The effect of spatial resolution on optical and near-IR studies of stellar clusters: Implications for the origin of the red excess
N. Bastian, A. Adamo, M. Schirmer, K. Hollyhead, Y. Beletsky, G. Carraro, B. Davies, M. Gieles, E. Silva-Villa
aa r X i v : . [ a s t r o - ph . GA ] A ug Mon. Not. R. Astron. Soc. , 1–8 (2014) Printed 9 October 2018 (MN L A TEX style file v2.2)
The effect of spatial resolution on optical and near-IRstudies of stellar clusters: Implications for the origin of thered excess
N. Bastian , A. Adamo , , M. Schirmer , K. Hollyhead , Y. Beletsky , G. Carraro ,B. Davies , M. Gieles , E. Silva-Villa , Astrophysics Research Institute, Liverpool John Moores University, 146 Brownlow Hill, Liverpool L3 5RF, UK Department of Astronomy, Stockholm University, Oscar Klein Centre, AlbaNova, Stockholm SE-106 91, Sweden Max Planck Institut f¨ur Astronomie, K¨onigstuhl 17, D-69117 Heidelberg, Germany Gemini Observatory, Casilla 603, La Serena, Chile Las Campanas Observatory, Carnegie Institution of Washington, Colina el Pino, Casilla 601, La Serena, Chile ESO, Alonso de Cordova 3107, Casilla 19001, Santiago de Chile, Chile Department of Physics, University of Surrey, Guildford GU2 7XH, UK (CRAQ) Universit´e Laval, 1045 Avenue de la M´edecine, G1V 0A6 Qu´ebec, Canada FACom-Instituto de Fsica-FCEN, Universidad de Antioquia, Calle 70 No. 52-21, Medell´ın, Colombia
Accepted. Received; in original form
ABSTRACT
Recent ground based near-IR studies of stellar clusters in nearby galaxies have sug-gested that young clusters remain embedded for 7 −
10 Myr in their progenitor molec-ular cloud, in conflict with optical based studies which find that clusters are exposedafter 1 − <
10 Myr) massive ( > ⊙ )clusters in the nearby spiral galaxy, M83, along with Hubble Space Telescope (HST)imaging in the optical and near-IR, and ground based near-IR imaging, to see how thecolours (and hence estimated properties such as age and extinction) are affected bythe aperture size employed, in order to simulate studies of differing resolution. We findthat the near-IR is heavily affected by the resolution, and when aperture sizes >
40 pcare used, all young/blue clusters move red-ward in colour space, which results in theirappearance as heavily extincted clusters. However, this is due to contamination fromnearby sources and nebular emission, and is not an extinction effect. Optical coloursare much less affected by resolution. Due to the larger affect of contamination in thenear-IR, we find that, in some cases, clusters will appear to show near-IR excess whenlarge ( >
20 pc) apertures are used. Our results explain why few young ( < V < Key words:
Due to the lower-extinction, as well as the near-IR focus ofthe next generation of ground and space based telescopes,the near-IR holds tremendous potential for stellar popula-tion studies. While simple stellar population models pre-dict that the colours of evolving stellar clusters do not varystrongly after 8 −
10 Myr, the near-IR offers the opportu-nity to pick out highly embedded clusters, both throughtheir stellar continuum as well as their nebular line emission (e.g., Br γ ). For example, mid-IR surveys such as the SpitzerGlimpse Survey (Churchwell et al. 2009) allowed the discov-ery of two (and subsequently many more) highly extinctedmassive young clusters located near the end of the Galacticbar (Figer et al. 2006; Davies et al. 2007).Grosbøl & Dottori (2012 - hereafter GD12) have usednear-IR imaging of a sample of relatively nearby (10 −
25 Mpc) spiral galaxies in order to study their cluster pop- c (cid:13) N. Bastian et al. ulations. Their sample contained a number of objects thathad red colours as well as Br γ -emission. This led the authorsto conclude that these clusters were young (ages < V > γ -emission werefound. This result is in apparent contradiction with a num-ber of optically based photometric and spectroscopic sur-veys that find a large number of clusters with ages less than7 Myr (e.g., Bastian et al. 2009, 2012; Whitmore et al. 2010;Adamo et al. 2011a). Additionally, studies of young mas-sive clusters in the Milky Way and nearby galaxies suggestthat clusters transition from being embedded in their natalmolecular cloud to be “exposed” in 1 − ∼
