The Massive Star-forming Regions Omnibus X-ray Catalog, Second Installment

The Chandra/ACIS observations used for MOXC2 are summarized in Table 2 and are identified by a unique Observation Identification (ObsID) number. All of these data sets are available in the Chandra archive. Observations are ordered by date for each target; target names as used in this paper are given in Column (1) in bold, followed by the original name assigned by the study's principal investigator (PI), noted in Column (11). Those observations where Gordon Garmire is listed as PI came from the ACIS Instrument Team Guaranteed Time Observation (GTO) program; those listing Stephen Murray as PI came from the HRC Instrument Team GTO program. We note this fact because the Chandra GTO programs provided many of the original seed observations for the MSFRs studied here and for many other Galactic Plane observations that we use in our wide-field mosaics.

Virtually all of the ACIS data we have previously published were taken with the Observatory in the ACIS imaging array (ACIS-I) configuration (Garmire et al. 2003), with the optical axis near the center of a 2 × 2 array of 1024 × 1024-pixel CCDs covering roughly 17' × 17' on the sky (with 0492 pixels). Although such observations commonly include data from two CCDs lying far off-axis in the ACIS spectroscopy array (ACIS-S), we previously chose to discard data from those CCDs due to the poor angular resolution of the Chandra mirrors at large off-axis angles.⁷

In order to widen the field coverage around MSFRs and to include more archival data, MOXC2 includes the off-axis CCDs in ACIS-I observations and imaging observations taken in the ACIS-S configuration (Garmire et al. 2003), as shown in the MOXC2 exposure map mosaics in Section 4. Four of our MSFRs were observed with the ACIS-S imaging configuration. One side effect of this more liberal data selection policy is that most observations presented here include both frontside-illuminated (FI) and backside-illuminated (BI) CCD detectors (Garmire et al. 2003). Because FI and BI detectors exhibit very different background spectra, their analysis must be done separately; thus we split each Chandra ObsID into separate FI and BI "virtual" observations within our data analysis workflow.

Our data reduction, diffuse emission analysis, and point source detection, extraction, and masking techniques employ several innovations beyond standard Chandra procedures, as discussed at length by Broos et al. (2010). These techniques were standardized for MSFRs by the CCCP and improved during the MYStIX and MOXC1 studies. Many of the CCCP data analysis steps were implemented by the ACIS Extract (AE) software package⁸ (Broos et al. 2012; Broos & Townsley 2016) and are described in detail in Broos et al. (2010). Nearly identical procedures were applied to the MOXC2 MSFRs, so we do not provide an exhaustive review of those procedures here. A few of the basic steps and improvements to methodologies previously published are described below.

The Chandra data analysis system, CIAO⁹ (Fruscione et al. 2006), the SAOImage DS9¹⁰ visualization tool (Joye & Mandel 2003), and the Interactive Data Language¹¹ (IDL) are used throughout our data analysis workflow, from data preparation through science analysis. Models of the local point-spread function (PSF) of each source, built by the MARX¹² observatory simulator (Davis et al. 2012), play a central role in source finding, constructing extraction apertures, accounting for source crowding when extracting local backgrounds, calculating energy-dependent aperture corrections for calibration data products, modeling detector pile-up effects (Section 3.2), and masking point sources for diffuse analysis (Section 2.2.4). As recommended by the Chandra X-ray Center,¹³ we tuned two MARX parameters to produce PSFs that match our own data: AspectBlur=007 and pix_adj=NONE for ACIS-I observations; AspectBlur=007 and pix_adj=EDSER for ACIS-S observations. Note that this tuning applies only to MARX version 5.3.0.

2.2.1. Point Source Detection Strategy

2.2.2. Astrometric Alignment

2.2.3. Validation of Source Candidates

2.2.4. Diffuse Emission

The primary data product of MOXC2 is a catalog of properties for 18,396 point sources found in our 13 ACIS mosaics of MSFRs and their surrounding Galactic Plane environments. This catalog provides the same source information as that given in MOXC1 and is similar to catalogs from CCCP (Broos et al. 2011), MYStIX (Kuhn et al. 2013), and SFiNCs (Getman et al. 2017).

Table 3 defines the columns of the MOXC2 point source catalog. This catalog is available in FITS format from the electronic edition of this article and may be available in many other formats from VizieR (Ochsenbein et al. 2000). All photometric quantities in this table are apparent (not corrected for absorption). The suffixes "_t," "_s," and "_h" on names of photometric quantities designate the total (0.5–8 keV), soft (0.5–2 keV), and hard (2–8 keV) energy bands. Energy-dependent correction for finite extraction apertures is applied to the ancillary reference file (ARF) calibration products (see Broos et al. 2010, Section 5.3); the SrcCounts and NetCounts quantities characterize the extraction and are not aperture-corrected. The only calibrated quantities in the catalog are apparent photon flux in units of photon cm⁻² s⁻¹ (see Broos et al. 2010, Section 7.4), and an estimate for apparent energy flux in units of erg cm⁻² s⁻¹ (Getman et al. 2010). Additional information regarding the definition of source properties is provided in the table notes.

We caution potential users of these results that we have not attempted to clean the point source catalog for sources that might be spurious reconstruction peaks (e.g., over-reconstructions of diffuse emission). This can be a problem for bright PWNe, SNRs, and dust-scattering halos around X-ray binaries. As we describe each target mosaic in Section 4 below, we will point out areas of particular concern for such spurious point sources.

As described in detail in MOXC1, ACIS detections of bright X-ray sources can suffer from a nonlinearity known as photon pile-up,¹⁸ where multiple X-ray photons are mistakenly detected as a single event because they fell close to each other on the CCD and arrived during the same readout frame. Pile-up causes photometry to be underestimated and the spectrum to be hardened. We check for pile-up in every observation of a source by estimating the observed count rate in a 3 × 3 pixel detection cell centered on the source position. (The highest rate found in all observations of the source is reported by the source property RateIn3x3Cell.)

For each source extraction in which RateIn3x3Cell exceeded a threshold of 0.05 count/frame, we modeled pile-up using a Monte Carlo approach that reconstructs a pile-up-free spectrum from a piled ACIS observation (Broos et al. 2011). Table 4 lists those extractions; column (8) characterizes the inferred level of pile-up.¹⁹ For those sources, several entries in Table 3 are expected to be biased by pile-up effects (in an energy-dependent way). Users are warned that photometric quantities for piled-up sources should be used with caution; higher pile-up corrections and narrower bandpasses should evoke the most caution. Alternatively, we provide pile-up corrected spectra for all source/ObsID entries in Table 4, as described below. Fitting those spectra will result in more meaningful source properties than possible from pile-up distorted photometry.

The Zenodo data repository²⁰ archives many MOXC2 reduced data products²¹ (Townsley & Broos 2017), including astrometrically aligned event lists and exposure maps, DS9 region files representing point source extraction apertures and PSF hooks, source photometry in 16 energy bands, reconstructed spectra for sources that suffer from photon pile-up (Section 3.2), lightcurve plots for the brighter sources, event lists and exposure maps with point sources masked, and smoothed images of diffuse emission (Section 4). For the piled sources in Table 4, we provide reconstructed spectra for each ObsID in which the source suffered from pile-up. The "README" file in the Zenodo archive describes the files there.

Many of the brighter X-ray point sources in MOXC2 also appear in the Chandra Source Catalog²² (Evans et al. 2010). Some may also have data available from the XMM-Newton mission's EPIC camera; catalogs of XMM  sources are available from the XMM-Newton Survey Science Centre.²³ In crowded regions such as MSFRs, care should be taken in matching MOXC2 sources to XMM  sources due to the mismatch in spatial resolution between the Chandra and XMM  telescopes.

Several short archival Chandra observations in and around MSFRs were obtained by the snapshot survey Chasing the Identification of ASCA Galactic Objects (ChIcAGO; Anderson et al. 2014). We use these data sets in the MOXC2 mosaics, and the brighter X-ray sources in these fields are contained in the ChIcAGO catalog. Several examples are mentioned in the descriptions of individual MSFRs below.

We have one target (RCW 120) in common with SFiNCs (Getman et al. 2017) and two targets (NGC 6334 and G333) for which a subset of MOXC2 ObsIDs were included in MOXC1; comparisons of these catalogs are mentioned briefly in the specific target descriptions of Section 4. A few other individual MOXC2 MSFRs have published Chandra point source catalogs. MOXC2 typically recovers all or most of these sources and goes on to find additional faint sources; again, details are given in Section 4.

Using the example of the massive filamentary infrared dark cloud (IRDC) known as "Nessie," Jackson et al. (2010) suggested that multiple star formation sites can form at regular intervals along a cylindrical IRDC via the "sausage instability." This instability was originally described by Chandrasekhar and Fermi (1953), and is essentially the analog of Jeans collapse in a self-gravitating sphere, applied now to a self-gravitating fluid cylinder. The dominant path for the intense star formation that results in MSFRs is now recognized to be such massive molecular filaments (Goodman et al. 2014)—structures up to 100 pc long formed by supersonic flows that compress the gas of the ISM into a web of dense, cylindrical structures criss-crossing GMCs (André et al. 2014). These pressurized "ridges" are crucibles for creating MSFRs; material flows down the filaments, feeding hot cores and clumps that eventually collapse to become massive stars and star clusters (Tackenberg et al. 2014).

