High-Resolution Observations of the Molecular Clouds Associated with the Huge HII Region CTB 102
Brandon Marshall, Sung-ju Kang, C.R. Kerton, Youngsik Kim, Minho Choi, Miju Kang
DDraft version April 2, 2019
Typeset using L A TEX default style in AASTeX62
High-Resolution Observations of the Molecular Clouds Associated with the Huge H II Region CTB 102
Brandon Marshall, Sung-ju Kang, C. R. Kerton, Youngsik Kim,
2, 3
Minho Choi, and Miju Kang Iowa State UniversityDept. of Physics & Astronomy, 2323 Osborne Dr.Ames, IA 50011-3160, USA Korea Astronomy and Space Science Institute776, Daedeokdae-ro, Yuseong-guDaejeon, 34055, Republic of Korea Daejeon Observatory213-48, Gwahak-ro, Yuseong-gu, Daejeon, 34128, Republic of Korea (Received; Revised; Accepted)
Submitted to ApJABSTRACTWe report the first high-resolution (sub-arcminute) large-scale mapping CO and CO observationsof the molecular clouds associated with the giant outer Galaxy H II region CTB 102 (KR 1). Theseobservations were made using a newly commissioned receiver system on the 13.7-m radio telescopeat the Taeduk Radio Astronomy Observatory. Our observations show that the molecular clouds havea spatial extent of 60 ×
35 pc and a total mass of 10 . − . M (cid:12) . Infrared data from WISE and were used to identify and classify the YSO population associated with ongoing star formationactivity within the molecular clouds. We directly detect 18 class I/class II YSOs and six transition diskobjects. Moving away from the H II region, there is an age/class gradient consistent with sequentialstar formation. The infrared and molecular-line data were combined to estimate the star formationefficiency (SFE) of the entire cloud as well as the SFE for various sub-regions of the cloud. We findthat the overall SFE is between ∼ Keywords:
ISM: clouds — ISM: individual objects (CTB 102) — stars: pre-main sequence — HIIregions INTRODUCTIONIn this paper we present the first CO and CO high-resolution (sub-arcminute) large-scale mapping observationsof the molecular clouds associated with the CTB 102 H II region. These observations provide basic data about theclouds, such as their sizes and masses, and are combined with archival infrared data to explore ongoing star formationactivity associated with the region.CTB 102 is an enormous H II region/bubble, with an estimated size of 100 – 130 pc, located in the outer Galaxy(J2000: 21 h m s , +52 ◦ (cid:48) (cid:48)(cid:48) ) at a distance d = 4 . ∼ (cid:48) resolution at 1420 MHz as part of the Canadian Galactic Plane Survey (CGPSTaylor et al. 2003). Arvidsson et al. (2009) used H89 α radio recombination line observations to show that the extensiveradio continuum structure visible in these surveys was all part of a single object with the primary structure having Corresponding author: Sung-ju [email protected] a r X i v : . [ a s t r o - ph . GA ] A p r Marshall et al. V , LSR = − . ± .
05 km s − and most of the surrounding filaments having | V − V | (cid:46) − . They alsodetermined that the total Lyman continuum photon emission rate from the region was N Ly ≥ (4 . ± . × s − ,consistent with a single early-type O star or with a cluster of several late-type O stars.In spite of its size, CTB 102 has not been studied at optical wavelengths because of both its distance and the fact itis hidden behind an extensive local region of high extinction (Lynds 1962; Fitzgerald 1968; Simonson & van SomerenGreve 1976). At near- and mid-infrared wavelengths our best view of the region comes from the Wide-field InfraredSurvey Explorer ( WISE ; Wright et al. 2010) all-sky survey. A portion of the region was also observed by
Spitzer during its GLIMPSE360 warm-mission survey (Whitney et al. 2008). Comparable resolution data for molecular gasemission, which is essential for understanding star-formation activity associated with the H II region, was not obtaineduntil this study.In section 2 we detail the new CO and CO observations as well as the archival data that has been collected forour study. In section 3 the physical properties of the newly mapped molecular clouds are described, and the youngstellar object (YSO) content of the clouds is determined using infrared data. The two studies then are combined toinvestigate the star-formation efficiency of the region. Finally, we discuss our findings and present our conclusions insection 4 and section 5. OBSERVATIONS2.1. CO and CO Observations
The CTB 102 H II region was observed using the Second Quabbin Optical Imaging Array in Taeduck Radio Astron-omy Observatory (SEQUOIA-TRAO) receiver system on the TRAO 13.7-m radio telescope. The data were obtainedbetween November 2016 and March 2017, during the first observing season after the SEQUOIA receiver array wasrelocated from the Five College Radio Astronomy Observatory (FCRAO) and adjusted to the TRAO system. Ourdata represent not just the first look at the CTB 102 region at CO and CO, but also some of the first data obtainedby the SEQUOIA-TRAO system.We observed CO (115.271 GHz, J = 1 −