85 %) of the optically detected clustershad radio counterparts, suggesting that the highly embed-ded phase lasts for only a few Myr.Using HST J and H-band imaging (F110W andF160W), Gazak et al. (2013) found a number of opticallydetected massive clusters with ages less than 6 Myr in thenearby spiral galaxy, M83. These clusters have very bluenear-IR colours, reflecting the lack of red-supergiants withinthem (c.f. Westmoquette et al. 2014). Why are such clustersmissing in other near-IR based cluster studies?In addition to the above paradox, a number of recentstudies have found an excess of flux in the near-IR, rela-tive to that expected from simple stellar population models,in studies of extragalactic young clusters (e.g., Reines etal. 2008, 2010; Adamo et al. 2010a,b, 2011a,b). Part of theexcess comes from the nebular emission from the ionised gasaround the clusters (Reines et al. 2010, Adamo et al. 2010a),which begins to significantly contribute to the integratedflux of the cluster from the I-band red-ward. However, evenwhen taking the nebular emission into account, a numberof clusters show near-IR excess that increases with increas-ing wavelength (e..g, Adamo et al. 2010a). The origin ofthis excess is still under debate, with potential contributionsfrom nearby young stellar objects (YSOs), heavily extinctednearby massive clusters, or contamination from nearby redsupergiants.A potential caveat to most near-IR based studies thathave been carried out so far, is the lower spatial resolu-tion achievable from the ground relative to optical, HSTbased surveys. For example, the study of GD12 of clus-ter complexes in a sample of grand design spiral galaxies,use an aperture radius of 0.5”, corresponding to a phys-ical radii of 23 to 63 pc at the distance of their targets(9 . −
26 Mpc). The authors acknowledge that their aper-tures do not cover just individual clusters, but rather containcluster complexes. In comparison, HST based optical stud-ies typically have aperture sizes of 3-10 pc (e.g., Whitmore Due to the resolution of their images, the authors generally didnot study individual clusters, but rather larger associations andcluster complexes, although we refer to them as “clusters” herefor simplicity. Throughout this work we use “nebular emission” to refer toboth line and continuum nebular emission. et al. 2010; Silva-Villa & Larsen 2011). Randriamanakotoet al. (2010) have tested the role of resolution in determin-ing the luminosity function index of cluster/complexes, andfound that decreasing the resolution led to mildly flatterluminosity functions ( ∼ . . α , and I-band mag-nitudes with simple stellar population models. We refer tothe above article for more information on the cluster sample.We compliment this study with ground based VLT-HAWK-I
J, H, K-band images, that allow us to perform photometrywith apertures of 11 to 87 pc, in order to see how the magni-tudes and colours change as a function of aperture size andwavelength.This paper is organised as follows. In § § § For the present work we use archival HST/WFC3 imagesfrom programme IDs 11360 (PI O’Connell) and 12513 (PI.Blair). The dataset consists of imaging with the F336W,F438W, F547M, F657N, F814W, F110W, and F160W fil-ters. We will refer to these filters as U, B, V, H α , I, J, andH, respectively, although no transformations were applied.For one of the fields (Field 1) the F555W filter was used in-stead of the F547M filter. The F110W filter images are onlyavailable for Fields 1 and 2. Further details on the data usedare given in a future work, Silva-Villa et al. (2014). In order to test the effects of resolution on cluster stud-ies, we use the catalogue of clusters and associations fromSilva-Villa et al. (2014). In that work, clusters were iden-tified on seven V-band WFC3 HST pointings that cover alarge fraction of the optical extent of the galaxy. Based on vi-sual inspection, sources were labelled as either class 1 (likely c (cid:13) , 1–8 patial resolution effects on cluster studies clusters, centrally concentrated, resolved objects) or class2 (likely associations, multiple centres, highly elongated).Ages, masses and extinctions were estimated through a com-parison of each sources U, B, V, H α and I-band magnitudeswith simple stellar population models (e.g., Adamo et al.2010a). In the present work, we use the Silva-Villa et al. cat-alogue and select class 1 and 2 sources with ages .