Our closest example is the G352 GMC, a 90-pc-long massive filament made famous by Herschel  (Russeil et al. 2013). G352 hosts the two multi-MSFR complexes NGC 6357 and NGC 6334 (Persi & Tapia 2008); such complexes are known as "clusters of clusters" (e.g., Bastian et al. 2007; Elmegreen 2008). Each of these sports a string of MSFRs with multiple bubbles and degree-sized "bowls" filled with hot plasma (MOXC1); they are situated on opposite sides of G352's main filament.

NGC 6334 is the best nearby example we have of the ongoing transfiguration of such a massive molecular filament into stars; it is thought to contain ∼175 O–B3 stars (Russeil et al. 2012) distributed among several MSFRs. It hosts a large number of H ii regions of various sizes and evolutionary states, spread across >2° of the Galactic Plane (Russeil et al. 2016). Many of its most massive clusters line up along a dense molecular ridge with a complicated filament morphology and many subfilaments. In some of these subfilaments, densities are high enough that gravity should dominate the energetics and they should be collapsing to form stars (Russeil et al. 2013). This is substantiated by a recent IR study (Willis et al. 2013) that identifies 2283 Class I and II young stellar objects in NGC 6334; these sources tend to cluster along the subfilaments and extend many parsecs beyond the main ridge. As shown in MOXC1 and earlier studies, the three ACIS pointings along the main ridge capture several collapsing subfilaments and add a large number of Class III (diskless) pre-MS stars to the census of young stars populating those regions (Ezoe et al. 2006; Feigelson et al. 2009; Broos et al. 2013).

Recent maser parallax measurements by the VLBA BeSSeL Survey (Wu et al. 2014) and VERA (Chibueze et al. 2014) both yield a distance of ∼1.3 kpc for NGC 6334, which we adopt for MOXC2. Readers should note, however, that some recent papers (e.g., Russeil et al. 2016; Tigé et al. 2017) use a distance of ∼1.75 kpc.

NGC 6334 serves as a transition target for us, illustrating the changes in our ACIS data analysis code and procedures between the MOXC1 analysis (circa 2013) and the current analysis for MOXC2. The same three ObsIDs analyzed for MOXC1 are re-analyzed here, augmented by several surrounding archival ACIS-I and ACIS-S pointings and our recent Chandra Cycle 17 General Observer (GO) data on the western side of NGC 6334's main filament, including the MSFR GM 24 (GM 1–24 in SIMBAD; Tapia & Persi 2008; Tapia et al. 2009).

Our final 14-wide ACIS mosaic is shown in Figure 1. Chandra typically characterizes diskless young stellar populations, both older pre-MS stars distributed across MSFRs and younger sources still embedded in their natal clouds. That generalization appears to hold true for NGC 6334; widely distributed X-ray sources are seen across the entire ACIS mosaic, extending far from the main ridge of MSFRs typically studied in NGC 6334.

Zoom In Zoom Out Reset image size

The original ACIS pointings on the main ridge have been described at length (Ezoe et al. 2006; Feigelson et al. 2009) and were presented in MOXC1, so we do not discuss them further here, except to note that they are pervaded by diffuse X-ray emission (Figure 2). Comparing the MOXC1 and MOXC2 catalogs in this region (where the two epochs of analysis used the same data), we find that both recover the same brighter sources (≥12 net counts) at all off-axis angles. For faint sources, the catalogs differ in several ways. Sometimes one analysis split a group of counts into a close pair of sources while the other found a single source. Sometimes faint sources were deemed valid in one analysis but not the other; this is especially true at large (>7') off-axis angles. These results are expected, given the strong interdependencies between faint candidate sources, the iterative nature of source validation, and the many decisions made, both algorithmic and by the analyst during visual reviews. Close examination of source positions shows that the inter-ObsID alignment is improved in MOXC2. Given this and the algorithmic improvements made over the years between MOXC1 and MOXC2, we recommend the use of the MOXC2 source list.

Zoom In Zoom Out Reset image size

4.1.1. GUM 61

Continuing down the G352 molecular filament southwest of the main ridge in NGC 6334, we find deep archival ACIS data on the large H ii region GUM 61 (Russeil et al. 2016; Figure 3). This hosts a populous diffuse cluster, as shown by the large number of X-ray sources seen across the entire ACIS-I field in this pointing (Figure 1(a)).

Zoom In Zoom Out Reset image size

At the center of the cluster (Figure 3(b)), the piled ACIS source c1934 (CXOU J171946.16-360552.2) matches HDE 319703A, a spectroscopic binary of type O7V((f)) + O9.5V (Maíz Apellániz et al. 2016). A spectral fit to the pile-up corrected spectrum of c1934 (from ObsID 13436) requires two thermal plasma components, TBabs(apec1 + apec2), for an acceptable fit. This yields N_H = 1.1 × 10²² cm⁻², kT1 = 0.7 keV, kT2 = 1.9 keV, and L_X = 1.6 × 10³² erg s⁻¹. The hard thermal plasma component dominates the spectrum. In ObsID 12382, obtained 4 days after ObsID 13436, c1934 shows slightly different spectral parameters (N_H = 1.4 × 10²² cm⁻², kT1 = 0.6 keV, kT2 = 2.5 keV) and is brighter, with L_X = 2.0 × 10³² erg s⁻¹. The softer plasma component dominates the spectrum below 2 keV. A hard, variable X-ray spectrum and changing X-ray luminosity often indicate wind interactions in binary systems; this may be yet another CWB discovered by Chandra.

Nearby is HD 319703B, a visual binary of type O6V((f))z (Sana et al. 2014), matching ACIS source c1800 (CXOU J171945.05-360547.0). It has 432 net counts and a median energy of 1.3 keV. The spectral fit yields N_H = 1.4 × 10²² cm⁻², kT = 0.5 keV, and L_X = 9 × 10³¹ erg s⁻¹. Source c1800 has four close neighbors in the ACIS data. We also detect HD 319703D, spectral type O9.5:Vn (Sana et al. 2014), as ACIS source c2162 (CXOU J171948.94-360602.8), with 98 net counts and a median energy of 1.4 keV. For the spectral fit, we find N_H = 1.9 × 10²² cm⁻², kT = 0.6 keV, and L_X = 3 × 10³¹ erg s⁻¹.

There is prominent diffuse X-ray emission near these massive stars and at the southwest (open) edge of the IR bubble (Figure 3(a)), where Russeil et al. (2016) note outflow in the kinematics of this region. This is probably a good example of diffuse X-ray emission tracing hot gas from massive star feedback in an H ii region, flowing out of that region to enrich and heat the surrounding ISM. It might be compared to M17, a spectacular example of this phenomenon (Townsley et al. 2003, MOXC1).

4.1.2. Southwest Pointings

Extending the ACIS mosaic southwest along NGC 6334's natal molecular filament past GUM 61, our ACIS pointing named G350.776+0.831 (ObsID 18081) captures a complex network of subfilaments (Figure 4(a)) and known clumps of pre-MS stars (Russeil et al. 2013; Willis et al. 2013). It is centered on the base of a large "bowl" of bright 8 μm emission seen opening to the northwest in Figure 1(b); the main molecular filament forming NGC 6334's backbone makes up the southern edge of this 30'-diameter structure. This pointing shows substantial shadowing or displacement of hot plasma in a large oval region at the base of the bowl, but other parts of the bowl contain diffuse X-ray emission, as does the large elongated cavity on the southern side of the main filament, opening from Gum 61 to the southwest.

Zoom In Zoom Out Reset image size

The ACIS mosaic continues down the main filament to the isolated, embedded IR cluster GM 24 (Figures 4(b) and (c)). This very young, midsized cluster (∼250 M_⊙; Tapia et al. 2009) is ionized by an O8 star (Bik et al. 2006); it sits at the southern edge of the Spitzer bowl, separated from the cluster-of-clusters complex by more than 40'. GM 24 could have formed as a result of the bowl's dynamical influence on NGC 6334's main molecular filament. GM 24 appears to support a strong bipolar outflow that is carving cavities in the surrounding GMC (Tapia et al. 2009). ACIS finds a strong concentration of X-ray point sources coincident with GM 24, plus a large distributed population. Hot plasma clearly fills the western bowl above the main filament; this pointing also captures the edge of the southwestern elongated cavity and again finds diffuse X-ray emission there.

The piled ACIS source GM24_c717 (CXOU J171701.53-362100.6), near the center of GM 24, has a reduced extraction aperture due to a close, faint neighbor (c715) to the northwest. The position for the O8 star from Bik et al. (2006) is 3'' north of GM24_c717, so it is not clear that this X-ray source is the counterpart of the O8 star. There is no X-ray source close to the O8 star's position.