0) and CO (110.201 GHz, J = 1 −
0) lines simultaneously using anOn-The-Fly (OTF) observation mode in a 1 . ◦ × . ◦ × (cid:48) × (cid:48) grid, centered on l = 93 . ◦ b = 2 . ◦
73. The RMS is ∼ . − . This compares favorably with the velocity resolution and RMS ofthe Outer Galaxy Survey (OGS; Heyer et al. 1998) and the Galactic Ring Survey (GRS; Jackson et al. 2006) bothobtained at FCRAO: 0.98 km s − and 0.65 km s − respectively, each with RMS ∼ . Otftool and the GILDAS software package
Class . The pointing of theobservation was calibrated using the SiO ν = 1 , J = 2 − ∼ (cid:48)(cid:48) . The antenna temperaturewas corrected automatically for the effects of atmospheric attenuation with a standard chopper-wheel method.The new radome of TRAO was installed in January 2017 to shield the antenna and receiver from the weather andother exterior environmental factors. Since the new radome affects the beam efficiency ( η mb ), we applied two differentbeam efficiencies at 115 GHz, for data acquired before ( η mb = 0.51) and after ( η mb = 0.46) the new radome installation.The full width at half maximum (FWHM) of the beam at 115 GHz ( CO) is 45 (cid:48)(cid:48) and at 110 GHz ( CO) is 48 (cid:48)(cid:48) .The integrated CO cube is shown in Figure 1, with CO contours overlain. For reference, the location of themolecular clouds are also indicated on the CGPS 1420 MHz image of the region shown in Figure 2.2.2.
Archival Infrared DataWISE scanned the entire sky in four bands centered at 3.4, 4.6, 12, and 22 µ m with a 5 σ sensitivity of roughly16.6, 15.6, 11.3, and 8.0 magnitudes respectively (Wright et al. 2010). We used the Infrared Science Archive (IRSA)to retrieved all of the sources in the WISE- based AllWISE catalog within a 29 arcminute radius around the center ofthe CTB 102 region at l = 93 . ◦ b = 2 . ◦
81. No additional constraints were used in the search and we retrieved 14814sources. The AllWISE catalog also provides the nearest coincident source ( ≤ (cid:48)(cid:48) ) from the Two Micron All Sky Survey( Skrutskie et al. 2006) All-Sky data release. This provides data in the J, H, and K S bands with magnitudelimits at approximately 16.0, 15.0, and 14.5 magnitudes respectively (Skrutskie et al. 2006). ANALYSIS he Molecular Clouds of CTB 102 Figure 1.
Integrated CO map of CTB 102. Integration was between V LSR = −
71 km s − to −
53 km s − . White ellipsesdenote the four main subdivisions of the molecular cloud (1, 2a, 2b, and 2c) along with the smaller 1a region that has a highconcentration of YSOs. Cyan contours show the integrated CO emission. Five contour levels were generated starting 3 σ abovethe median background. The morphology of the molecular cloud is very similar in both the CO and CO maps.
The CTB 102 Molecular Clouds
We adopt the distance to CTB 102 of 4.3 kpc from Arvidsson et al. (2009). This is essentially a kinematic-baseddistance that accounts for known non-circular motions in the second quadrant of the Galaxy (Brand & Blitz 1993).Assuming this, we estimate the overall size of the molecular cloud associated with the region to be approximately60 ×
35 pc. Figure 1 shows the CO emission integrated over −
71 km s − < V LSR < −
53 km s − , and indicates theextent of four distinct subdivisions of the molecular cloud, which were identified from visual inspection of the datacube. The semi-major and semi-minor axes of region 1 measure approximately 14 × × × × × ×
16 pc) encompassing regions 2a, 2b, and 2c.The mass of the molecular clouds associated with CTB 102 can be measured by using the X CO factor to convertthe integrated CO intensity into an H column density, then integrating the column density over the cloud area.Column density was calculated using N (H ) = 2 . × (cid:90) T dV cm − , (1) Marshall et al.
Figure 2. b ) in K. where we have used the Milky Way average X CO = 2.3 × cm − (K km s − ) − , which is a robust value applicableto clouds found in a wide range of Galactic environments (Bolatto et al. 2013; Sz˝ucs et al. 2016; Gong et al. 2018).In addition to a statistical uncertainty of ± . CO is known to vary systematically withGalactic environment due to changes in both metallicity and local star formation rate, and it is important to considerhow significantly the environment around CTB 102 may differ from the Galactic average.X CO is known to increase with decreasing metallicity, and a very sharp upturn is observed for metallicities below0.5 Z (cid:12) (see e.g. Figure 9 in Bolatto et al. 2013). For R (cid:12) = 8 . R G ) of 9.7 kpc (9.3 kpc), so ∆ R G = R G − R (cid:12) ≈ . − .