10 Myrand masses > ⊙ . The mass cut was applied in or-der to limit the effects of stochasticity of the stellar IMFin the broad band properties of the sources (e.g., Barbaro& Bertelli 1977; Fouesneau & Lan¸con 2010; Silva-Villa &Larsen 2011; de Meulenaer et al. 2013). VLT/HAWK-I covers a field of view of 8 ′ × ′ with a pixelscale of 0.104”, using a mosaic of 4 HAWAII-2 detector ar-rays. In order to cover M83, a 5 ′ wide dither pattern waschosen. Hence the central 4 ′ − ′ of M83 received 4 timesmore integration time than its outskirts. Data were takenover 11 nights between 2009-01-02 and 2009-03-25 (ESOprogram ID 382.D-0181; PI Gieles). Blank sky fields wereinterspersed for background subtraction.The source density of common astrometric referencecatalogs in this area is insufficient for full distortion cor-rection and image registration. We thus worked with a sec-ondary reference catalog from archival R -band data takenin good seeing with the Wide Field Imager at the 2.2mMPG/ESO telescope (program ID 69.C-0426, PI: Alves). Acommon astrometric solution was derived for all HAWK-I exposures, with an internal uncertainty of ∼ ∼
10 bright and isolated sourcesper filter, per pointing. Aperture photometry for the ∼ , final = Mag , observed + Mag , correction , Table 1.
Characteristics of the VLT/HAWK-I data set. Since thedata were taken over 11 different nights, and a wide dither patternwas used, the image seeing is not constant across the field.Band t exp [s] Seeing J ×
60s 0.36 − H ×
60s 0.33 − K ×
60s 0.37 − BrG × − whereMag , correction = c ∗ (Mag , observed − Mag , observed ) + c , and c and c are constants derived from all of the 2MASSsources used for calibration per filter ( ∼
40 isolated sources).The same approach was used for the 10 pixel apertures.Using this method, the standard deviation of the finaladopted magnitude, relative to the 2MASS magnitude, was ∼ .
06 mag for each filter and aperture size.
Figure 1 shows the H − K vs. J − H colours of our clustersample for different aperture sizes. In each panel, the 11 pcaperture is shown as filled black circles, and a larger aper-ture is shown with filled red triangles, and the two pointsare connected with solid (blue) lines in order to see howeach cluster is affected by the aperture size. Additionally,in each panel we highlight the clusters with blue near-IRcolours with open circles. These are the young, relativelylow extinction clusters (some examples are shown in Fig. 4).Note how these blue clusters move to redder colours as largerapertures are used. The dashed lines (with diamonds) showthe Yggdrasil simple stellar population models (SSP) (Za-ckrisson et al. 2011), from 1 to 10 Myr (from low to high J − H colour) for solar metallicity. Two models are shown,one including nebular (continuum and line) emission (theredder model in H − K ) and one of pure stellar continuum.Additionally, in Fig. 2 we show the same as in the bottompanel of Fig. 1, however only including the clusters withblue near-IR colours (circled). The red-ward trend is clearerin this representation.For the blue clusters in the 11 pc aperture (those cir-cled), note how they move to the red as larger apertures areused. For aperture sizes of 44 pc, nearly all of them havemoved to the “red cloud”, centred at H − K = 0 . J − H = 0 .