We fit GM24_c717's pile-up corrected spectrum (from ObsID 18876) with the same simple spectral model used for other massive stars. Unlike the GUM 61 sources described above, GM24_c717 has a very hard spectrum, with N_H = 4 × 10²² cm⁻², kT = 9 keV, and L_X = 2 × 10³² erg s⁻¹. Any additional soft thermal plasma component could easily be absorbed away by this high column, so the luminosity we report is a lower limit. In the second GM 24 data set taken 3 days later (ObsID 18082), the X-ray luminosity of GM24_c717 is lower by a factor of 20 but the spectrum remains very hard. Such extremely hard X-ray spectra and variability have certainly been seen in other young stars (e.g., W51 IRS2E in MOXC1 and two early-B stars in IRAS 20126+4104; Anderson et al. 2011), but this is still an exceptional source due to its high luminosity and extreme variability; it deserves further attention.

4.1.3. H ii 351.2+0.5

East of GUM 61 lies another prominent H ii region south of NGC 6334's main ridge, called H ii 351.2+0.5 by Russeil et al. (2016). Comparatively few X-ray point sources are found in this part of our ACIS mosaic, even though it is captured by three ACIS pointings, because it is only observed far off-axis in each of these pointings (Figure 5(a)). This region does serve to illustrate the usefulness of including off-axis CCDs in our mosaics; diffuse X-ray emission is seen throughout this area. It appears at the center of H ii 351.2+0.5 and at the bubble rim, perhaps indicating hot plasma flowing out its southeast side.

Zoom In Zoom Out Reset image size

The brightest X-ray source in this part of the ACIS mosaic is c6709 (CXOU J172052.63-360420.6); it is the X-ray counterpart to the massive binary HD 156738 (Sana et al. 2014), spectral type O6III((f)) (Sota et al. 2014), located near the center of H ii 351.2+0.5. Source c6709 has 948 net counts and a median energy of 1.2 keV. As for ACIS source c1934 above, c6709 requires a two-temperature fit, TBabs(apec1 + apec2), although the plasmas are much softer here. Fit parameters are N_H = 1.4 × 10²² cm⁻², kT1 = 0.2 keV, kT2 = 0.5 keV, and L_X = 5.5 × 10³² erg s⁻¹. These soft plasmas are consistent with instability-driven wind shocks in the individual stars.

4.1.4. SNR G351.2+0.1

W75N is part of the great concentration of star formation in the Cygnus X North molecular cloud complex (Reipurth & Schneider 2008). It contains several UCHiiRs ionized by early-B stars (Shepherd et al. 2004) and exhibits complicated dynamics from multiple molecular outflows (Shepherd et al. 2003; Davis et al. 2007). Persi et al. (2006) find at least 25 cluster members based on their IR-excess; we find that several of these have X-ray counterparts. Kumar et al. (2007), also using near- and mid-IR data, find a distributed population of young stars across the region, tracing the molecular filaments.

The single ACIS observation of W75N (Figure 6) shows a concentration of hard sources near the aimpoint. ACIS source c202 (CXOU J203836.47+423733.7) appears to be the X-ray counterpart to VLA_3 (Bb), an UCHiiR with Lyman continuum flux consistent with ionization by a B1 star (Shepherd et al. 2004). Source c202 has only five counts, but its median energy is 6.7 keV and all five counts have energies >6 keV. The spectral fit gives N_H ∼ 30 × 10²² cm⁻², kT ∼ 7 keV, and L_X ∼ 8 × 10³⁰ erg s⁻¹, but the fit parameters are not well-constrained. This is just a rough characterization given the limited counts, but there is no doubt that this is an extreme X-ray source, exceptionally hard and highly obscured; such emission is not expected from a single B1 star.

Zoom In Zoom Out Reset image size

Diffuse X-ray emission is seen on the west side of W75N despite the high obscuration and short (<30 ks) observation. It is brightest in the southwest corner of the ACIS-I image, outside the large dark arc in the Spitzer images that runs vertically at R.A. ∼ 20:38:15. In Figure 6(b) we have scaled the ACIS image to show the fainter diffuse emission surrounding the central cluster and filling apparent cavities in the Spitzer emission. Bright diffuse emission is seen far off-axis on the ACIS-S CCDs; this may be associated with the wider Cygnus X star-forming complex. The central region also shows extensive diffuse X-ray emission (Figure 6(c)) extending all the way down to the cluster center (Figure 6(d)).

The innocuous visual H ii region RCW 120 was catapulted to fame by the Spitzer GLIMPSE survey (Churchwell et al. 2006; Zavagno et al. 2007), which revealed it to be a canonical example of a single, dusty bubble hosting massive star formation (Deharveng & Zavagno 2008). It has since received much attention at long wavelengths, especially from Herschel, which detects several very young stars forming in dense, cold material around the edge of the bubble (Zavagno et al. 2010; Figueira et al. 2017). The bubble appears to be disrupted at its northern edge and is leaking photons at several points around its rim (Zavagno et al. 2007; Deharveng et al. 2009; Anderson et al. 2015). Such "leaky" H ii regions may demonstrate that ionizing radiation from massive stars extends well beyond classical H ii region boundaries, influencing star formation over a much wider area (Zavagno et al. 2007).

The Chandra observations of RCW 120 were presented in the SFiNCs (Star Formation in Nearby Clouds) survey (Getman et al. 2017). SFiNCs studies X-ray point sources in Chandra observations of nearby, typically lower-mass star-forming regions, so this is the only target that MOXC2 and SFiNCs have in common. We included RCW 120 in MOXC2 (Figure 7), primarily to facilitate study of the diffuse X-ray emission in this iconic Spitzer bubble, but this common target does give us the opportunity to compare our point source catalog with that of SFiNCs. This is a useful check, since the two projects use similar (but not identical) analysis methods. The SFiNCs analysis is a few years older than the MOXC2 analysis.

Zoom In Zoom Out Reset image size

In MOXC2, we find 999 X-ray sources in the main ACIS-I array observations of RCW 120 itself; SFiNCs found 678 X-ray sources in the same data. Matching the two catalogs, we find 604 formal matches, 395 MOCX2 sources not matched, and 74 SFiNCs sources not matched. One goal of MOXC2 is to detect fainter sources; thus we expect to have many unmatched MOXC2 sources. In several cases, either MOXC2 or SFiNCs found a pair of sources while the other catalog found a single source. Many of the unmatched sources (in both catalogs) are far off-axis and faint. For such sources, the source validity metric²⁵ can be sensitive to slight source position differences, algorithmic changes to the background calculation, and differences in the catalogs (neighboring sources affect the background calculation). Because source candidates are pruned by a simple threshold on this noisy validity metric, two reductions of the same data will not produce identical catalogs near the detection limit.

For our ACIS mosaic of the RCW 120 region (Figure 7(a)), we included neighboring observations of SNRs to get a wider context for RCW 120's diffuse X-ray emission: ObsID 6721 on CTB 37A (Pannuti et al. 2014) and ObsID 6692 on CTB 37B (Aharonian et al. 2008). Large masks were applied to these SNRs to keep them from dominating our smoothed diffuse X-ray image of the mosaic (Figure 7(b)). For completeness, we include the point sources we found in these fields in our X-ray source list for RCW 120.

Extensive diffuse X-ray emission is seen in RCW 120 (Figure 7(c)). It appears to trace hot plasma through a breach in the northeast side of the bubble and out into the surrounding ISM. The whole bubble apparently shadows background diffuse X-ray emission.

Clumps of X-ray point sources are shown in Figures 7(d)–(f). Panels (d) and (f) show groups of sources on the bubble rim; panel (e) shows the central cluster. The main ionizing source for RCW 120 is LSS 3959 (CD-28 11636 in SIMBAD), an O8V star (Zavagno et al. 2007). Its X-ray counterpart is ACIS source c1121 (CXOU J171220.84-382930.3, prominent in Figure 7(e)), with 682 net counts and a median energy of 1.4 keV. The spectral fit requires two thermal plasma components, TBabs(apec1 + apec2); the softer plasma dominates the spectrum. Parameters are N_H = 1.4 × 10²² cm⁻², kT1 = 0.5 keV, kT2 =2 keV, and L_X = 1.4 × 10³² erg s⁻¹. The harder plasma component is not expected from a single late-O star and is perhaps suggesting magnetic activity or binarity.

Like W75N described above, the IRAS 20126+4104 MSFR is another component of the extensive and diverse star formation complex in Cygnus (Reipurth & Schneider 2008). Its distance is well-determined by maser parallax (Moscadelli et al. 2011). IRAS 20126+4104 itself is a famous massive young stellar object (MYSO) with a prominent molecular outflow that has long been the subject of radio study (e.g., Cesaroni et al. 2005). Cesaroni et al. (2014) have imaged this MYSO at 1.4 mm, finding a distorted disk and associated jet rotating around a proto-B star. De Buizer et al. (2017) presents a recent SOFIA detection of the MYSO; he finds that SOFIA and Herschel  data show elongation in the direction of the outflow. The Chandra data were first presented by Anderson et al. (2011), with details on the X-ray counterparts to radio sources I20S and I20Var (Hofner et al. 2007). These authors note the detection of 150 point sources in the Chandra data and suggest that IRAS 20126+4104 is surrounded by a young stellar cluster. Those sources are cataloged in a recent paper by the same group (Montes et al. 2015), who determine that ∼80 of the 150 X-ray sources are likely members of this MSFR. They find that the surface density of X-ray sources peaks on the MYSO; thus IRAS 20126+4104 is a young cluster forming an early-B star at its center.