05 dex/kpc (Balser et al. 2012); this results in an essentially negligible decrease in metallicity to0.9 Z (cid:12) at the distance of CTB 102. In contrast, a hypothetical far outer Galaxy molecular cloud at R G = 16 kpcwould have, using the same metallicity gradient, Z ≈ . (cid:12) , and the Galactic average X CO would clearly not beappropriate.The radiation environment of portions of the CTB 102 molecular clouds clearly deviate from the Galactic averageinterstellar radiation field (ISRF) due to the proximity of the H II region. To roughly quantify this, Cloudy 17.00(Ferland et al. 2017) was used to model the radiation field surrounding an O5 V star. The energy density of theradiation field incident on a molecular cloud located at a distance of 1, 3, and 10 pc is ∼ − .So, while an extremely strong ISRF is limited to the very close proximity of the O star, increases in the strength of he Molecular Clouds of CTB 102 II region. The effect onchanging the ISRF strength on X CO has been investigated via simulations. Sz˝ucs et al. (2016) models show that X CO is only weakly sensitive to the ISRF strength at Z ∼ Z (cid:12) (see their models (d), (e) and (j) corresponding to an ISRF ofstrength G = 1, 10, and 100). Clark & Glover (2015) investigated the effect of increased star formation rate, whichwas modeled as an increase in both the ISRF strength and the cosmic ray ionization rate. They found that X CO doesincrease with increased SFR, but, for clouds with properties probably closest to ours ( M ∼ M (cid:12) , n o ∼
100 cm − )the dependence on SFR is weak (a factor of a few increase over two orders of magnitude increase in SFR). Sz˝ucs et al.(2016) provide a succinct summary of the relationship between X CO and SFR: at best the increase is sub-linear, andincreased density and velocity dispersion within the molecular clouds can act to offset the increase associated withincreasing SFR.Elliptical annuli were used to define the extent of each cloud and to calculate an average background level. Amean molecular weight of 2.3 was used for the total mass calculation. The systematic uncertainties related to ISRFstrength and metallicity, which we think are minimal in the case of CTB 102, will tend to make our mass estimatesunderestimates. Results are shown in Table 1 along with the results of other mass calculation techniques described inthis subsection.We can also estimate the H column densities using the integrated map for CO. Following the procedure describedin Rohlfs & Wilson (2004), the CO column density is related to the integrated CO intensity by: N ( CO) = 7 . × (cid:90) T dV cm − . (2)In applying Equation 2 we are assuming that the CO emission is optically thin and that LTE applies. We alsoassume identical excitation temperatures for CO and CO of 10 K (Sz˝ucs et al. 2016; Simon et al. 2001). We adoptthe conversion factors CO / CO = 6 . × D GC (kpc) + 18 .
71 and CO / H = 8 × − from Milam et al. (2005)and Blake et al. (1987), respectively, assuming D GC = 10 . ∼ CO mass estimate to be a lowerlimit accurate to within a factor of a few (Simon et al. 2001; Arvidsson et al. 2009; Sz˝ucs et al. 2016).Finally, the viral mass of a molecular cloud is often a good estimate of the true mass of the cloud even when the gasis not truly in virial equilibrium. Sz˝ucs et al. (2016) found that, in the context of their numerical models of molecularclouds, the underlying assumption of this mass determination technique is that the observed line widths are dominatedby turbulent broadening rather than thermal broadening. The virial masses were determined using M vir = 1040 × R pc × σ v , (3)where σ v is the CO velocity dispersion in km s − , and R pc is the geometric mean size (Equation 12 in Sz˝ucs et al.2016). The values used for each region are given in Table 1.3.2. YSO Population of the CTB 102 Molecular Clouds
We used
WISE observations of the region to explore the YSO content of the molecular clouds. There are multipletypes of sources that can mimic the colors of YSOs, so we must first take steps to remove these contaminants todetermine the actual number of YSO candidates in the region. We use w , w , w
3, and w WISE µ m band magnitudes from the AllWISE catalog respectively.To identify likely background star-forming galaxies (SFG) we use the color cuts from Kang et al. (2017, K17hereafter). These cuts follow Koenig & Leisawitz (2014, KL14 hereafter), but have more conservative magnitudecuts to reduce the number of spurious transition disk source candidates: w − w > . w − w < . w − w < . × ( w − w − . w > . w > . . (4) Marshall et al.
Table 1.