6. Hence, studies that use apertures of this sizewould not be expected to find blue clusters.False composite colour images of a sample of clustersand their surroundings are shown in Fig. 3. These clusterswere chosen to have blue near-IR colours (when small aper-tures were used), and also to display a range in H α mor-phology. Such morphology will be discussed in more detailin Hollyhead et al. (in prep.). In the upper panel we show J,H, K-band HAWK-I colour composites, whereas in the mid-dle panels HST/WFC3 B, V, and H α composites are shown.For the near-IR composites, we note that most clusters have c (cid:13) , 1–8 N. Bastian et al. a source within 44 pc with a brightness that is similar to, orhigher than, the central cluster.We note that a ∼
23 pc radius is the smallest apertureused in the GD12 study. This explains why young clusters intheir survey (those with Br γ emission) have red colours. It isnot due to extinction, but rather contamination from nearbysources that are bright in the near-IR as well as the nebularemission from the ionised shell (that is often seen aroundyoung clusters with sizes between 10 −
50 pc - Hollyhead etal. in prep.).Additionally, we explored the effect of adopting differ-ent annuli surrounding the clusters in order to subtract thebackground flux. For older clusters ( >
10 Myr) the mea-sured colours and magnitudes were not strongly affected bythe size of the annuli. For younger ( <
10 Myr) clusters wefound the following results. As expected, the measured mag-nitudes and colours displayed the least scatter when smallannuli ( <
20 pc) were used. For annuli larger than 100 pc we also found that the magnitudes and colours displayedsmall scatter (i.e. differences between using annuli of 100 and150 pc). For intermediate annuli (20 −
100 pc) the scatter wasthe largest. The reason for this is that young clusters tendnot be born in isolation, but rather often have surroundingclusters and/or young field populations around them (e.g.,Bastian et al. 2005). Hence, whether or not a bright nearbysource (or a number of them) enter the background annulican have a significant effect on the measured colours andmagnitudes.While the main goal of the present work is to explore theeffect that resolution plays in the study of young clusters,we also investigated the effect on older clusters. To do this,we carried out similar tests on a sample of clusters withages between 20 and 200 Myr, taken from the Bastian etal. (2012) sample. We found that for these clusters the near-IR (and optical) colours were not dependent on the aperturesize used. The reason for this is that these clusters are rela-tively isolated, and are no longer associated with their highlystructured natal star-forming region. These star-forming re-gions dissolve into the field on ∼
10 Myr timescales (e.g.,Gieles & Portegies Zwart 2011), so finding an older cluster( >
20 Myr) near a younger ( <
10 Myr) cluster is much rarerthan finding a young cluster ( < −
10 Myr). Hence, we conclude that the young( <
10 Myr) regions will be much more affected by the spa-tial resolution than older clusters ( >
20 Myr).
We carried out a similar experiment in the optical and showthe results in Fig. 4. Here we only show the results for the11 and 87 radius apertures, as even at these extremes, it isclear that the effect of resolution is much less in the opticalthan in the near-IR. Additionally, we are only showing thoseclusters which appear blue in the 11 pc aperture near-IRcolour-colour plot (i.e., those circled in the top panel Fig. 1).The youngest, non-extincted clusters ( < V − I = 0 . U − B = − . V > − -0.5 0.0 0.5 1.0 1.5H - K-0.50.00.51.01.5 J - H Aperture: 11 pcAperture: 22 pc-0.5 0.0 0.5 1.0 1.5H - K-0.50.00.51.01.5 J - H Aperture: 11 pcAperture: 44 pc-0.5 0.0 0.5 1.0 1.5H - K-0.50.00.51.01.5 J - H Aperture: 11 pcAperture: 87 pc
Figure 1.
The effect of aperture size on the resulting near-IRcolours of young ( .
10 Myr) clusters. The black dots show thecolour of clusters when a ∼
11 pc aperture is used. In the up-per panel we highlight blue clusters with open circles. Red dotsrepresent the same clusters but now measured with larger aper-tures, connected by solid (blue lines). Note that the blue clustersbecome redder and join the main (i.e. the “red cloud”) locus ofpoints as larger apertures are used.c (cid:13)000
11 pc aperture is used. In the up-per panel we highlight blue clusters with open circles. Red dotsrepresent the same clusters but now measured with larger aper-tures, connected by solid (blue lines). Note that the blue clustersbecome redder and join the main (i.e. the “red cloud”) locus ofpoints as larger apertures are used.c (cid:13)000 , 1–8 patial resolution effects on cluster studies M a gn i t ud e ( V e g a )
11 pc22 pc44 pc87 pc
UB V H a I J H
UB V H a I J H
Wavelength [micron]
UB V H a I J H
UB V H a I J H
UBV H a I J H
Figure 3.
False colour composite images of a sample of clusters used in the current work.
Upper panels:
HAWK-I J, H, K-bandcomposites.