Our analysis of the same Chandra data found more faint point sources, as designed (Figure 8); we recovered all 150 X-ray sources reported by Montes et al. (2015). Many X-ray sources near the center of the cluster resolve into close pairs in our analysis (Figure 8(d)). Anderson et al. (2011) described the exceptional X-ray spectra of the ACIS counterparts to I20S and I20Var; we find similar spectral fit results. They concluded that I20S was an early-B MYSO and I20Var a likely IMPS. We concur with these findings and emphasize their implications: in some cases, stars that will become X-ray-quiet on the main sequence are extraordinarily hard, bright X-ray emitters in their youth, perhaps due to magnetic activity similar to that found in lower-mass pre-MS stars (Povich et al. 2011; Gregory et al. 2016).

Zoom In Zoom Out Reset image size

We find prominent diffuse X-ray emission in this ACIS field (Figures 8(b)–(d)), with a bright ridge running the length of the ACIS-I array to the east of the cluster (Figure 8(b)). The image center shows an arc of X-ray emission just to the east of the large pillar that hosts the embedded cluster (Figure 8(c)); between this and the eastern ridge is an absence of diffuse X-rays coincident with faint WISE 22 μm emission. This may be material shadowing the diffuse X-rays. Faint diffuse X-ray emission pervades the pillar as far west as the central cluster (Figure 8(d)); then drops off sharply. The western side of the ACIS-I field shows a complex mix of fainter diffuse X-ray emission anticoincident with bright WISE structures. This fades to only minimal diffuse X-ray emission on the off-axis ACIS-S CCDs. Again, shadowing or displacement of hot gas is likely here.

The vicinity of the famous giant H ii region W31 has several Chandra/ACIS observations of a variety of targets at a wide range of distances. As for other MOXC2 targets, we have combined them into a wide-field mosaic (Figures 9(a) and (b)) to sample the X-ray point source populations and diffuse emission present in large Galactic Plane fields.

Zoom In Zoom Out Reset image size

Deharveng et al. (2015) showed that the northern component of the W31 complex (W31N, which they call G010.32-0.15) is actually a foreground MSFR unassociated with the other major MSFRs along this sightline, at a distance of just 1.75 kpc. W31N (G10.3-0.1) is a bipolar H ii region ionized by an O5–O6 star (Bik et al. 2005) and containing an IR cluster; it is triggering a menagerie of second-generation massive stars and their UCHiiRs in the dense filament at its waist (Deharveng et al. 2015b; Dewangan et al. 2015).

At this modest distance, it makes a rich Chandra target; we find >50 X-ray point sources in the central cluster and 490 sources across the ACIS-I field centered on W31N. As demonstrated below, with these data we can study both the first-generation ionizing cluster and the onset of X-ray emission in massive stars just formed. Figure 9(c) shows that W31N also contains diffuse X-ray emission and strongly shadows (or displaces) the diffuse emission seen in the wider W31 field.

The original 6.5 ks ACIS-S observation (ObsID 10518) of W31N detected the ionizing O5 star (source ChI J180857-2004_2 in Anderson et al. 2014); these authors suggest that it could be a CWB. In our data (Figure 9(d)), this is source c1345 (CXOU J180859.12-200508.4), with 553 net counts and a median energy of 3.3 keV. A spectral fit gives N_H = 4.9 × 10²² cm⁻², kT = 3.8 keV, and L_X = 2.4 × 10³² erg s⁻¹. Once again, this high column could be hiding an additional soft plasma component, so this luminosity is a lower limit. We concur with Anderson et al. (2014); this hard spectrum suggests a CWB.

Our source c1447 (CXOU J180901.48-200507.5) is the MYSO (YSO #4) in clump C2 of Deharveng et al. (2015). It has 43 net counts, a median energy of 4.6 keV, and is variable; fit parameters are N_H = 21 × 10²² cm⁻², kT = 5.5 keV, and L_X = 5 × 10³¹ erg s⁻¹. This extremely high column explains the high median energy. The hard thermal plasma and variability demonstrate that dramatic X-ray emission turns on early in the formation process of at least some massive stars.

Near the center of our ACIS mosaic of the W31 complex (Figure 9(b)), the conspicuous MSFR not yet observed by Chandra is G10.2-0.3, a giant H ii region at a distance of 3.4 kpc (Blum et al. 2001). It contains four O stars and four MYSOs in a very young (0.6 Myr) cluster (Furness et al. 2010), several UCHiiRs (Ghosh et al. 1989), and XMM  source detections (Nebot Gómez-Morán et al. 2015); it would make an excellent Chandra target. Southwest of G10.2-0.3, our mosaic includes the piled-up soft gamma repeater SGR 1806-20; we have masked its bright X-ray halo of dust-scattered light (Kaplan et al. 2002) in our image of diffuse X-ray emission.

At the southeast corner of our ACIS mosaic, we have included a 9.9 ks ACIS-I observation (ObsID 10713) of the radio-bright SNR G9.95-0.81 (Brogan et al. 2006). This field also includes the large GLIMPSE bubble CN139, with a near kinematic distance of 4.3 kpc (Churchwell et al. 2007; Watson et al. 2009). Quite surprisingly (given the short observation), Figure 9(e) clearly shows diffuse X-ray emission coincident with this bubble, somewhat offset to the northwest from the center of the radio SNR. Spectral analysis to determine the absorbing column to this X-ray emission should help determine whether it is really associated with the bubble.

Through further serendipity, the globular cluster 2MASS GC02 is captured far off-axis on the S3 CCD in ObsID 10713, visible (by zooming) at the bottom of the Spitzer/IRAC 8 μm image in Figure 9(b). ACIS source c2200 (CXOU J180936.55-204645.1) is centered on 2MASS GC02. It has 38 net counts and a median energy of 4.0 keV. It is most likely a composite of emission from multiple members of the globular cluster, but its properties might help to establish the feasibility for a longer ACIS observation of this target.

Starting with a 20 ks observation of the MSFR IRAS 19410+2336, we built a 4-pointing ACIS-S mosaic of overlapping short observations from archival Chandra data in this region (Figures 10(a) and (b)). This is the shallowest mosaic in MOXC2 (just 2–20 ks). This mosaic also features a second MSFR, NGC 6823, presumably not associated with IRAS 19410+2336, although at a similar distance. This is the first of three examples of ACIS mosaics in MOXC2 that feature two unrelated MSFRs that just happen to be projected near each other on the sky.

Zoom In Zoom Out Reset image size

IRAS 19410+2336 is a young, compact, embedded MSFR made up of two main star-forming clumps (Beuther et al. 2002; Rodón et al. 2012), both exhibiting masers (Beuther et al. 2002) and a large number of outflows (Beuther et al. 2003). The Chandra observation of IRAS 19410+2336 (ObsID 1868, 20 ks) was analyzed by Beuther et al. (2002), who tabulated 13 X-ray sources in the vicinity of the two clumps (Figure 10(d)). They find these sources to have hard X-ray spectra and to be heavily obscured, with fairly high intrinsic luminosities. They conclude that the X-ray sources are mainly IMPS; as noted above, subsequent work has confirmed that such sources can be strong X-ray emitters (Povich et al. 2011; Gregory et al. 2016).

Beuther et al. (2002) included the full 20 ks exposure in their analysis; our standard processing of this data set leaves just 8.7 ks of usable data on the aimpoint CCD S3 and on S1, the other backside-illuminated device, due to space weather resulting in high backgrounds for these chips. The ACIS frontside-illuminated devices in operation for this observation (I2, I3, S2, and S4) were unaffected by these background flares, so they retain the full integration time in our analysis. Unfortunately, the two main embedded clumps in IRAS 19410+2336 are imaged on CCD S3, so we report source detections in only 8.7 ks of data for these clumps. As we will see below, this is comparable to the ACIS integration time we have on the second MSFR imaged in this multi-pointing ACIS-S mosaic, NGC 6823.

As shown in Figures 10(c) and (d), we find X-ray sources in the same regions as Beuther et al. (2002), although not always the same sources. The southern clump is well-populated with X-ray sources, but the northern one is not; rather, a second group of X-ray sources sits to the east of the northern clump. Diffuse X-ray emission appears between the two groups of X-ray sources.