Properties of the CTB 102 molecular cloudsRegion 1 1a 2 2a 2b 2c TotalSize (pc) a × × × × × × × R pc b · · · CO Avg. V LSR (km s − ) − − − − − − − σ v (km s − ) 1.72 1.44 1.53 1.56 1.31 1.60 1.85Mass (log M (cid:12) ) 4.27 3.81 4.62 4.00 4.06 4.14 4.78 CO Avg. V LSR (km s − ) − − − − − − − σ v (km s − ) 1.20 1.16 1.23 1.03 1.29 1.15 1.22Mass (log M (cid:12) ) 3.28 2.64 3.99 3.52 3.56 3.61 4.07Virial Mass (log M (cid:12) ) 4.54 4.04 4.68 4.26 4.22 4.40 4.95 c a Semi-major and semi-minor axes of the elliptical apertures used to determine the mass of the region from theintegrated CO and CO data cubes b Geometric mean size used in the virial mass calculation c Mass determined from the sum of regions 1 and 2
Active galactic nuclei (AGN) with unresolved broad-line regions are another source of false YSOs as their mid-infrared colors are very similar. These AGN however are expected to be fainter than a typical YSO closer than ∼ w > . × ( w − w
3) + 4 . w > . w > . w > w − w . . (5)After the elimination of the red extragalactic contaminants, there are two sources of galactic contaminants that weconsider. We once again follow the method from K17 to determine whether any of the objects are unresolved shockemission knots or resolved PAH emission. For shock emission we follow the criteria: w − w > . w − w < . , (6)and for PAH emission we use: w − w < . w − w > . w − w < . w − w > . . (7)Figure 3 illustrates the contaminant identification procedure. Starting with the 14814 sources selected from theAllWISE catalog, 13223 sources are identified as contaminants, and the remaining 1591 sources move on to the YSOcandidate classification process described in subsubsection 3.2.1.3.2.1. YSO Classification
After removing all of the likely contaminants, we are ideally left with just field stars and YSO candidates. To identifyand classify YSO candidates we followed the KL14 procedure. For convenience we show the YSO classification scheme he Molecular Clouds of CTB 102 Figure 3.
WISE color-color diagram for
WISE µ m bands showing the different contamination cuts describedin subsection 3.2. All 14814 sources extracted from the AllWISE catalog using a spatial search around CTB 102 are plotted.Sources categorized as shock objects, star forming galaxies and PAH emission are found within the boundaries of the black lineson the plot. Source color refers to the WISE µ m band magnitude. below, and refer the reader to KL14 for details on how the color cuts were developed. Class I objects, essentially YSOswith significant infalling envelopes of material, are defined by: w − w > . w − w > − . × ( w − w
3) + 2 . w − w > − . × ( w − w − . w − w < . . (8)Class II objects, corresponding to YSOs with a significant amount of material in a circumstellar disk, are definedby: w − w > . w − w < . × ( w − w − . w − w > − . × ( w − w
3) + 2 . w − w > . × ( w − w − . w − w < . . (9)Transition disk objects, which perhaps are a more evolved Class II object, are defined by: w − w > . . < w − w < . Marshall et al.
Figure 4.
RGB image of CTB102 in
WISE
24, 12, and 4.6 µ m. Red circles correspond to class I sources, magenta diamondsto class II, and cyan boxes to transition disks. Source type comes from the K17 classification given in Table 2. Also shown, aswhite ellipses, are the molecular cloud regions identified in Figure 1. w − w > . × ( w − w − . w ≤ . . (10)As most of the sources have corresponding counterparts, KL14 also developed techniques to identify ClassI and Class II objects using and WISE data for sources not classified using only
WISE photometry.
Class I objects are defined by: H − K S > − . × ( w − w
2) + 2 . . (11) Class II YSOs are defined by: H − K S > . H − K S > − . × ( w − w
2) + 0 . H − K S < (0 . / . × ( w − w − . w ≤ . . (12)We are left with the choice whether or not to consider the χ and signal-to-noise cuts from KL14, which are designedto minimize the number of fake sources that could be potentially classified as YSO candidates. The cuts are thefollowing: WISE band 1 (3.4 µ m): non-null value for w sigmpro and w chi < ( w snr − / WISE band 2 (4.6 µ m): non-null w sigmpro , he Molecular Clouds of CTB 102 WISE band 3 (12 µ m): w snr ≥ w chi < ( w snr − / . < w chi < . WISE band 4 (22 µ m): non-null w sigmpro and w chi < (2 × w snr − / snr , chi
2, and sigmpro are the source signal-to-noise, χ , and photometric uncertainty respectively.Each band cut is applied to the source as that band is required for the classification. For example, only transitiondisk objects must pass the WISE band 4 cut, as these are the only object that require w χ and signal-to-noise cuts and use a visualinspection to identify spurious YSO candidates after the initial classifications are made. Using this procedure weinitially found 128 YSO candidates. The subsequent visual inspection removed 104 sources leaving a total of 24 YSOs(see Figure 5).The quality checks from KL14 are very effective at removing spurious sources, and are best used when investigatingvery large spatial regions. For a smaller area, such as in this study, visual inspection and removal of spurious sourcesis still practical and results in a better picture of the YSO population. We find 18 Class I and II YSO candidatesusing this technique, 11 more than from using the KL14 technique. We also find 6 transition disk objects, whereasKL14 identifies none. A number of the transition disk YSOs appear around a 24 µ m bubble-like structure as seen inFigure 4; however, due to the uncertainty in the evolutionary stage of transition disks, the focus the rest of our studywill be on the 18 Class I and Class II YSO candidates (our ‘best candidate’ sample). The YSO classifications for thissample based on WISE and colors are shown in column 2 of Table 2. Photometry and source designations forthis sample and the 6 transition disk candidates are listed in Table 4 and Table 5 in the Appendix.3.2.2.