Middle panels:
HST WFC3 B, V, H α -band composites. The apertures used (11, 22, 44, 87 pc) are shown. In most of thenear-IR images, a source with a brightness similar to or larger than the central source is contained within the 44 pc aperture. However, inthe optical images, the central source often remains the dominant source even within the 87 pc aperture. Lower panels:
The HST based(including J and H-band) spectral energy distribution for the five sources above for different aperture sizes, each has been normalised tothe flux in the V-band. Note that for larger aperture sizes, the flux in H α increases strongly, due to picking up the emission outside thehole that has formed around the central massive cluster. Also, note that all clusters display significant near-IR (beginning in the I-band)excess as larger apertures are used, while the optical/UV magnitudes remain largely unchanged. -0.5 0.0 0.5 1.0 1.5H - K-0.50.00.51.01.5 J - H Aperture: 11 pcAperture: 87 pc
Figure 2.
The same as the bottom panel of Fig. 2 except here, forclarity, only clusters with blue colours (when a ∼
11 pc apertureis used), are shown. colour-colour space would still lead to the same conclusion,i.e. that they are young clusters with high extinction.In order to quantify this, we fit all the clusters shown inFig. 4 to SSP models using the method discussed in Adamoet al. (2010a,b,2011a,b; see also § V of 0.2 mag (and a standard deviation of 0.25 mag). By us-ing the larger apertures, the mass of the cluster was over-estimated by 0.8 dex, highlighting the significants amountof flux coming from nearby sources when studying young( <
10 Myr) clusters. Hence, we conclude that the estimatedage and extinction of the clusters, based on optical and UVphotometry, are not strongly affected by aperture size.The use of larger apertures does result in (at least)one clear change in the optical spectral energy distribution(SED), namely the strength of emission lines. As seen in themiddle panels of Fig. 3, the young clusters often clear outa hole in the surrounding ISM, due to the combination ofstellar winds, SNe, and their ionising flux. These combine to c (cid:13) , 1–8 N. Bastian et al. -0.5 0.0 0.5 1.0 1.5 2.0F547M (or F555W) - F814W-2.0-1.5-1.0-0.50.00.51.0 F W - F W Aperture: 11 pcAperture: 87 pc
Figure 4.
Similar to Fig. 1 but now for optical colours. We onlyshow the most extreme case (apertures of 11 and 87 pc) as even inthis extreme case, it is clear that the colour changes would not beenough to significantly shift the best fitting age and extinction. drive an ionised shell into the surrounding ISM. When theapertures cover the ionised ring (the edge, in projection, ofthe 3D shell) the amount of flux in emission lines goes updrastically.
As discussed above, the near-IR colours and magnitudes arefound to be more strongly affected by neighbouring sourcesand nebular emission than optical colours. This could ex-plain (at least partially) the origin of the near-IR flux ex-cess (relative to SSP models with or without the inclusion ofnebular emission) that has been found in a number of stud-ies of young extragalactic clusters (e.g., Reines et al. 2008,Adamo et al. 2011a). As seen in the bottom panels of Fig. 3,all the clusters begin to display significant amounts of near-IR excess when apertures greater than 20 pc are used. Wenote that the excess near-IR emission found by Reines etal. (2008) is largely due to nebular emission (Reines et al.2010), whereas Adamo et al. (2010, 2011a,b) used modelsthat included nebular emission, suggesting that the excessnear-IR found in those studies has a different origin.In Fig. 5 we investigate the cause of this by looking atthe predicted flux in different bands as a function of agefor a simple stellar population (we use the
Yggdrasil
SSPmodels for solar metallicity, and without nebular emission).The models are normalised to the flux at an age of 3 Myr.After ∼ − ∼ −
30 Myr (the secondary population), the combinedflux will be dominated by the primary population in theoptical and the secondary population in the near-IR.In addition to contamination from neighbouring stellarsources, another potential source of the infrared excess is thecontribution from the nebular emission, including both the line and continuum (e.g., Reines et al. 2010). As young clus-ters emerge from their natal cocoon, their collective stellarwinds, SNe, and ionising flux drive a shell of ionised ma-terial away from the cluster (see examples shown in Fig. 3and also Whitmore et al. 2011). While the cluster is still par-tially embedded, it should show the red-excess phenomenon(relative to SSP models that do not include nebular emis-sion) even with small apertures. As the shell expands, thenebular emission will drop in small apertures. However, asthe aperture increases the contribution of the nebular can in-crease rapidly as the ionised shell enters the aperture. Hence,nebular emission also moves the clusters from blue near-IRcolours to the “red cloud” observed in Figure. 1.A useful comparison can be made with the 30 Doradusregion in the LMC. In addition to the massive ( ∼ × M ⊙ )young (2 − ii region, a number of older clusters (and stellar populationsin the field) are nearby. One of these clusters is Hodge 301,a ∼ −