About 26' south of IRAS 19410+2336 we find the other MSFR in our mosaic, NGC 6823 (Prato et al. 2008), captured on two short ACIS-S observations (Figure 10(e)). It features a tight grouping of massive stars at its center (Shi & Hu 1999; Riaz et al. 2012), with pre-MS stars distributed out to the cluster radius of 162 (Kharchenko et al. 2005). Its earliest star is HDE 344784 A, with a spectral type of O6.5 V((f))z (Arias et al. 2016). Anderson et al. (2014) detect its X-ray counterpart as their source ChI J194310+2318_5. In our analysis, this is Chandra source c228 (CXOU J194310.96+231745.5), with 72 net counts and a median energy of 1.1 keV. Our simple thermal plasma fit yields N_H = 0.9 × 10²² cm⁻², kT = 0.6 keV, and L_X = 1.9 × 10³² erg s⁻¹.

Anderson et al. (2014) list several other X-ray sources in this region; we also find many other sources. Despite having just 7.9 ks total integration on this target and with the cluster center placed 8' off-axis on the 6 ks ObsID, ACIS still clearly detects this MSFR. Several of our X-ray sources are counterparts to young stellar objects tabulated in Riaz et al. (2012).

In the fourth pointing of this mosaic (ObsID 8164, 2.7 ks), we have two sources at the aimpoint: the piled-up source c19 (CXOU J194154.60+225112.3) and its close neighbor c20 (CXOU J194154.60+225112.8). Due to crowding, extraction apertures are reduced to 45% for c19 (25 net counts, median energy 3.3 keV) and just 36% for c20 (10 net counts, median energy 2.4 keV). With so few counts, we cannot correct c19 for pile-up. Rough spectral fits to these sources show them both to be very hard (kT ∼ 10 keV) but with very different absorbing columns and absorption-corrected total-band fluxes: N_H = 4 ×10²² cm⁻², F_X ∼ 8 × 10⁻¹³ erg s⁻¹ cm⁻² for c19, and N_H =1 × 10²² cm⁻², F_X ∼ 2 × 10⁻¹³ erg s⁻¹ cm⁻² for c20. Anderson et al. (2014) find a single X-ray source (ChI J194152+2251_2) at this location (a blend of our c19 and c20), using this same data set; they fit it with a power law with N_H = 0.9 × 10²² cm⁻², slope Γ ∼ 0.4, and absorption-corrected total-band flux F_X ∼ 7 ×10⁻¹³erg s⁻¹ cm⁻². They noted that the near-IR image of this source appears extended, consistent with a blend of sources.

Despite the very short integration times, diffuse X-ray emission is seen throughout this mosaic. It is brightest around NGC 6823 (Figure 10(e)), likely indicating a mix of unresolved pre-MS stars in the cluster and perhaps hot plasma from the winds of the O6 star. We see highly structured diffuse emission suffusing the deeply embedded IRAS 19410+2336 (Figure 10(d)) and pervading the surrounding field (Figure 10(c)). The wider field contains SNR G59.5+0.1 (Xu & Wang 2012), which may also be contributing to diffuse X-ray emission in our mosaic.

In a recent Chandra GO+GTO program, we targeted the newly formed MSFRs W42 and W33, both offering rich samples of very young (<0.2 Myr) massive stars ionizing UCHiiRs, to study the emergence of X-ray emission in these objects and their influence on the early evolution of MSFRs. At the same time we explored two of the Galaxy's extremely rare YMCs, with >10⁴ M_⊙ (Portegies Zwart et al. 2010), Red Supergiant Cluster 1 (RSGC1), age 10 Myr (Froebrich & Scholz 2013), and Cl 1813-178, age 4.5 Myr (Messineo et al. 2011). YMCs persist for >10 Myr because of their extraordinarily large masses; they allow us to explore changes in X-ray emission from massive stars as they approach the ends of their lives, along with the neutron stars and SNRs that they leave behind.

Here our ACIS-I mosaic (Figure 11) captures both the nearby MSFR W42 (Blum et al. 2000) and the unrelated background YMC RSGC1 (Figer et al. 2006), along with two pulsars and their PWNe (Gotthelf & Halpern 2008). We recover all X-ray sources tabulated by Gotthelf & Halpern (2008). We find a dense cluster of X-ray sources at the center of W42 (Figure 11(c)).

Zoom In Zoom Out Reset image size

W42 hosts at least 4 UCHiiRs (Urquhart et al. 2013) and contains more than 500 candidate young stellar objects based on IR-excesses (Dewangan et al. 2015a). The brightest near-IR source, W42 No. 1, is an O5–O6 star (Blum et al. 2000). W42's edge-on orientation illustrates the connection between molecular filaments and MSFRs: part of a massive filament has collapsed to form the MSFR that is now blowing a bipolar bubble normal to the filament's long axis (Deharveng et al. 2010; Dewangan et al. 2015a). We find few X-ray sources along this filament south of the W42 cluster.

The bright yellow X-ray source slightly northeast of field center in Figure 11(c) is c1575 (CXOU J183815.27-064758.8), the counterpart to W42 No. 1 (Blum et al. 2000). It has 148 net counts and a median energy of 2.4 keV; the spectral fit gives N_H = 2.3 × 10²² cm⁻², kT = 3.3 keV, and L_X = 5.5 × 10³¹ erg s⁻¹. Again this thermal plasma temperature is too hard for a single mid-O star, although the luminosity is modest compared to many CWBs. Alternatively, the W42 complex may be more distant than the 2.2 kpc (Blum et al. 2000) we have assumed; Dewangan et al. (2015a) use a distance of 3.8 kpc. That distance increases the X-ray luminosity of source c1575 to L_X = 1.6 × 10³² erg s⁻¹.

RSGC1 (Figer et al. 2006) is a 10 Myr old obscured cluster (Davies et al. 2008; Froebrich & Scholz 2013), the fifth most massive Galactic YMC known (Richards et al. 2012). It has >200 massive stars, far more than the other known RSGCs (Froebrich & Scholz 2013), that form a compact cluster. The X-ray-bright pulsar AX J1838.0-0655 is thought to be a member of RSGC1 (Gotthelf & Halpern 2008). Despite almost 74 ks of ACIS-I integration, we still do not detect any of the red supergiants in RSGC1 (Table 1 in Figer et al. 2006). There are many X-ray sources in this region of the mosaic, however, and several are quite hard. These highly absorbed sources may well be members of RSGC1.

Gotthelf and Halpern (2008) analyzed ObsID 6719 and noted that their soft X-ray source #12 is the high-proper-motion spectroscopic binary HD 171999. Our source c747 (CXOU J183758.77-064822.2) is their source #12. Nearby in our data is the similarly soft X-ray source c738 (CXOU J183758.67-064826.0). ObsID 16673 was obtained 9.5 years after ObsID 6719. Using SIMBAD proper motions for HD 171999, we calculate that this source should have moved about −12 in R.A., −38 in decl., in the 9.5 years between ACIS observations. Separations between c747 and c738 are about −15 in R.A., −38 in decl. Thus we conclude that our sources c747 and c738 are in fact the same X-ray source, the counterpart to HD 171999.

Diffuse X-ray emission around the pulsars AX J1838.0-0655 and AX J1837.3-0652 (Gotthelf & Halpern 2008) extends far beyond the brightest regions that we masked and dominates our ACIS mosaic of this field (Figure 11(b)). The bright PWN around pulsar AX J1838.0-0655 (our piled source c989) may have been over-reconstructed into point sources by our machinery; thus the six sources surrounding c989 may be spurious.

Fainter X-ray emission pervades much of the wider field, filling bubbles and voids in the Spitzer images. The brightest unresolved emission at the center of the W42 cluster (Figure 11(c)) undoubtedly includes a substantial contribution from faint cluster members; spectral analysis will be required to establish if any of this emission is truly diffuse.

The other targets in our recent GO program to study MSFRs at a range of ages were the very young W33 complex and the older, more distant YMC Cl 1813-178. As we found in the ACIS mosaic of W42 and RSGC1, here again we have a superposition of unrelated MSFRs separated by just a few arcminutes on the sky (Figure 12). In this case, Cl 1813-178 is so big and so well-populated with X-ray sources that it will be a challenge to assign membership in the part of the field where it overlaps with the foreground W33 complex.

Zoom In Zoom Out Reset image size

The W33 MSFR consists of a chain of very young stellar clumps and sparse clusters. It appears that these clusters formed separately and nearly simultaneously (Beck et al. 1998); this may indicate that they formed at sausage-instability density enhancements along a massive molecular filament (Chandrasekhar & Fermi 1953; Jackson et al. 2010). The distance to W33 has recently been firmly established at 2.4 kpc via maser trigonometric parallax (Immer et al. 2013).

The ACIS-I aimpoint for our study of W33 was centered on W33 Main (Figure 13), which contains at least five mid-O stars (Beck et al. 1998) ionizing at least three distinct UCHiiRs, separated by dense material. We find a clear concentration of X-ray sources consistent with the cluster position as given by Immer et al. (2014) (Figure 13(b)), but no counterparts to the infrared sources IRS1, IRS2, or IRS3 from Dyck & Simon (1977), using SIMBAD positions.