Classification via SED Modeling
We also classify the YSO candidates in our ‘best candidate’ sample using a slightly modified version of the techniquedescribed by Alexander et al. (2013). This technique uses the Robitaille et al. (2007) spectral energy distribution(SED) fitting tool to match candidate YSO SEDs to the suite of YSO models from Robitaille et al. (2006). Initially,we attempt to fit reddened stellar models from Kurucz (1993) to the SEDs to ensure that the YSO candidates arenot actually reddened stellar photospheres. We allowed A V to vary from 0-40 mag. and the distance to vary from 4.0to 4.6 kpc. As in Alexander et al. (2013), we defined a good fit when χ / n data ≤
2, where n data is the number ofpoints in the observed SED. Results of testing each best YSO candidate found that none of the sample SEDs meetthis condition, with typical χ ∼ χ − χ /n data <
3. We then used the criteria from Robitaille et al. (2007) to classify each acceptable model basedon the values of the mass accretion rate from the YSO envelope ( ˙ M env ), the stellar mass ( M (cid:63) ), and the circumstellardisk mass ( M disk ). Class I YSOs have ˙ M env /M (cid:63) > − yr − , Class II YSOs have ˙ M env /M (cid:63) < − yr − and M disk /M (cid:63) > − , and Class III YSOs have ˙ M env /M (cid:63) < − yr − and M disk /M (cid:63) < − .This classification is not always clear-cut since it is possible to find both Class I and Class II type models, or modelswith different masses, that are acceptable fits to the source SED. For example, in Table 2 we see that the best-fitmodel to source 1983 is a Class II YSO ( χ = 2 .
8) with a mass of 4.49. There are also 56 other acceptable models(91.8% of all acceptable models) that are also Class II YSOs, and there are another 5 acceptable models (8.2% of allacceptable models) that are Class I YSOs. To quantify this inherent fuzziness in the YSO classification we calculatea weighted average class-type and mass as in Alexander et al. (2013). Weights ( P i ) for each acceptable model ( i ) arecalculated using: P i = exp (cid:18) − χ i − χ (cid:19) , (13)where χ i is the model χ , and χ is the χ for the best-fit model. The weighted average ( ¯ X ) of parameter X (inour case this is either the model mass or YSO class) is then calculated using:¯ X = Σ X i P i Σ P i , (14)where the summation is over all acceptable models.0 Marshall et al.
Figure 5.
WISE color-color diagram for
WISE µ m bands for our WISE classified YSOs (see Table 2, Table 4,and Table 5). The dashed lines denote color criteria for different YSO classes as determined by
WISE photometry. Red sourcesin the top box are classified as class I YSOs, Magenta sources are class II, and cyan sources are transition disks. Note that themagenta sources not confined to a dashed line region were classified by their photometry.
To illustrate the usefulness of procedure we consider two of the YSO candidates, source 1983 and source 11025, inmore detail. The best-fit and acceptable model SEDs are shown in Figure 6 and Figure 7. In the case of source 1983,we find an average YSO class of 1.97, which reflects the fact that almost all of the acceptable models are Class II. Theweighted average mass of 4.41 M (cid:12) is also similar to the best-fit model mass of 4.49 M (cid:12) . For source 11025, the bestSED classification is a class II YSO, but in this case 40% of the acceptable models have a class I SED. The averageYSO class of 1.62 reflects this mix of acceptable models. We also see that the weighted average mass of 4.53 M (cid:12) isdifferent from the best-fit model mass of 5.45 M (cid:12) . Since the average decimal class (column 8 in Table 2) provides aquantitative measure of the uncertainty of any given classification we prefer to use it over the strict binary class I orclass II classification from the WISE colors.We see in Table 2 that there is a slight trend in the average YSO classification depending on the region where theYSO is located. Regions 1, 1a, 2a, 2b, and 2c have average YSO classifications of 1.77, 1.78, 1.48, 1.33, and 1.66,respectively. We notice that the more evolved YSOs (class II) tend to be in region 1 whereas younger YSOs tend tobe found in region 2. This may be indicative of separate (sequential) generations of star formation, where regions 2a,2b, and 2c are younger because they are located farther from the main H II region. We hesitate to make any strongerstatement on this point though due to the small sample sizes in regions 2a, 2b, and 2c. The SED analysis was alsoperformed with a fixed distance of 4.3 kpc rather than allowing the distance to vary ± . Star Formation Efficiency he Molecular Clouds of CTB 102 λ (µm) -12 -11 -10 -9 λ F λ ( e r g s / c m / s ) = 2.844 A V = 8.9 Scale = 0.60 Figure 6.
