THE MASSIVE STAR-FORMING REGION CYGNUS OB2. II. INTEGRATED STELLAR PROPERTIES AND THE STAR FORMATION HISTORY

While X-ray observations are excellent at separating young sources from older field stars, this method is not infallible since late-type MS stars of all ages have been observed to have X-ray emission (e.g., Vaiana et al. 1981), albeit at generally lower levels. Nearby MS stars, although intrinsically fainter, could therefore have similar X-ray brightnesses as distant pre-MS stars, and thus be a potential contaminant in our catalog. To remove any foreground sources, we employ one of the interesting features of the IPHAS (r′−Hα, r′ − i′) color–color diagram (CCD) where the MS does not redden onto itself like it does in the near-IR CCD, but sweeps out an area in the color–color plane (Drew et al. 2005). This allows differently reddened stellar populations along the line of sight to be separated (e.g., Drew et al. 2008; Sale et al. 2009).

In sightlines toward the Cygnus region, McCuskey & Houk (1971) found a high concentration of stellar sources between 200 and 300 pc and a strong decline in stellar density between 500 pc and 1 kpc, with the stellar populations making up Cygnus OB2 and Cygnus X beyond that. Figure 1 shows IPHAS CCDs for all IPHAS sources and for just those with Chandra X-ray counterparts in the areas covered by the Chandra observational footprints. It is clear from the CCD, as noted by Drew et al. (2008), that the majority of foreground stars in the direction of Cyg OB2 are nearby and only mildly reddened. The main Cyg OB2 population, as traced by Chandra, can clearly be seen between A_V = 4–7 in the central field and A_V = 2.5–5 in the northwestern field.

Zoom In Zoom Out Reset image size

It is possible that pre-MS members of Cyg OB2 with Hα emission may appear above the reddened MSs and closer to the foreground population (Hα emission, frequently observed in young stars, causes the r′−Hα color to increase). To preclude the removal of genuine members of Cyg OB2 we compare the positions of candidate foreground stars in the near-IR color–magnitude diagram (CMD) with a foreground MS. Those sources whose positions in both diagrams were appropriate for pre-MS sources at the distance of Cyg OB2 and potentially experiencing Hα emission were retained. It should be noted that the majority of removed sources fall on a thin strip in the IPHAS CCD that would exhibit a larger spread if their positions in the CCD were due to the effects of Hα emission.

For sources with only near-IR associations this method was not available to us and we had to rely on the positions of sources in the near-IR CMD. Due to the smaller effects of extinction in the near-IR and the fact that near-IR isochrones redden onto themselves it is not simple to separate the foreground population in a similar manner. Instead we may only identify sources with very blue colors as likely foreground sources and we have removed a small number of outlying sources in this way. In total we removed 46 sources (∼3%) from our catalog as likely X-ray emitting foreground sources, which are listed in Table 1. These were predominantly removed from the 757 sources (∼50% of the original catalog) with IPHAS photometry. Based on the fraction of IPHAS sources removed and the number of sources without IPHAS photometry we estimate that a small fraction <1% of foreground contaminants may remain in the catalog, but will be impossible to identify without spectra or further photometry. Finally, we note that the spatial distribution of sources removed is uniform and does not exhibit a clustered distribution as the overall X-ray source distribution does (see Paper 1). We therefore confidently remove these sources from further consideration.

Figure 2 shows (J, J − H) CMDs for sources in the two fields. We use the J − H color instead of the H − K color because it is less affected by circumstellar dust and has a smaller degeneracy between color and magnitude for pre-MS isochrones. Applying the photometric criteria discussed above to the J- and H-band photometry, there are 1076 and 325 valid sources, 96% and 86% of sources in each field. Both CMDs show a clear sequence with large numbers of sources down to J ∼ 19. The greater depth in the central field is due primarily to the longer exposure of the Chandra observation, although the completeness of the UKIDSS survey in this field is also a benefit. The sources in the central field are also redder, as expected considering the greater extinction in this direction.