Zoom In Zoom Out Reset image size

W33A, a deeply embedded MSFR that features a well-studied MYSO (Davies et al. 2010), is captured at the northeast corner of the ACIS-I field. W33A may be forming due to the intersection of two massive molecular filaments (Galván-Madrid et al. 2010). We find five faint X-ray sources in a clump at the location of W33A; all have median energies above 3 keV, indicating that they are highly obscured. If these sources are associated with star formation in W33A, they indicate either that X-ray emission turns on very early in the star formation process or that W33A has experienced multiple epochs of star formation, with today's MYSO and dusty cores representing current and future star formation, while the X-ray sources represent an older population.

The ACIS data cover four more clusters associated with H ii regions across the W33 complex: radio region W33B and IR clusters cl1, cl2, and Mercer 1 (Messineo et al. 2011). While all of these regions contain X-ray sources, none of them shows a clear central concentration indicative of an X-ray clump, as we saw in W33 Main and W33A.

The monolithic YMC Cl 1813-178 was discovered serendipitously (Messineo et al. 2008), just ∼13' west of W33 Main, but Messineo et al. (2011) estimate that it is much more distant (3.8–4.8 kpc) and hence unrelated to W33. Cl 1813-178 is one of fewer than 20 YMCs known in the Milky Way (Clark et al. 2013), with a mass of 10⁴ M_⊙ and many known massive stars (Messineo et al. 2011).

Helfand et al. (2007) analyzed the original 30 ks ACIS-I observation of this field (which targeted SNR G12.82-0.02) and cataloged 75 X-ray sources; we recover all of those sources in our analysis. Our combined (ACIS GTO + archival) 59 ks Chandra data set provides a wealth of candidate members of Cl 1813-178 (Figure 14), with 572 X-ray sources within the reported cluster radius of 35 (Messineo et al. 2011) and hundreds more beyond that.

Zoom In Zoom Out Reset image size

We have X-ray counterparts to at least 16 of the 25 massive stars listed in Messineo et al.'s (2011) Table 1. They find two Wolf-Rayet (WR) stars in this cluster: their #4 and #7, both of type WN7. The first of these, #4, is our source c807 (CXOU J181314.20-175343.4), with 348 net counts and a median energy of 3.6 keV; a spectral fit gives N_H = 8 × 10²² cm⁻², kT = 2.8 keV, and L_X = 1.2 × 10³³ erg s⁻¹. The second of these, #7, is our source c1281 (CXOU J181322.48-175350.2), with 68 net counts and a median energy of 2.8 keV. It is crowded with the faint source c1286 (CXOU J181322.56-175350.1), with 6 net counts and median energy 3.7 keV. A spectral fit to c1281 gives N_H = 4 × 10²² cm⁻², kT = 2.6 keV, and L_X = 1.3 × 10³² erg s⁻¹. Based on the high thermal plasma temperatures and X-ray luminosities for these WR stars, we concur with Messineo et al. that they are likely CWBs.

We also fit the X-ray spectrum of Messineo et al. (2011) source #5, which they find to be a late-O star but with an unusually hard X-ray spectrum. This is our source c1346 (CXOU J181323.71-175040.5), with 265 net counts and a median energy of 3.0 keV; fit parameters are N_H = 2.0 × 10²² cm⁻², kT > 10 keV, and L_X = 3.5 × 10³² erg s⁻¹. This spectrum is equally well fit by a power law (TBabs*pow in XSPEC) with N_H = 2.0 × 10²² cm⁻², slope Γ = 1.5, and L_X = 3.6 × 10³² erg s⁻¹. As Messineo et al. suggest, this source may be another CWB. Its spectrum is so hard that it might be compared to that found for source I20S in IRAS 20126+4104 (Anderson et al. 2011).

Our ACIS mosaic reaches approximately the same sensitivity in both W33 and Cl 1813-178; W33 is closer but more obscured. As shown in all figures in this section, diffuse X-ray emission pervades the entire W33 mosaic. As we saw in our W42 mosaic, the brightest diffuse emission here is associated with the PWN around a pulsar (PSR J1813-1749) and its SNR (G12.82-0.02; Funk et al. 2007; Helfand et al. 2007; Gotthelf & Halpern 2009). Some of our cataloged point sources that are superposed on this bright diffuse emission could be artifacts of image reconstruction; identifying counterparts to these X-ray sources at longer wavelengths is the best way to set aside such a concern. Fainter diffuse emission extends to the west across the entire Cl 1813-178 ACIS pointing. It appears at the W33 Main and Cl 1813-178 cluster centers and around the edges of the W33 ACIS-I field. It is faint but present on the easternmost ACIS-S CCD. The W33 MSFR probably shadows background diffuse emission across most of the W33 pointing.

NGC 7538 is a clumpy Outer Galaxy MSFR, made up of several separate star-forming sites with a range of ages (McCaughrean et al. 1991; Balog et al. 2004; Ojha et al. 2004), similar to W33. It hosts the O3V star IRS6 and the O9V star IRS5 (Puga et al. 2010), along with a variety of MYSOs that are well-studied in the radio (e.g., Beuther et al. 2017).

The Chandra data were analyzed recently as part of two multiwavelength studies (Mallick et al. 2014; Sharma et al. 2017), finding 182 and 190 X-ray point sources respectively, but neither of these papers gives full catalogs for their Chandra sources. Mallick et al. tabulate 27 X-ray sources with NIR counterparts in the IRS1–3 and IRS9 regions; they provide some X-ray properties for that subset of their X-ray detections. We recover all of these sources in our analysis and go on to find almost 500 X-ray sources and diffuse X-ray emission in this single 30 ks ACIS-I observation (Figures 15(a) and (b)).

Zoom In Zoom Out Reset image size

In fact we find clumps of X-ray sources associated with all the well-known star-forming regions in NGC 7538: IRS1–3, IRS4–8, IRS9, IRS11/NGC 7538S, and the globules described by Sharma et al. (2017). ACIS images of most of these regions are presented in Figures 15(c)–(f); the IRS9 region is omitted because it has few X-ray sources and the IRAC 8 μm image is dominated by the single bright source IRS9 itself. Sharma et al. (2017) give short summaries of these star-forming regions; they all harbor massive stars or MYSOs, many ionizing various types of H ii regions. Here we briefly describe X-ray counterparts to the famous IRS sources.

IRS1–3 (Figure 15(c)). We find many X-ray sources in this young MSFR. Chandra source c343 (CXOU J231345.32+612810.9) is the counterpart to the MYSO IRS1; it has just five X-ray events but all have energies >3 keV, with a median energy of 4.4 keV. This indicates that the X-ray source is highly obscured, as expected for this massive protostar. Chandra source c351 (CXOU J231345.47+612820.1) is the counterpart to IRS2, an O9.5V star; its eight counts have a median energy of 1.6 keV. IRS3 has no X-ray counterpart in our data.

IRS4–8 (Figure 15(d)). This is the main star-forming complex in NGC 7538; it is well-populated with X-ray sources. Chandra source cc228 (CXOU J231330.22+613010.7) is the counterpart to the O9V star IRS5, with nine counts and a median energy of 1.4 keV, similar to IRS2. Chandra source c222 (CXOU J231334.40+613014.6) is the counterpart to the O3V star IRS6, with 400 counts and a median energy of 1.4 keV. A spectral fit yields N_H = 1.7 × 10²² cm⁻², kT = 0.8 keV, and L_X = 8 × 10³² erg s⁻¹. Sources IRS4, IRS7, and IRS8 have no X-ray counterparts.

IRS9. The faint Chandra source c460 (CXOU J231401.78+612718.8) is the counterpart to the MYSO IRS9. It has just eight counts and an extreme median energy of 5.0 keV; event energies span the wide range of 1.5–7.5 keV. Sandell et al. (2005) find multiple outflows in this region, suggesting an embedded cluster rather than a single MYSO, but we find just a few X-ray sources here.

IRS11/NGC 7538S (Figure 15(e)). Chandra source c322 (CXOU J231343.92+612657.6) may be a match to the MYSO IRS11. It has 17 counts and a median energy of 4.0 keV, with event energies ranging over 2.6–5.0 keV. The spectral fit is not well-constrained, but the source is certainly highly obscured, with N_H > 20 × 10²² cm⁻², kT ∼ 1 keV plasma with L_X ∼ 1 × 10³³ erg s⁻¹. We do not detect an X-ray counterpart to NGC 7538S.

Globules (Figure 15(f)) as defined by Sharma et al. (2017). These host small groups of X-ray sources at the eastern edge of NGC 7538. Sharma et al. noted that they resemble bright-rimmed clouds and point back to the O3 star IRS6.

The brightest diffuse X-ray emission in the ACIS observation is far off-axis on the ACIS-S chips. On-axis, a diffuse enhancement is coincident with the central cavity around the cluster. Faint patches of diffuse X-rays extend north and south to the edges of the ACIS-I field, anticoincident with the WISE mid-IR emission. Luisi et al. (2016) find ionizing radiation leaking past the primary photodissociation region (PDR) to a second PDR northeast of the main bubble, demonstrating the influence of MSFRs on the surrounding ISM. Diffuse X-ray emission extends at least as far as the second PDR to the northeast. It also spreads to the south and southwest, past the main NGC 7538 bubble seen in the WISE data, once again demonstrating that the hot ISM around MSFRs is more complex and extensive than might be predicted from IR data alone.

