonE. F., Werner M., 1992, ApJ, 390, 499Melena N. W., Massey P., Morrell N. I., Zangari A. M.,2008, AJ, 135, 878Mewe R., Kaastra J. S., Liedahl D. A., 1995, Legacy, 6, 16Meyer M. R., 1996, PASP, 108, 380Mitchell G. F., Johnstone D., Moriarty-Schieven G., FichM., Tothill N. F. H., 2001, ApJ, 556, 215Moffat A. F. J., Corcoran M. F., Stevens I. R., SkalkowskiG., Marchenko S. V., M¨ucke A., Ptak A., Koribalski B. S.,Brenneman L., Mushotzky R., Pittard J. M., PollockA. M. T., Brandner W., 2002, ApJ, 573, 191Morse J. A., Winkler P. F., Kirshner R. P., 1995, AJ, 109,2104Muno M. P., Law C., Clark J. S., Dougherty S. M., de GrijsR., Portegies Zwart S., Yusef-Zadeh F., 2006, ApJ, 650,203Nagata T., Woodward C. E., Shure M., Pipher J. L., OkudaH., 1990, ApJ, 351, 83Najarro F., Figer D. F., Hillier D. J., Kudritzki R. P., 2004,ApJL, 611, L105Oey M. S., 1996, ApJ, 467, 666Oskinova L. M., Gruendl R. A., Ignace R., Chu Y.-H.,Hamann W.-R., Feldmeier A., 2010, ApJ, 712, 763Pang X., Grebel E. K., Allison R. J., Goodwin S. P., Alt-mann M., Harbeck D., Moffat A. F. J., Drissen L., 2013,ApJ, 764, 73Park B.-G., Sung H., 2002, AJ, 123, 892Park S., Hughes J. P., Slane P. O., Burrows D. N., GaenslerB. M., Ghavamian P., 2007, ApJL, 670, L121Parkin E. R., Pittard J. M., 2010, MNRAS, 406, 2373Pellegrini E. W., Baldwin J. A., Brogan C. L., HansonM. M., Abel N. P., Ferland G. J., Nemala H. B., ShawG., Troland T. H., 2007, ApJ, 658, 1119Pittard J. M., 2011, MNRAS, 411, L41Pittard J. M., Parkin E. R., 2010, MNRAS, 403, 1657Povich M. S., Churchwell E., Bieging J. H., Kang M., Whit-ney B. A., Brogan C. L., Kulesa C. A., Cohen M., BablerB. L., Indebetouw R., Meade M. R., Robitaille T. P., 2009,ApJ, 696, 1278Povich M. S., Stone J. M., Churchwell E., Zweibel E. G.,Wolfire M. G., Babler B. L., Indebetouw R., Meade M. R.,Whitney B. A., 2007, ApJ, 660, 346Purcell C. R., Minier V., Longmore S. N., Andr´e P., WalshA. J., Jones P., Herpin F., Hill T., Cunningham M. R.,Burton M. G., 2009, A&A, 504, 139Rauw G., Naz´e Y., Gosset E., Stevens I. R., Blomme R.,Corcoran M. F., Pittard J. M., Runacres M. C., 2002,A&A, 395, 499Raymond J. C., Smith B. W., 1977, ApJS, 35, 419Reynoso E. M., Green A. J., Johnston S., Dubner G. M.,Giacani E. B., Goss W. M., 2003, MNRAS, 345, 671Rockefeller G., Fryer C. L., Melia F., Wang Q. D., 2005,ApJ, 623, 171Rodr´ıguez L. F., Gonz´alez R. F., Montes G., Asiri H. M.,Raga A. C., Cant´o J., 2012, ApJ, 755, 152Rodr´ıguez-Gonz´alez A., Cant´o J., Esquivel A., Raga A. C.,Vel´azquez P. F., 2007, MNRAS, 380, 1198Rogers H., Pittard J. M., 2013, MNRAS, 431, 1337Smith B. J., Struck C., Nowak M. A., 2005, AJ, 129, 1350Stevens I. R., Hartwell J. M., 2003, MNRAS, 339, 280Tanaka S. J., Takahara F., 2013, MNRAS, 429, 2945 Tothill N. F. H., Gagn´e M., Stecklum B., Kenworthy M. A.,2008, The Lagoon Nebula and its Vicinity. p. 533Townsley L., Feigelson E., Montmerle T., Broos P., Chu Y.-H., Garmire G., Getman K., 2005, in Sjouwerman L. O.,Dyer K. K., eds, X-Ray and Radio Connections Parsec-Scale X-ray Flows in High-Mass Star-Forming RegionsTownsley L. K., Broos P. S., Chu Y.-H., Gagn´e M., GarmireG. P., Gruendl R. A., Hamaguchi K., Mac Low M.-M.,Montmerle T., Naz´e Y., Oey M. S., Park S., Petre R.,Pittard J. M., 2011, ApJS, 194, 15Townsley L. K., Feigelson E. D., Montmerle T., Broos P. S.,Chu Y.-H., Garmire G. P., 2003, ApJ, 593, 874Wada Q., Tsujimoto M., Ebisawa K., Miller E. D., 2013,PASJ, 65, L2Wang J., Townsley L. K., Feigelson E. D., Broos P. S.,Getman K. V., Rom´an-Z´u˜niga C. G., Lada E., 2008, ApJ,675, 464Wang Q. D., Dong H., Lang C., 2006, MNRAS, 371, 38Weaver R., McCray R., Castor J., Shapiro P., Moore R.,1977, ApJ, 218, 377Winkler P. F., Kirshner R. P., 1985, ApJ, 299, 981Winkler P. F., Twelker K., Reith C. N., Long K. S., 2009,ApJ, 692, 1489Winston E., Wolk S. J., Bourke T. L., Megeath S. T.,Gutermuth R., Spitzbart B., 2011, ApJ, 743, 166Winston E., Wolk S. J., Bourke T. L., Megeath S. T.,Gutermuth R., Spitzbart B., 2012, ApJ, 744, 126Woermann B., Gaylard M. J., Otrupcek R., 2000, MNRAS,317, 421Wolk S. J., Bourke T. L., Smith R. K., Spitzbart B., AlvesJ., 2002, ApJL, 580, L161Wolk S. J., Bourke T. L., Vigil M., 2008, The EmbeddedMassive Star Forming Region RCW 38. p. 124Wolk S. J., Broos P. S., Getman K. V., Feigelson E. D.,Preibisch T., Townsley L. K., Wang J., Stassun K. G.,King R. R., McCaughrean M. J., Moffat A. F. J., Zin-necker H., 2011, ApJS, 194, 12Wolk S. J., Spitzbart B. D., Bourke T. L., Alves J., 2006,AJ, 132, 1100Xu Y., Moscadelli L., Reid M. J., Menten K. M., Zhang B.,Zheng X. W., Brunthaler A., 2011, ApJ, 733, 25Yusef-Zadeh F., Law C., Wardle M., Wang Q. D., FruscioneA., Lang C. C., Cotera A., 2002, ApJ, 570, 665 c (cid:13)000