Zoom In Zoom Out Reset image size

To determine stellar properties of the population of X-ray sources we have attempted to fit a number of pre-MS isochrones to our data, as shown in Figure 2. We used Siess et al. (2000) isochrones for late spectral types (M <8 M_☉) converted onto the 2MASS system using the data given by Kenyon & Hartmann (1995) and Bessell (2005). For earlier types we used the MS absolute magnitudes in Cramer (1997) and Martins et al. (2005) and colors from Koornneef (1983) and Kenyon & Hartmann (1995). Goodness-of-fit tests, weighting each source using its color and magnitude errors, were performed to estimate the best-fitting isochrone and mean extinction of the sources in each field. Fits were attempted at a range of ages (at intervals of 0.25 Myr) and reddenings (at intervals in A_V of 0.25). To assess the validity of these results, 95% confidence intervals were calculated for all fitted ages and extinctions using a bootstrapping technique in which our data set was resampled 1000 times and re-fit. The range of fitted values provides a reliable estimate of the confidence intervals on each fit. Both the age and extinction were left as free parameters during the refitting.

We note first that, regardless of age, the lower mass stars are better fit by a slightly redder isochrone than the higher mass stars. This effect is most prominent for stars <2 M_☉, and is apparent in both fields, although it is stronger in the northwestern field. This could be attributed to inaccuracies in the isochrones or near-IR colors for massive stars (particularly since a set of empirical 2MASS colors for massive stars does not exist in the literature), but we note that such a discrepancy has not been noted for other SFRs. The use of two different sources for the near-IR photometry raises the question of whether the different photometric systems may have caused this discrepancy, but since the discrepancy is apparent in both fields and the northwestern field uses primarily 2MASS photometry, the different photometric catalogs are unlikely to be responsible. The effect is also visible in the CMD presented by Albacete Colombo et al. (2007, their Figure 4), who used only 2MASS photometry in their analysis of the central Chandra field. The origin of this discrepancy is unclear. The high and variable extinction toward Cyg OB2 and complex star formation history (see discussion in Section 6.1) makes this difficult to resolve. Despite the apparent difference in reddening however, we show below that the difference between the fitted extinctions of the high- and low-mass populations are not significant.

Because of this we fit the high- (8–20 M_☉) and low-mass (<8 M_☉) populations separately. The extinction was allowed to vary for both fits, while the age of the region was only varied for the low-mass population because the high-mass population should already be on the MS and therefore, excluding the highest mass stars which may have evolved off the MS and which we do not include in our fit, are insensitive to age variations. In the central field the high-mass stars are best fit at a reddening of A_V ∼ 7.25^+0.5_−0.5, with the lower-mass stars redder with A_V ∼ 7.5^+0.5_−0.75. In the northwestern field the high-mass stars are fit with A_V ∼ 4.75^+0.5_−0.75 with the low-mass stars having A_V ∼ 5.5^+1.5_−1.0. Since there is no degeneracy between extinction and assumed distance, these values are independent of our choice of distance modulus. Compared to the mean extinctions calculated in Paper 1 from X-ray spectral fits these values are slightly larger, though within the uncertainties calculated here.

When fitting the low-mass component with pre-MS isochrones we attempted fits at both the older distance of 1.8 kpc (DM = 11.2; Massey & Thompson 1991) and the revised distance of 1.45 kpc (DM = 10.8) estimated by Hanson (2003). Using DM = 11.2 the low-mass stars are best fit with a 2.5^+1.0_−1.25 Myr pre-MS isochrone, in agreement with previous studies based on the evolutionary state of the most massive stars in Cyg OB2 (e.g., Hanson 2003). At DM = 10.8 the population is better fit with a slightly older 3.5^+0.75_−1.0 Myr isochrone (as shown in Figure 2). This is older than the upper limit determined by Hanson (2003) based on MS O6 and O7 dwarfs, but within the 95% confidence interval.

In the northwestern field we find consistently older ages for the stellar populations. At DM = 11.2 the low-mass population is best fit at an age of 4.0^+1.75_−1.25 Myr and at DM = 10.8 at an age of 5.25^+1.5_−1.0 Myr. For both fields the use of the closer distance proposed by Hanson (2003) requires an older pre-MS isochrone to fit the data. These ages will be discussed in more detail in Section 6.1.