Acceptable model SEDs for source 1983. The best-fit model is shown as a black line. Points are from and
WISE . The best-fit model and
WISE / photometry both classify the source as a class II YSO. The average class is 1.97due to the presence of a relatively small number of acceptable class I models. λ (µm) -12 -11 -10 -9 λ F λ ( e r g s / c m / s ) = 2.560 A V = 2.6 Scale = 0.65 Figure 7.
As Figure 6 for source 11025. The best-fit model and
WISE / photometry both classify the source as a classII YSO. In this case though, the average class is 1.62 as ∼
40% of the acceptable SED models are class I.
The YSO candidates in the CTB 102 molecular clouds detected by
WISE are primarily intermediate- and high-massobjects. Best-fit masses range from 1.71 – 14.6 M (cid:12) , and weighted masses range from 3.84 – 14.6 M (cid:12) . To explore ifour
WISE data are simply not sensitive to lower-mass YSOs, we used the YSO model photometry from Robitailleet al. (2006) to calculate the average apparent magnitude in the
Spitzer µ m band ([I2]) for Class I and Class IIYSOs in two mass bins (0.5 – 1 M (cid:12) and 1 – 2 M (cid:12) ). Since the Spitzer µ m and the WISE µ m filters are verysimilar (e.g., Jarrett et al. 2011) we assume that the magnitudes are identical for the purpose of this calculation. Eachbin contained approximately 1000 models sampling a wide range of disk and envelope properties as well as samplingten different viewing angles. The average values found for each bin were: Class I (1 – 2 M (cid:12) ) 14.0 (2.3); Class I (0.5– 1 M (cid:12) ) 14.9 (2.3); Class II (1 –2 M (cid:12) ) 14.4 (1.2); and Class II (0.5 – 1 M (cid:12) ) 15.6 (1.3). The uncertainties, givenin the parentheses after the average, were calculated by adding in quadrature the standard deviation due to modelvariations and the average standard deviation caused by the variable viewing angle. WISE has a 5 σ sensitivity at4.6 µ m of w ∼ . WISE data.To determine the total stellar mass, we can extrapolate our observed sample of YSOs down to sub-solar massesassuming a power-law initial mass function (IMF). Fitting a power law to our sample of 18 YSOs using the weightedaverage mass values, we obtain Γ ≡ d (log N )) /d (log M ) = − . ± .
99. Using the best-fit masses, we obtain Γ =2
Marshall et al.
Table 2.
YSO classifications for best candidate sampleBest Fit Alternate Fit AverageID K17 Class c % Match d Mass e A V Class f Mass g Class h Mass i A V Region a Class b ( χ ) ( (cid:12) ) (mag.) ( χ ) (M (cid:12) ) (M (cid:12) ) (mag.)871 Class II (W) I (1.5) 93.6% (137) 4.94 4.18 II (18.7) 4.38 1.00 4.03 3.8 2b1205* Class I (W) I (5.7) 60.0% (51) 5.15 16.6 II (15.1) 11.7 1.01 6.56 18.2 2b1983* Class II (W) II (2.8) 91.8% (56) 4.49 8.9 I (8.4) 3.60 1.97 4.41 8.4 2a1988 Class II (M) II (26) 100% (15) 14.6 6.6 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · a All sources were verified by eye; * sources also meet the photometry quality cuts from KL14. b Source classification based on the criteria in K12, KL14, and K17. W denotes the classification was based solely on
WISE data, and M denotes that and
WISE data were used. c Source classification based on YSO models best fit to the source SEDs and classification scheme from Robitaille et al. (2006)and the χ of the fit. d Percentage of good fit YSO models that are the same classification type as the best fit and the number of models that fit. e Stellar mass of best fit YSO model. f The classification of the best fit model with a different YSO class than the best fit along with the χ of the fit. In somecases alternate classifications are not present because each well fit model are the same class as the best fit. g Stellar mass of the alternative classification best fit YSO model. h Weighted average decimal classification based on the χ for each good fit model. i Weighted average mass for each good fit model. . ± .