25 Myr, ∼ ⊙ cluster located ∼
40 pc awayfrom R136 in projection, which contains a number of red su-pergiants (Grebel & Chu 2000). Based on the models shownin Fig. 5, Hodge 301 would be expected to contribute 50%as much flux as R136 in the K-band, and around 1% asmuch in the U-band. Additionally, the 30 Doradus regionis known to host stellar populations (both between clustersand within the field) with significant age differences (tensof Myr - e.g., Walborn & Blades 1997). These populationswould contribute significantly to the flux of R136 if aperturesgreater than ∼
20 pc were used. The above example high-lights the role that other nearby clusters may have on theestimated properties of a massive clusters if a large apertureis used, however, a substantial field population (e.g., indi-vidual RSGs or AGBs) would have the same result. Efremov& Elmgeen (1998) have shown that the age spread presentin a region scales with the size of the region under study,so larger apertures will naturally sample larger age spreadswithin a region.Most studies that have found the near-IR flux excess arebased on distant galaxies, where large (physical) aperturesare required due to resolution restrictions (e.g., Reines etal. 2008, Adamo et al. 2011b who used apertures of ∼
40 and ∼
35 pc, respectively). Based on the results presented here,it appears that the basic properties of the clusters (whichare estimated largely from optical colours) such as age andextinction are likely to be representative of the dominantcluster. However, the near-IR integrated flux may be con-taminated by neighbouring sources and nebular emission.This excess flux accounts for some fraction of the observednear-IR excess, the exact amount will depend on the sur-roundings of the clusters. In a future work, we will directlyaddress the source of the NIR excess using high-spatial reso-lution VLT/SINFONI near-IR integral field spectroscopy ofvery young star clusters in M83 and Haro 11 (Adamo et al.,in prep.). However, of the five clusters shown in Fig. 3, threedisplay strong H α emission and the red excess phenomenonwhen large apertures are used. We have carried out SED fitting for the five clusters shownin Fig. 3 using the same method and SSP models (
Yggdrasil , c (cid:13) , 1–8 patial resolution effects on cluster studies R e l a t i v e f l u x U BVIJHK
Figure 5.
Predictions from the
Yggdrasil
SSP models of the rel-ative flux (normalised to the flux at an age of 3 Myr) in differ-ent bands as a function of age. Note that the flux in the near-IR (JHK) increases strongly after ∼ solar metallicity, Kroupa 2001 stellar IMF) as in Adamo etal. 2010a,b, 2011a,b), for each of the apertures used. Themodels include nebular emission (both line and continuumemission), and we assume that 50% of the ionising photonsare absorbed and re-emitted within the aperture. We onlyused the observed fluxes in the optical for the fits, and thencompared the predicted J (F110W) and H (F160W) fluxesof the best fitting model to the HST observations.In general, we find that for the smallest (11 pc) aper-tures, the observed J and H-band fluxes are significantlyfainter than the predictions. This is due to the fact thatthe young clusters (as seen in Fig. 3) have cleared most ofthe natal gas cloud away (at least to radii larger than theaperture), meaning that significantly less gas is present thanthe model assumes, leading to lower emission (both in thecontinuum and line emission) than the model predicts. Asthe apertures become larger (22 pc), the predicted and ob-served near-IR fluxes become comparable. For the (at leastpartially) embedded cluster in Fig. 3 (10438) the observedvs. predicted fluxes in the near-IR agree for the largest aper-tures (87 pc), suggesting that at this radius, roughly half ofthe ionising photons are absorbed. For the other clusters, thelargest apertures have significant amounts of excess emissionabove the model fit, showing that neighbouring sources areaffecting the photometry.The extreme sensitivity of the near-IR (and the opti-cal emission lines) on the adopted ’covering factor’ of theemission (i.e., the fraction of ionising photons absorbed)shows that care must be taken when including these filtersin broad-band fits for stellar population properties. The testperformed on the M83 clusters suggests that, in some cases,when only the optical bands are fitted, our adopted models(which include stellar and nebular emission) are not able to reproduce the observed near-IR fluxes in larger apertures(i.e., a near-IR excess is present).Hence, there are two effects that contribute to the near-IR excess, nebular emission and nearby contaminating stel-lar populations. Quantifying