The elongated G333 GMC hosts several obscured MSFRs, spread over 80 pc across its long axis (e.g., Russeil et al. 2005; Nguyên et al. 2015). G333 has formed several massive young clumps and clusters, all at nearly the same time and with nearly regular spacing along the cloud. These clusters are too young to have hosted supernovae and their wind-blown bubbles do not overlap significantly, so it is unlikely that we are witnessing a "sequential triggering" process (Elmegreen & Lada 1977).

G333 may be a more evolved version of the 80-pc-long massive molecular filament and IRDC "Nessie" (Jackson et al. 2010). Near-IR dust extinction maps (Kainulainen et al. 2011) show that IRDCs are often surrounded by molecular clouds with >10 times more mass; thus a massive (but not extraordinary) IRDC of 10⁴M_⊙ can easily represent just the densest part of a typical 10⁵M_⊙ GMC. So G333 may be a rare example of a very massive IRDC with an unusually favorable sky orientation that has nearly completely collapsed into regularly spaced sites of coeval massive star formation. Soon supernovae in these embedded clusters will destroy the last vestiges of the original IRDC and start blowing superbubbles to feed the hot ISM, as we see in Carina (Townsley et al. 2011a, 2011b) and NGC 6357 (MOXC1). But since the underlying elongated structure still remains in G333, we can study the transformation of a massive molecular filament into a cluster of clusters. Similar to G352 described above but more distant (so the entire GMC can be captured in a few ACIS-I pointings), G333 provides an important observational and evolutionary link between giant molecular filaments such as Nessie and cluster-of-clusters star-forming complexes such as Carina.

This study required four 60 ks ACIS-I pointings to mosaic MSFRs across the entire G333 GMC. We included additional archival Chandra data on surrounding pointings to create a wide-field, high-resolution X-ray mosaic of the entire G333 GMC (Figure 16). The young SNR RCW 103 lies just off the southwest edge of our G333 mosaic. We have omitted its ACIS data from our mosaic because there is a recent analysis of those data (Frank et al. 2015). It is also extremely bright and swamps the fainter diffuse X-ray emission in our mosaic. As in MOXC1, we continue to assume a distance of 2.6 kpc to G333 (Figuerêdo et al. 2005; Roman-Lopes et al. 2009). Recent papers (e.g., Lo et al. 2015; Nguyên et al. 2015) use a distance of ∼3.6 kpc to this GMC; assuming this larger distance would increase our X-ray source luminosities by nearly a factor of two. Below we describe each of our four main ACIS-I pointings on G333 MSFRs.

Zoom In Zoom Out Reset image size

4.10.1. G333.6-0.2

The most prominent MSFR in G333 is the bipolar bubble and giant H ii region G333.6-0.2 (Figure 17(a)). We presented a single-ObsID (60.1 ks) ACIS-I study of this region in MOXC1. The MOXC2 analysis used slightly more data (61.6 ks). In both epochs of analysis, ACIS reveals more than 100 X-ray sources in a central compact cluster powering the giant H ii region, a distributed population of over 500 more point sources across this ACIS-I pointing, and soft diffuse X-ray emission tracing hot plasma apparently leaking from the embedded cluster (Figure 17(b)). As we found for the main ridge in NGC 6334, MOXC1 and MOXC2 catalogs are the same for brighter sources but find different sources at the faint limit. As before, we favor the MOXC2 results because they reflect improvements in our analysis methodologies over the years.

Zoom In Zoom Out Reset image size

4.10.2. G333.3-0.4

This pointing (Figure 18) contrasts an IR-bright MSFR with the only part of the G333 GMC that is not bright at 24 μm. The upper half of Figure 18(a) shows this dark bay; we see no diffuse X-ray emission from this region. Water masers are found even in the IR-dark part of the field, indicating ongoing massive star formation (Breen et al. 2007). A near-IR study of G333.3-0.4 (Roman-Lopes et al. 2003) found 41 reddened massive stars in a 1' cluster behind $\langle {A}_{V}\rangle \sim 28$ mag, with spectral types ∼O5.5–B5. We find X-ray counterparts to at least 13 of those massive stars, sometimes resolving two or three X-ray sources at the location of the IR source.

Zoom In Zoom Out Reset image size

The MSFR G333.3-0.4 is associated with IRAS 16177-5018, powered by an O3If⁺ star at 2.6 ± 0.7 kpc (Roman-Lopes et al. 2009). Our X-ray counterpart to that star is c5623 (CXOU J162131.60-502508.3), with 13 net counts and a median energy of 3.2 keV; it has two close neighbors (each <12 away). With so few counts, spectral fit parameters for c5623 are not well-constrained, but they are roughly N_H ∼ 13 × 10²² cm⁻², kT ∼ 0.9 keV, and L_X ∼ 4 × 10³² erg s⁻¹.

In Figure 18(a) we do see diffuse emission from the G333.3-0.4 cluster and around the edges of this field, sometimes filling Spitzer bubbles. The center of the cluster (Figure 18(b)) contains a substantial number of X-ray sources; the overdensity of hard events not captured by detected sources implies that there are many more X-ray sources just below the detection limit. Diffuse X-ray emission is seen around the cluster but is fainter in the center, perhaps shadowed by the high absorbing column in this region. The X-ray sources are typically hard and faint, again probably due to high absorption. No other major grouping of X-ray sources is found in this pointing.

4.10.3. G333.1-0.4

This pointing (Figure 19) captures two powerful H ii regions: G333.12-0.45 and G333.03-0.45 (GAL 333.1-00.4 and GAL 333.0-00.4 in SIMBAD). A near-IR study of G333.12-0.45 also finds a cluster distance of 2.6 kpc, a range of extinctions (A_V ∼ 12–32 mag), several MYSOs with spectral types O4–B4, and a total cluster mass of >10³M_⊙ (Figuerêdo et al. 2005), but no source positions are provided. G333.12-0.45 shows direct evidence for large-scale infall (Lo et al. 2015), implying that it is very young.

Zoom In Zoom Out Reset image size

The grouping of X-ray sources closest to G333.12-0.45 is shown in Figure 19(b), in the northeast region of this pointing. Bright diffuse X-ray emission is seen here. This region is several arcminutes off-axis, where ACIS PSFs are large and sensitivity is reduced; thus point sources are typically faint. The diffuse emission may be due partially to unresolved pre-MS stars from this obscured cluster.

Another group of X-ray sources is shown in Figure 19(c), in the eastern part of this ACIS pointing. Faint diffuse X-ray emission pervades the field. There is an UCHiiR candidate here (Mookerjea et al. 2004). These X-ray sources may be part of GLIMPSE IR cluster #78 (Mercer et al. 2005).

The H ii region G333.03-0.45 is not as well-studied as G333.12-0.45 but shows a large number of dust cores and UCHiiR candidates (Mookerjea et al. 2004). It is as bright in Lyman continuum as the other G333 MSFRs (Conti & Crowther 2004). G333.03-0.45 is at the center of this ACIS pointing. It shows diffuse X-ray emission distributed throughout a fairly loose grouping of point sources (Figure 19(d)).

4.10.4. RCW 106

The final pointing (Figure 20) in our G333 mosaic of MSFRs again shows diffuse X-ray emission at the periphery and in a Spitzer bubble, with strong shadowing at the center (Figure 20(a)). We find two main clumps of X-ray sources. The first is on the northeast side of the pointing (Figure 20(b)). Two bright X-ray sources are found here: c2691 (CXOU J162026.75-505417.1; 398 net counts, median energy 1.7 keV) and c2412 (CXOU J162022.75-505420.4; 110 net counts, median energy 4.2 keV, variable). A spectral fit to c2691 requires two thermal plasma components, with N_H = 2.2 × 10²² cm⁻², kT1 = 0.5 keV, kT2 > 3 keV, and L_X = 9.0 × 10³² erg s⁻¹. The harder source, c2412, can be fit with a one-temperature thermal plasma, although that temperature is so high that it is not well-constrained by the ACIS data: N_H = 10 × 10²² cm⁻², kT > 10 keV, and L_X = 1 × 10³² erg s⁻¹.

Zoom In Zoom Out Reset image size

The second clump of X-ray sources is on the southwest side of the pointing, in a prominent Spitzer bubble (Figure 20(c)). These sources may be associated with GLIMPSE IR cluster #76 (Mercer et al. 2005).