Figure 3 shows the (J − H, H − K_s) CCD for the 1357 sources that meet our photometric requirements (fewer stars than present in the CMD in Figure 2, on account of requiring satisfactory K_s magnitudes also). For reference we also show the MS track from Kenyon & Hartmann (1995) and the classical T-Tauri star (CTTS) locus from Meyer et al. (1997), both with reddening vectors extending from them. As noted above the majority of sources trace a reddened MS, with a number of sources significantly more reddened still. The higher photometric errors of the 2MASS-dominated northwestern field photometry is also evident when comparing the two fields and from the mean color error shown in each figure. In the central field there are two sources with large J − H colors and low H − K_s colors that do not correspond to plausible stellar colors and as such may have spurious photometry.

Zoom In Zoom Out Reset image size

The near-IR CCD is often useful for estimating the evolutionary class of young stellar objects because sources with circumstellar material will be shifted to higher H − K_s colors due to near-IR excesses in the K_s band. These are Class ii objects (Lada 1987), such as T-Tauri or Herbig Ae/Be stars, which are pre-MS stars with circumstellar accretion disks that are less evolved than the Class iii objects that have lost their inner circumstellar disks and clearly dominate our sample. We find 63 and 23 sources that fall within this region of the color–color plane, 5.9% and 7.9% of our catalog for the central and northwestern fields, respectively. The fraction for the central field is slightly larger than the 4.4% determined by Albacete Colombo et al. (2007), which can likely be attributed to our use of UKIDSS data to probe lower-mass stars that retain circumstellar disks for longer periods of time (e.g., Hillenbrand et al. 1998; Haisch et al. 2001a). To the right of the reddened CTTS is a region typically occupied by less-evolved Class i protostars, or evolved stars, with significant amounts of circumstellar material. We find only one object in this region, and from its isolation and lack of IPHAS photometry we are unable to determine whether it is a true member of Cyg OB2 or a background source.

The fractions of Class i and ii sources are significantly lower than that found from studies of other similar age SFRs in the near-IR such as M17 (37%; Hoffmeister et al. 2008) and the Orion Nebula Cluster (ONC; >55%; Hillenbrand et al. 1998).⁴ Albacete Colombo et al. (2007) suggested that the low fraction of circumstellar disks in Cyg OB2 was due to the strong UV radiation field from the large number of OB stars in the vicinity that causes photoevaporation of the disk material. Although predicted by theory (e.g., Johnstone et al. 1998), an inverse relationship between disk fractions and UV radiation field has only been mildly observed in SFRs to date (e.g., Balog et al. 2007; Guarcello et al. 2007). To a certain extent our results support such a theory, with the fraction of circumstellar disks lower in the center of the SFR where the UV radiation field from OB stars is stronger. However, the difference is both statistically small and this interpretation over-simplifies the dynamical and three-dimensional structure of the region (e.g., Balog et al. 2007) and therefore does not allow an accurate comparison between disk fractions and the UV radiation field. However, If Cyg OB2 contains a significant fraction of sources with ages ⩾5 Myr, then the disk fractions of 5%–10% found here are perfectly reasonable (e.g., Hillenbrand 2005). Given the growing evidence for an older generation of star formation in the region (see discussion in Section 6.1) we suggest that this is a more likely explanation for the low fraction of circumstellar disks than invoking UV photoevaporation from OB stars.

The use of near-IR colors is however an imprecise method of tracing the population of disk bearing stars since a K_s-band excess will only reveal hot dust from the inner parts of a circumstellar disk. The combination of near- and mid-IR photometry is necessary to produce a more complete picture of the evolution of circumstellar disks in a SFR. Forthcoming observations of the Cygnus region from the Spitzer Space Telescope will allow this matter to be probed in more detail.

Feigelson et al. (2005) showed that the median energy of X-ray sources is a reliable indicator for the absorbing column density, log N_H, obtained from spectral fitting procedures. In Figure 4, we show the distribution of median energies of Chandra sources in Cygnus OB2 compared to that of the well-studied Chandra Ultra-deep Orion Project (COUP; Getman et al. 2005b) of the ONC. The peak median energy of Cygnus OB2 sources is higher than that of Orion because of the higher extinction toward Cyg OB2 compared to Orion.