the amount that each type con-tributes is difficult. Even models that take nebular emissioninto account may miss the actual flux, as they need to as-sume a covering fraction (i.e. the fraction of ionising photonsthat are absorbed within the aperture used). Throughoutthis work, we have adopted the 50% covering factor modelsto estimate the cluster parameters. From Fig. 3 we can seethat some clusters have blow ionised bubbles around them,meaning that the covering fraction is lower than assumed inthe centre (i.e. for small apertures) but may be appropriatefor larger apertures. Still other clusters appear to have “blowouts”, where a large fraction of the ionising photons may beescaping to large distances. In principle, by modelling theSED with multiple emission lines included, one may be es-timate the fraction of ionising photons that are escaping asa function of radius. In practice, however, such modelling islimited by the uncertainties in the ionising photon output ofmassive star models, making the conversion between the ob-served emission line strength to the ionising photon escapefraction highly uncertain (e.g., Doran et al. 2013). We find that the apparent contradiction between the prop-erties (age/extinction) of young cluster populations in thenear-IR and optical reported in the literature (with near-IR studies finding that clusters remain highly extincted for ∼ ∼ γ associated with them) will appear red, and will be assignedhigh extinction values.On the other hand, optical studies do not appear to beas heavily affected by resolution. This is due to two effects.The first is that the natural extent of cluster colours in theoptical is larger, ranging ∼ . U − B colour,relative to ∼ . J − H or H − K . Secondly, con-tamination from nearby (older) sources is much stronger inthe near-IR than in the optical (Fig. 5). This contamina-tion forces the measured integrated colours (of the clusterand surroundings) to the “red cloud” of points. This in turnleads to erroneous inferences about the extinction (and pos-sibly age, although this is usually determined by the pres-ence/absence of emission line flux) of these clusters. Hence,unless high resolution ( .
20 pc) apertures can be employed,it is not possible to infer the age or extinction distributionsof clusters with broadband photometry in the near-IR, po-tentially invalidating recent results based on larger apertures(45 −
100 pc - e.g., Grosbøl & Dottori 2012, 2013). In gen- c (cid:13) , 1–8 N. Bastian et al. eral, using aperture sizes closest to the size of the clusters(radii of 5 −
10 pc) leads to the best results.Contamination from nearby sources has a larger effecton near-IR than optical colours. For young clusters, we foundthat using aperture sizes between 11 and 87 pc did notfundamentally alter the optical colours, and hence the esti-mated properties, of YMCs. This is in agreement with pre-vious photometric studies of clusters from the ground andHST, that found consistent results (e.g., Larsen 2002).By studying how the spectral energy distribution ofclusters change as a function of aperture size, we found thatlarger apertures result in a significant near-IR flux excess insome young clusters. This is caused by the contribution ofneighbouring sources as well as the nebular emission fromnearby (often related) ionised gas which is stronger at redderwavelengths (e.g., Reines et al. 2010; Adamo et al. 2010a).Models that include nebular emission can explain some ofthis excess near-IR emission, however, for some clusters themodels cannot account for the additional flux. It is in thesesources that nearby stellar contamination in the near-IR isdominant.Finally, we note that increasing the aperture size signif-icantly affects the amount of nebular line emission found forthe clusters (see the bottom panel in Fig. 3). This is due tothe fact that young clusters have cleared out a large amountof the (ionised and neutral) gas/dust around them, leavinga shell surrounding the cluster, the size of which should de-pend on the age and mass of the cluster and well as thedensity of the surrounding ISM. When larger apertures areused, more of the shell is included, leading to larger amountsof ionised gas in the aperture, and strong nebular emission.In a future work (Adamo et al. in prep.) we will usenear-IR integral field spectroscopy of many of the regionspresented here, to quantify the amount of the “red excess”that is due to ionised gas and stellar contamination. Addi-tionally, in Hollyhead et al. (in prep.), we will explore theeffect of including emission lines (e.g., H α ) in the SED fit-ting, in particular in cases where the clusters have drivenlarge ionised bubbles outside the size of the aperture. ACKNOWLEDGMENTS
We thank Preben Grosbøl and Horacio Dottori for insight-ful discussions and the referee for a careful reading of themanuscript and for helpful suggestions. NB and MG are par-tially funded by a Royal Society University Research Fellow-ship.