32. Both of these slopes are flatter than the canonical Salpeter (1955) IMF (Γ = − . − − − − (cid:12) (the range of mass values in Table 2). For each different cluster IMF, we simulatedsample sizes of 3, 10, 18, 30, 100, 300, 1000, and 3000 YSOs. Each simulation was run 1000 times per sample size.We found that the observed IMF slope does not approach the actual value until the sample reaches approximately 100in size (see Figure 8). For very small YSO sample sizes, it is essentially impossible to get the true IMF slope fromthe observations. For a Salpeter IMF, we find that a random selection of 18 YSOs will produce Γ = − . ± . he Molecular Clouds of CTB 102 . . . . . . . . . Log N(YSOs) − . − . − . − . . . . Γ Γ= − . − . − . − . Figure 8.
Derived IMF slope Γ ≡ d (log N )) /d (log M ) as a function of simulated cluster size for clusters following differentunderlying IMFs. Red dots correspond to a cluster with an intrinsic Γ = − .
35, purple follow Γ = − .
10, blue follow Γ = − . − .
35. Error bars correspond to the standard deviation of the simulated clusters. The black squarecorresponds to our observed IMF slope based on the best-fit YSO masses, and the cyan corresponds to the average mass values. observed IMF is consistent with other power laws within the large standard deviation for small sample sizes, but wefind no reason to not follow Γ = − .
35 in estimating the cluster mass in our case. Using Γ = − .
35, we use the 6 – 7M (cid:12)
YSOs from Table 2 in each region to normalize our IMF, and extrapolate down to the sub-solar mass regime todetermine the cluster mass.The star formation efficiency (SFE), M Cluster / ( M Cluster + M Gas ), was found by considering three different clustermasses. We first look at only our sample of detected YSOs to determine the cluster masses within each region,thus setting a lower limit of the SFE (SFE
Min in Table 3). We also use the Salpeter IMF, with Γ = − .
35, for0 . (cid:12) ≤ M < M (cid:12) , Max = 7 M (cid:12) and Γ = 0 for 0 .
08 M (cid:12) < M < . (cid:12) (SFE in Table 3) . Finally, we derived analternative cluster mass using the same IMF, but ignoring YSOs with M < . (cid:12) (SFE Alt. in Table 3).Table 3 shows our calculated star formation efficiencies for each of the subdivided regions and in total. We remindthe reader that systematic effects discussed in subsection 3.1, which we believe are small in this case, may result inthe SFE values being overestimated. DISCUSSIONThe molecular clouds associated with CTB 102 are likely the substantial remnants of an even larger initial GMCthat was the birthplace of CTB 102. They are 60 ×
35 pc in extent, and our CO observations provide a lower limitmass estimate of 10 . M (cid:12) , while our CO observations give a mass range of 10 . − . M (cid:12) .We used archival WISE and photometry to identify and classify 18 class I and class II YSOs within themolecular clouds. SED fitting was used to estimate YSO masses and to refine the YSO classifications. Star formationactivity is not spread uniformly throughout the clouds, rather our study reveals pockets of intermediate-mass starformation, closely associated with the high CO and CO concentrations seen in Figure 1 and Figure 4. The lack ofobserved low-mass YSOs (M < (cid:12) ) is very likely a selection effect due to the sensitivity of the WISE bands andthe distance of the molecular cloud.4
Marshall et al.
Table 3.
Cluster Mass and Star Formation EfficiencyRegion 1 1a 2 2a 2b 2c Total a (M (cid:12) ) 75.2 35.5 27.2 11.4 6.6 9.2 108.2SFE Min (cid:12) ) 3.64 3.52 3.5 3.20 3.20 3.19 3.87SFE ( CO) b c (cid:12) ) 3.5 3.34 3.32 3.03 3.03 2.68 3.70SFE Alt ( CO) 13.7% 25.3% 4.8% 9.7% 8.5% 3.4% 7.8%SFE
Alt (virial) 7.4% 16.6% 3.9% 5.6% 6.1% 1.9% 5.4% a Sum of average YSO masses in the region (see Table 2). b Star formation efficiency based on the cluster mass and the CO mass. c Star formation efficiency based on the cluster mass and the virial mass.