Zoom In Zoom Out Reset image size

The COUP sample also shows a smaller population with a higher median energy around 3 keV that is indicative of a population of embedded sources with higher extinction. However, such a second population is not clearly visible in the Cyg OB2 distribution with the high median energy sources appearing to constitute the tail-end of the standard distribution. This can be quantified as the fraction of sources in the two catalogs that would be classed as “heavily obscured” by some authors, i.e., with 〈E〉 ⩾ 3.0 keV. In Cyg OB2 only 10% of the 1696 sources fall in this class, compared to 21% of Orion sources. This relative lack of embedded sources is observed in both Chandra fields. It is also not a property of the typically higher mass of our Cyg OB2 sources compared to those in the deeper low-mass COUP sample: filtering both samples on X-ray luminosity (which correlates well with stellar mass) and comparing only those sources with L_X > 29.5 erg s⁻¹ (our approximate X-ray completeness limit, see Section 4.2) also shows a deficit of embedded Cyg OB2 sources.

While it appears there is a moderate evolution in stellar X-ray luminosity with age (L_X decreasing with age, but L_X/L_bol increasing; Preibisch & Feigelson 2005), and there might also be an associated change in spectral hardness as stars age, we expect such effects to be very small compared to the gross changes observed in median photon energy as a result of the different absorption characteristics of the ONC and Cygnus OB2 regions. We conclude that the relative lack of sources with high median X-ray energies indicates that, unlike the ONC, the central region of Cyg OB2 does not contain an embedded population of recently formed stars and that star formation in the vicinity of Cyg OB2 has declined significantly.

Different young SFRs, such as Cepheus OB3 (Getman et al. 2006), M17 (Broos et al. 2007), and NGC 6357 (Wang et al. 2007), have all been observed to have X-ray luminosity functions (XLFs) that exhibit similar lognormal shapes. Since the XLF is effectively a convolution of the IMF and the X-ray luminosity–mass relation for pre-MS stars, it may be used to study the properties of a stellar population without making any assumptions about the region or the objects being studied (e.g., by using pre-MS evolution models). Furthermore, if we assume that the IMF is invariant in young clusters, at least as far as can be measured, the completeness of our sample may be estimated by comparing the XLF with that determined from deeper observations of other clusters, such as the COUP study which is complete to ∼0.1 M_☉.

Figure 5 shows a comparison between the XLF of lightly obscured (〈E〉 ⩽ 3.0 keV) Cygnus OB2 sources and those from the well-studied Orion Nebula (we show only the lightly obscured populations since Cyg OB2 has a smaller heavily obscured population compared to the ONC, and this will influence the XLF). Error bars for the Cyg OB2 XLF showing the 1σ confidence level were estimated from Monte Carlo simulations of the XLF, where individual X-ray luminosities were drawn from a Gaussian distribution with mean equal to the measured X-ray luminosity and variances based on the X-ray luminosity error. The COUP XLF has been re-normalized to fit the Cyg OB2 XLF at the intermediate- to high-mass end.

Zoom In Zoom Out Reset image size

Between log L_X ∼ 29.5–31.5 erg s⁻¹ the ONC and Cyg OB2 XLFs are in good agreement. At lower luminosities the two XLFs differ, with fewer Cyg OB2 X-ray sources compared to the ONC. This represents our completeness limit, from which we determine a mass limit of 1 M_☉ using the X-ray luminosity to mass correlation in Preibisch et al. (2005).⁵ Beyond this our completeness drops off rapidly toward lower masses with our sample only 50% complete at 0.75 M_☉ and <20% complete at 0.5 M_☉. Considered separately the two fields show similar XLFs down a similar completeness limit, despite an exposure difference of a factor of 2. This is a product of the L_X–mass relation, which is constantly between 1 and 10 M_☉ and then drops rapidly at lower masses with a high dispersion (Preibisch et al. 2005). Beyond the completeness limit, the XLF falls off quicker in the northwestern field than it does in the central field due to the difference in X-ray exposure and the lack of UKIDSS photometry in the northwestern field.

The MFs of very young SFRs are often described as IMFs, although this is rarely true. The IMF may be disrupted within as little as ∼1–2 Myr due to the ejection of members through interactions and the demise of the highest mass members. Considering the ages estimated in Section 3.1 and evidence in the literature for ejected massive stars from Cyg OB2 (e.g., Comerón & Pasquali 2007), it is important to make this distinction and understand its implications.