REFERENCES
Adamo, A., ¨Ostlin, G., Zackrisson, E., et al. 2010a, MNRAS, 407,870Adamo, A., Zackrisson, E., ¨Ostlin, G., & Hayes, M. 2010b, ApJ,725, 1620Adamo, A., ¨Ostlin, G., & Zackrisson, E. 2011a, MNRAS, 417,1904Adamo, A., ¨Ostlin, G., Zackrisson, E., & Hayes, M. 2011b, MN-RAS, 414, 1793Barbaro, C., & Bertelli, C. 1977, A&A, 54, 243Bastian, N., Gieles, M., Efremov, Y. N., & Lamers, H. J. G. L. M.2005, A&A, 443, 79 Bastian, N., Trancho, G., Konstantopoulos, I. S., & Miller, B. W.2009, ApJ, 701, 607Bastian, N., Adamo, A., Gieles, M., et al. 2012, MNRAS, 419,2606Churchwell, E.,Babler, B. L., Meade, M. R., et al. 2009, PASP,121, 213Davies, B., Figer, D. F., Kudritzki, R.-P., et al. 2007, ApJ, 671,781de Meulenaer, P., Narbutis, D., Mineikis, T., & Vanseviˇcius, V.2013, A&A, 550, A20Doran, E. I., Crowther, P. A., de Koter, A., et al. 2013, A&A,558, A134Efremov, Y. N., & Elmegreen, B. G. 1998, MNRAS, 299, 588Erben, T., Schirmer, M., Dietrich, J. P., et al. 2005, AN, 326, 432Figer, D. F., MacKenty, J. W., Robberto, M., et al. 2006, ApJ,643, 1166Fouesneau, M., & Lan¸con, A. 2010, A&A, 521, A22Gazak, J. Z., Bastian, N., Kudritzki, R.-P., et al. 2013, MNRAS,430, L35Gieles, M., & Portegies Zwart, S. F. 2011, MNRAS, 410, L6Grebel, E. K., & Chu, Y.-H. 2000, AJ, 119, 787Grosbøl, P., & Dottori, H. 2012, A&A, 542, A39Grosbøl, P., & Dottori, H. 2013, A&A, 551, L13Larsen, S. S. 2002, AJ, 124, 1393Longmore, S. N., Kruijssen, J. M. D., Bastian, N., et al. 2014,PPVI in press (arXiv:1401.4175)Reines, A. E., Johnson, K. E., & Hunt, L. K. 2008, AJ, 136, 1415Reines, A. E., Nidever, D. L., Whelan, D. G., & Johnson, K. E.2010, ApJ, 708, 26Schirmer, M. 2013, ApJS, 209, 21Silva-Villa, E., & Larsen, S. S. 2011, A&A, 529, A25Silva-Villa, E., Adamo, A., Bastian, N., Fouesneau, M., & Zack-risson, E. 2014, MNRAS, 440, L116Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131,1163Walborn, N. R., & Blades, J. C. 1997, ApJS, 112, 457Westmoquette, M. S., Bastian, N., Smith, L. J., et al. 2014, ApJ,789, 94Whitmore, B. C., & Zhang, Q. 2002, AJ, 124, 1418Whitmore, B. C., Chandar, R., Schweizer, F., et al. 2010, AJ,140, 75Whitmore, B. C., Chandar, R., Kim, H., et al. 2011, ApJ, 729, 78Zackrisson, E., Rydberg, C.-E., Schaerer, D., ¨Ostlin, G., & Tuli,M. 2011, ApJ, 740, 13
This paper has been typeset from a TEX/ L A TEX file preparedby the author. c (cid:13)000