The average YSO classification depends on position within the molecular cloud. We found the average YSO classfor region 1 and 2 to be 1.74 and 1.47 respectively. This may be indicative of separate generations of star formationpotentially stemming from the O star(s) powering CTB 102. A stronger statement on the existence of an age/classgradient is precluded because of the small YSO sample size: seven YSOs in region 2 and nine YSOs in region 1.A number of transition disk objects were identified, and, although we do not study them in detail, we saw that theyappear in a region without any significant amount of CO or CO emission. Interestingly, the majority of theminstead appear associated with a bubble-like structure seen in the mid-infrared
WISE wavelengths and at 1420 MHzoutside of regions 1 and 2c (see Figure 1 and Figure 2). Again, while the sample size is small, the spatial locationof these objects, which are presumably more evolved YSOs than class II objects, is consistent with the gradient inage/class mentioned in the previous paragraph.The total SFE
Min of the CTB 102 molecular cloud is 0 . WISE keeps us from observing low-mass YSOs that are likely there. When we use the M Cluster values, derived using IMF extrapolations from the observed YSOs (see subsection 3.3), the total SFE and SFE
Alt. values increase and range between ∼ to be between 3 – 6%, and Evans & Lada (1991) found the SFE of L1630 to be 3 – 4% over a 40 × ∼ × ∼ ρ ∗ > (cid:12) pc − , and region1a has ρ ∗ = 10 −
20 M (cid:12) pc − using a third spatial dimension of 10 – 5 pc. They also have observed SFEs rangingbetween 8 – 33%, with higher values apparently associated with more evolved clusters (Lada & Lada 2003). Perhapsthe best nearby ( d < CONCLUSIONS he Molecular Clouds of CTB 102 CO and CO observations of the molecular clouds associatedwith the enormous CTB 102 H II region, which were obtained using the recently commissioned SEQUOIA-TRAOsystem. The conclusions of our study are summarized below:1. We find that what remains of the original molecular cloud has separated into at least two main regions, as seenby the strong divide in the CO and CO maps (see Figure 1), and contains a total mass of 10 . − . M (cid:12) .2. The molecular cloud contains ongoing star formation as seen by the presence of 18 intermediate-mass YSOswithin the cloud. When comparing the pockets of star formation, we found that there is a difference in the averageYSO class between them, suggesting there are at least two separate generations of star formation.3. The SFE of the molecular cloud as a whole is comparable with efficiencies found for other Galactic GMCs.However, with a SFE between 17 and 37%, the region 1a SFE is much higher than what would be expected for aregion only 5 × WISE andNEOWISE are funded by the National Aeronautics and Space Administration.This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project ofthe University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology,funded by the National Aeronautics and Space Administration and the National Science Foundation.Miju Kang was supported by Basuc Science Research Program through the National Research Foundation of Ko-rea(NRF) funded by the Ministry of Science, ICT & Future Planning (No. NRF-2015R1C2A1A01052160).APPENDIX
WISE and photometric data for our best YSO candiate sample and for the transition disk candidates areprovided in Table 4 and Table 5 respectively. REFERENCES
Alexander, M. J., Kobulnicky, H. A., Kerton, C. R., &Arvidsson, K. 2013, ApJ, 770, 1Arvidsson, K., Kerton, C. R., & Foster, T. 2009, ApJ, 700,1000Baudry, A., Lucas, R., & Guilloteau, S. 1995, A&A, 293,594Blake, G. A., Sutton, E. C., Masson, C. R., & Phillips,T. G. 1987, ApJ, 315, 621Bolatto, A. D., Wolfire, M., & Leroy, A. K. 2013, ARA&A,51, 207Brand, J., & Blitz, L. 1993, A&A, 275, 67Clark, P. C., & Glover, S. C. O. 2015, MNRAS, 452, 2057Evans, II, N. J., & Lada, E. A. 1991, in IAU Symposium,Vol. 147, Fragmentation of Molecular Clouds and StarFormation, ed. E. Falgarone, F. Boulanger, & G. Duvert,293Evans, II, N. J., Dunham, M. M., Jørgensen, J. K., et al.2009, ApJS, 181, 321 Ferland, G. J., Chatzikos, M., Guzm´an, F., et al. 2017,RMxAA, 53, 385Fitzgerald, M. P. 1968, AJS, 73, 177Gong, M., Ostriker, E. C., & Kim, C.-G. 2018, ApJ, 858, 16Heyer, M. H., Brunt, C., Snell, R. L., et al. 1998, ApJS,115, 241Jackson, J. M., Rathborne, J. M., Shah, R. Y., et al. 2006,ApJS, 163, 145Jarrett, T. H., Cohen, M., Masci, F., et al. 2011, ApJ, 735,112Kallas, E., & Reich, W. 1980, A&AS, 42, 227Kang, S.-J., Kerton, C. R., Choi, M., & Kang, M. 2017,ApJ, 845, 21Koenig, X. P., & Leisawitz, D. T. 2014, ApJ, 791, 131Kurucz, R. L. 1993, VizieR Online Data Catalog, 6039Lada, C. J., & Lada, E. A. 2003, ARA&A, 41, 57Lada, E. A. 1992, ApJL, 393, L25Lynds, B. T. 1962, ApJS, 7, 1 Marshall et al.
Table 4.
Best YSO Candidate SampleWISEA ID