Figure 6 shows MFs for both fields in Cyg OB2 and for the combined fields, all of which reach a maximum at ∼1 M_☉, our estimated completeness limit (Section 4.2). Excluding the bins spanning the range 3–10 M_☉ that contain A and late B-type stars not expected to emit X-rays (see Section 2 and our discussion of the completeness of our sample) the MFs show a clear power-law slope, steepening at higher masses. Excluding the most massive stars we find a power-law slope of Γ = −1.09 ± 0.13 in the combined fields, in agreement with the value of Γ = −1.3 ± 0.3 estimated by Kroupa (2001) from accumulated observations, and only slightly shallower than the canonical Salpeter value. When the two fields are considered separately we find negligibly different slopes of Γ = −1.08 ± 0.15 in the central field and Γ = −1.09 ± 0.10 in the northwestern field. When considering only the O-type stars we find a much steeper slope with Γ = −2.72 ± 0.52.

Zoom In Zoom Out Reset image size

Previous measurements of the MF⁶ in Cyg OB2 have mostly been derived from spectroscopy of the massive stars. Massey & Thompson (1991) measured Γ = −1.0 ± 0.3 for all stars more massive than 9 M_☉ and Γ = −0.9 ± 0.5 using only the highest quality photometry and stars more massive than 15 M_☉. Compared to our high-mass MF this is a much shallower slope, although the masses used were derived from older models for massive stars. More recently, Kiminki et al. (2007) derived a slope of Γ = −2.2 ± 0.1 for spectroscopically studied stars more massive than 15 M_☉ across a wide area of Cyg OB2, deriving masses from updated stellar evolutionary models and the revised O star stellar parameters presented by Martins et al. (2005). Our high-mass MF is in good agreement with their value, mainly because of our use of similar models and stellar parameters. However, we have shown here that the MF at lower masses is not as steep and is closer to the canonical value. We suggest that the steep high-mass slope of the MF is not intrinsic but is a product of the older stellar populations in the region having lost their most massive members, thereby steepening the MF.

The only previous measurement of the MF at lower masses is due to Knödlseder (2000) who found Γ = −1.6 ± 0.1 down to 2 M_☉ from a photometric analysis of 2MASS observations in Cyg OB2. However, his use of MS magnitudes and colors for a pre-MS stellar population puts the derived masses into question. Our measurement is therefore the first deep measurement of the pre-MS MF in Cygnus OB2 and its agreement with the Kroupa (2001) value, supporting the view that the IMF is invariant.

The uncertainty in our estimation of the age of Cyg OB2 has repercussions for stellar masses derived from fitting pre-MS isochrones to the near-IR CMD. We also cannot assume that all the stars within a single field were formed coevally (see the discussion in Section 6.1). To test the impact on the MF slope of this age uncertainty we altered the estimated age and regenerated the individual stellar masses and the MF. We find that when the estimated age is varied by ±1 Myr the derived MF slope changes by less than 10%, with the MF in the northwestern field varying much less than in the central field. Even varying the age over a large range (by ±2 Myr) only causes a MF slope variation of ∼20% (for the central field) and <10% in the northwestern field. Age variations larger than this cause much greater variations in the MF slope, particularly in the central field, but are well outside the 95% confidence intervals.

We also studied how the MF would vary if all the stars were not formed coevally but formed over a range of ages (an age spread that might explain the wide spread in the CMD, see discussion in Section 6.1). Since the spread in the CMD could be caused by both a spread in extinction and a spread in ages we did this by selecting ages for each source that would minimize the resulting spread in extinction. This is the opposite process to that used above where a single age was used, thereby minimizing the spread in ages (the true situation is likely to be somewhere between these two extremes). The resulting MF has a slope of Γ = −1.03 ± 0.11 in the combined fields, Γ = −0.98 ± 0.13 in the central field, and Γ = −1.05 ± 0.06 in the northwestern field. These small variations in the slope of the MF strengthen its validity despite the uncertainty over the age of the region.