4.10.5. Archival Pointings

We included several archival Chandra data sets in the G333 mosaic, to broaden the field and provide a wider context for the diffuse X-ray emission. ObsIDs 8141, 8161, 9602, and 10507 were from the ChIcAGO project (Anderson et al. 2014). ObsID 10929 re-observed a bright ChIcAGO source, which was determined to be the magnetar PSR J1622-4950 (Levin et al. 2010; Anderson et al. 2012). This is our source c8972 (CXOU J162244.91-495052.8); it is piled up in ObsID 8161. We find prominent diffuse X-ray emission surrounding the magnetar (Figure 21(a)). This could be the X-ray signature of its SNR; Anderson et al. (2012) found a radio arc just north of this region (G333.9+0.0) and suggested that it might be an SNR associated with PSR J1622-4950. Our diffuse X-ray emission could be tracing hot plasma filling that radio-bright partial shell. Additionally, Anderson et al. showed PSR J1622-4950 to be coincident with a potentially nonthermal diffuse radio source ("Source A") that may be a PWN associated with the magnetar. Since the diffuse X-ray emission is concentrated in the vicinity of PSR J1622-4950 and the diffuse radio source, this adds further support for the existence of a magnetar wind nebula.

Zoom In Zoom Out Reset image size

ObsID 5471 targeted IGR J16195-4945, which is a likely high-mass X-ray binary (Tomsick et al. 2006). This is our source c439 (CXOU J161932.20-494430.6); it is piled up, but there are too few counts to perform pile-up correction. There is bright diffuse X-ray emission in this region (Figure 21(b)). Pandey et al. (2006) note extended radio emission in the vicinity of IGR J16195-4945, so perhaps we are once again seeing X-ray emission from a SNR.

AFGL 2591 is an embedded MYSO with prominent outflows and maser emission, surrounded by a small cluster (Trinidad et al. 2003); it is seen in the direction of the Cygnus X star-forming complex (Reipurth & Schneider 2008), but it is now believed to be substantially behind Cygnus X (at 3.33 kpc), from parallax of its water masers (Rygl et al. 2012). The radio-bright source VLA-3 provides most of the IR emission and is thought to be a late-O-type MYSO (Sanna et al. 2012).

Parkin et al. (2009) analyzed the Chandra data on AFGL 2591, finding 62 sources on the S3 CCD; our analysis of the same data (Figure 22) yields 152 sources on S3 and 288 sources in the entire observation. In the vicinity of VLA-3, Parkin et al. find four X-ray sources but do not detect the MYSO. We recover those four X-ray sources and add several more in our analysis (Figure 22(d)), but again we do not detect VLA-3. Parkin et al. interpreted a small excess of counts close to VLA-3 as possible diffuse emission from the winds of the proto-O star. We find a few of those counts to be consistent with a point source east of VLA-3; the remaining small number of counts could come from other point sources (including VLA-3) fainter than our detection limit, from the background, or from other sources of diffuse emission. Firmly establishing diffuse X-ray emission from VLA-3 wind interactions would require much more Chandra data.

Zoom In Zoom Out Reset image size

Extensive diffuse X-ray emission is seen in this ACIS field, including off-axis on the I2 and I3 CCDs. Figure 22(c) demonstrates that diffuse emission surrounds AFGL 2591; Figure 22(d) shows that faint structures extend down to the embedded core. Some of this diffuse emission could be associated with the foreground Cygnus X complex; future X-ray spectral fitting to infer the absorption across the field may help us locate it along the line of sight.

The first compact IRDC targeted by Chandra was G014.225-00.506 (Povich & Whitney 2010), part of a large molecular cloud complex associated with the M17 giant H ii region known as the M17 Southwest Extension (Elmegreen et al. 1979), located at a distance of 2.0 kpc (Xu et al. 2011). Using the same analysis methods as those presented here, we cataloged 840 X-ray sources in a 99 ks ACIS-I observation of G014.225-00.506 (Povich et al. 2016). The second IRDC observed by Chandra was the more well-studied object G34.4+0.23 (Rathborne et al. 2005; Shepherd et al. 2007), observed with ACIS-I for just 63 ks under the assumption that it lies at a maser parallax distance of 1.56 kpc (Kurayama et al. 2011), substantially closer than G014.225-00.506. This distance was soon questioned, however (Foster et al. 2012), and recent papers (Foster et al. 2014; Xu et al. 2016) favor the kinematic distance of around 3.7 kpc (Rathborne et al. 2005).

This Chandra observation (Figure 23) reveals few X-ray sources along the G34.4+0.23 IRDC compared to G014.225-00.506 (Povich et al. 2016), again suggesting that the kinematic distance is correct and that the 63 ks ACIS exposure was too short to detect much of the young stellar population embedded in this IRDC. Nevertheless, we do find a large number of X-ray point sources in this field (Figure 23(c)). Some will be foreground and background contaminants, but many hundreds are likely pre-MS stars associated in some way with the star formation going on in this region.

Zoom In Zoom Out Reset image size

We have X-ray counterparts to a few of the young stellar objects listed in Shepherd et al. (2007). In particular, we find three faint, very hard X-ray sources in close proximity to source #11 from Shepherd et al., at the center of the UCHiiR (Figure 23(d)): c284 (CXOU J185318.71+012449.1, eight net counts, median energy 4.1 keV); c283 (CXOU J185318.71+012446.3, six net counts, median energy 4.5 keV); cc554 (CXOU J185318.54+012445.2, four net counts, median energy 4.2 keV).

Foster et al. (2014) found a distributed population of low-mass protostars associated with the IRDC but located outside the densest parts of the filament. Perhaps our X-ray sources sample an even older and more distributed stellar population. Further multiwavelength analysis is necessary to understand this X-ray source population and the history of star formation across this region.

The G34.4+0.23 Chandra mosaic shows abundant diffuse X-ray emission. A small patch at field center may contain unresolved point sources from the central proto-cluster (Figure 23(d)). This is largely surrounded by dark regions, perhaps due to shadowing by the IRDC and surrounding molecular material. Diffuse emission is seen again at the field edges, especially at the east and west corners of the ACIS-I image. As often demonstrated in other MOXC2 targets, the off-axis ACIS-S CCDs hint at more diffuse X-ray structures far off-axis. This emission is anticoincident with moderately bright Spitzer 8 μm emission to the southeast of G34.4+0.23.

The more developed MSFR G34.26+0.15 was partially imaged off-axis in this observation, at the southern corner of the ACIS-I array (Figure 23(b)). A few X-ray point sources are found there (likely limited by low sensitivity and source confusion), and the region is surrounded by diffuse X-ray emission. A Chandra observation centered on this MSFR would probably reveal a rich young stellar population and an interesting network of hot plasma from massive star feedback in G34.26+0.15.

We finish our tour of MOXC2 MSFRs with Westerlund 1 (Wd1), the most massive YMC known in the Galaxy (Andersen et al. 2017). It is a monolithic cluster, not part of a complex of multiple MSFRs, as we have seen in many MOXC2 targets. The center of Wd1 is known to be elongated (Gennaro et al. 2011), but recent work (Gennaro et al. 2017) finds that its massive stars are not strongly concentrated toward the center (mass-segregated); these authors conclude that Wd1 has changed little in size or density since its formation. Its combination of mass and age explains its uniquely large population of evolved massive stars, including red supergiants, yellow hypergiants, and WRs; Wd1 may have weathered ∼100 supernovae already (Clark et al. 2008). Twelve WRs (Skinner et al. 2006) and a magnetar (Muno et al. 2006) were among the 241 point sources detected in the original 57 ks ACIS-S observation (Clark et al. 2008).

We recover all but six of the X-ray sources tabulated in Clark et al. (2008). We find many hundreds more X-ray sources and diffuse emission across our ACIS mosaic (Figure 24). The magnetar (CXOU J164710.2-455216) suffers from photon pile-up in these observations. Clark et al. (2008) presented X-ray spectral fits for the brightest ACIS sources, so we will not report further spectral fitting results here.

Zoom In Zoom Out Reset image size

Wd1 is four times as massive as NGC 3603 (featured in MOXC1) but eight times lower in central density (Figer 2008). This and the relatively short Chandra exposure on Wd1 means that our X-ray point source list suffers much less confusion than we found in NGC 3603 (MOXC1). After excising resolved point sources, the ACIS event data still show a substantial overdensity of counts at the center of Wd1 (Figure 24(c)); a longer Chandra observation would undoubtedly resolve out a large number of additional X-ray point sources in the cluster core.

The second pointing, to the south of Wd1 in our ACIS mosaic (ObsID 11836, 10 ks), is an ACIS-I GTO observation of 0FGL J1648.1-4606, part of a snapshot survey of unidentified Fermi/LAT GeV sources. PSR J1648-4611 in this field is listed as a non-detection in Table 2 of Kargaltsev et al. (2012). This source remains undetected in our analysis.

Diffuse X-ray emission from Wd1 was studied by Muno et al. (2006) (Chandra) and Kavanagh et al. (2011) (XMM). In our ACIS mosaic (Figure 24(b)), we see diffuse emission across much of the field, including the cluster center (Figure 24(c)). The unresolved X-ray emission at the northwest corner of the mosaic is due in large part to the bright LMXB GX340+0, located off the field. We have masked the brightest of this scattered light from our diffuse images. Northeast of Wd1, far off-axis on the S0 chip of ObsID 6283, we have another serendipitous detection of diffuse X-ray emission apparently associated with a bubble structure in the Spitzer data (Figure 24(d)).