The main constraints we have on the age of Cyg OB2 are derived from observations of its most massive stars. Hanson (2003) estimated an age of 2 ± 1 Myr to accommodate the observed mix of MS and evolved OB stars: very young sources such as A37 from Comerón et al. (2002), which Hanson (2003) classified as an O5V(f), cannot be older than 1 Myr (based on evolutionary tracks); while supergiants such as Cyg OB2 #9, type O3If (Massey & Thompson 1991), are believed to be at least 3 Myr old. The presence of a MS extending from type O6, which also cannot be older than 3 Myr, supports this age estimate.

There has been growing evidence for a large number of evolved supergiant stars in the region (Massey & Thompson 1991; Comerón et al. 2002). Hanson (2003) noted that as the search for OB stars in the region was extended to larger radii the fraction of evolved massive stars increased, leading her to suggest that our view of Cyg OB2 is becoming contaminated by non-members. Many of the new evolved massive stars identified by Comerón et al. (2002) were found to the south of the original association in a region where Drew et al. (2008) have since uncovered a population of A-type dwarf stars with an age of 5–7 Myr. Drew et al. (2008) suggest that this region may represent a larger and older parent cluster.

In Section 3.1, we derived ages of 3.5^+0.75_−1.0 and 5.25^+1.5_−1.0 Myr for the central and northwestern fields, both with large spreads in the CMD that cannot solely be explained by photometric errors. Such wide spreads in CMDs have previously been interpreted as due to wide age spreads in association members (e.g., Palla & Stahler 2000). However, other authors have suggested that the observed CMD spreads are due to a mixture of photometric errors, variability in pre-MS stars (up to several magnitudes for CTTSs, Herbst et al. 1994), binarity (Preibisch & Zinnecker 1999), episodic accretion phases (Baraffe et al. 2009) or a spread in line-of-sight distances or extinctions. Many of these factors are not well understood, and in the case of Cyg OB2 particularly the spread in extinction. Sale et al. (2009) measured the rise of extinction with distance toward Cyg OB2 and found a substantial increase in extinction at ∼1.5 kpc, which they estimate could be confined to a depth of ⩽100 pc. If a significant spread of extinction did exist across Cyg OB2, it could, combined with the effects of binarity and pre-MS variability, explain a large proportion of the spread in the CMD. Furthermore, given the uncertainties on the two ages derived in this work, the age difference between the two fields is not significant in itself and could fit with a picture of a single burst of star formation in the region.

However, combining the ages derived here with the existing evidence for very young (∼1 Myr) and slightly older (∼5–7 Myr) populations in Cyg OB2 suggests a picture of a region with a large spread in ages. In such a picture the central field, for which we derived an age of 3.5^+0.75_−1.0 Myr, merges with the young OB-star dominated association, while the northwestern field is biased to a greater age. We cannot yet know to what extent there are two distinct generations of stars present, or whether there is a real age spread (with the youngest stars co-located with the OB stars).

As shown in Section 2, the catalog presented in Paper 1 is complete to 1 M_☉ within the caveats of X-ray emitting stars. We can therefore combine the MF determined here and the stellar density within our completeness limits to estimate the total size and mass of the entire Cyg OB2 association. For the purposes of this exercise we will consider the Cygnus OB2 association to consist of all sources young enough to be detected in X-rays (i.e., <100 Myr). While this may include multiple generations of star formation that may have occurred on different spatial scales, until the different populations can be separated this is the only option available to us. Also, because our observations only sample two small fields in the center of the association, we must combine our results with previous estimates of the total spatial extent of the SFR available in the literature.

Knödlseder (2000) analyzed the morphology of Cyg OB2 by fitting a King profile (King 1962) to 2MASS star count data, estimating it to be a spherically symmetric association of ∼2° diameter. This analysis necessarily relied heavily on performing a background subtraction using the stellar content of a nearby field, despite the inherent difficulties involved in doing this along complicated sightlines in the Galactic plane. An improved morphological analysis may be made using the radial stellar density of Chandra sources in the direction of the northwestern field where a radial length of 30′ is possible using the two fields (see Figure 1 in Paper 1).

We find that the stellar density drops off much more quickly than estimated by Knödlseder (2000), with approximately one-third as many stars detected by us in the 1–2.5 M_☉ mass range (where our data remain complete) at distances of >15′, compared with the King profile determined by Knödlseder (2000). Using our radial density profile and assuming that Cyg OB2 is spherically symmetric we have estimated a new King profile for Cyg OB2. Using our peak stellar density and the MF slope estimated above, extrapolated down to 0.01 M_☉ using the multi-stage power-law IMF from Kroupa (2001) we calculate a total stellar content of 5.7 × 10⁴ stars and a total mass of 2.8 × 10⁴ M_☉. However, it is clear from the spatial distribution of A-type dwarfs in Cyg OB2 presented by Drew et al. (2008) that the structure of Cyg OB2 is far from spherical. The authors note a distinct lack of A-type dwarfs to the east of Cyg OB2 and a pronounced concentration toward the southern part of the cluster that may increase the size of the association. We therefore estimate a total stellar mass of (3 ± 1) × 10⁴ M_☉ for the entire association.

Using this estimated density profile and our derived MF we can estimate the expected number of stars of different masses in Cyg OB2. A total mass of 3 × 10⁴ M_☉ implies a total of ∼1200 OB stars, approximately half that predicted by Knödlseder (2000). Our current MF predicts a total number of O-type stars of ∼75, only slightly larger than the currently known number of 65 O-type stars in Cyg OB2 (Massey & Thompson 1991; Hanson 2003; Kiminki et al. 2007; Comerón et al. 2008; Negueruela et al. 2008). This may suggest that there are more O-type stars to be discovered in Cyg OB2. If we assume that the high-mass MF slope has been influenced by the demise of the most massive stars in the region and that it was originally Γ = −1.3 (Massey 1998), we estimate the number of stars that have evolved to their end states to be ∼12. There are no known supernova remnants in Cyg OB2, although there is some evidence for stellar remnants: Bednarek (2003) and Abdo et al. (2009) identify two pulsars within Cyg OB2, with ages ∼2 × 10⁴ yr (Camilo et al. 2009), that likely represent recently expired massive stars, while Comerón & Pasquali (2007) identified an O4If runaway star that appears to have been ejected from Cyg OB2 ∼1.6 Myr ago, possibly following the supernova explosion of a more massive binary companion.

Table 2 lists some of the properties of Cyg OB2 estimated in this work along with those of a number of other massive SFRs. With the exception of the ONC the majority of these regions have not been studied to as low a stellar mass, with properties determined in different ways by different authors (the number of O-type stars is particularly poorly known for many of these regions). Despite this some comparisons may be made. Cyg OB2 is certainly a very high-mass cluster, comparable to the Arches and Quintuplet clusters near the Galactic center or the giant H ii regions Westerlund 2 and NGC 3603 in the Carina spiral arm. Only regions such as Westerlund 1 cluster, W49A, and R136 in the Large Magellanic Cloud appear to be more massive.

Despite their similar masses, these SFRs show significant differences in their stellar densities, ranging from ∼10² M_☉ pc⁻³ for Cyg OB2 and the Quintuplet cluster to ∼10⁵ M_☉ pc⁻³ for NGC 3603 and the Arches cluster. This may represent an inherent difference between two varieties of massive SFR: that of the compact, massive cluster with a very high density core on one hand and the lower density and spatially more extended region on the other. This could be an evolutionary effect since Cyg OB2 and the Quintuplet cluster are both older than NGC 3603 and the Arches cluster, although the age differences are much shorter than the time periods necessary for such a change (Gieles & Bastian 2008). It is more likely that these differences are real and are related to how the molecular cloud collapsed: low density cores suggest formation by triggering or the interaction of gas motions, while gravitational collapse typically results in higher densities (Smith 2008).

Homeier & Alves (2005) note that W49A, which also has a low core density, has a very subclustered structure, a picture that could apply to the wider Cygnus X giant molecular cloud (GMC) with subclusters such as Cyg OB2 (Bica et al. 2003) and Cyg OB9 (e.g., Uyanıker et al. 2001). This structure supports the hierarchical picture of star formation where young clusters and associations represent the dense inner regions of a hierarchy emerging from the GMC in which they formed (e.g., Elmegreen et al. 2006).