YOUNG STELLAR OBJECTS AND TRIGGERED STAR FORMATION IN THE VULPECULA OB ASSOCIATION

The Spitzer Space Telescope (Werner et al. 2004) observed over 270 deg² of the inner Galactic plane in six wavelength bands as part of two legacy programs: the Galactic Legacy Infrared Mid-Plane Survey Extraordinaire (GLIMPSE; PI: E. Churchwell; Benjamin et al. 2003) and the MIPS GALactic plane survey (MIPSGAL; S. Carey PI; Carey et al. 2009). The Vulpecula region was covered by GLIMPSE I (PID 188) and MIPSGAL I (PID 20597). It was imaged with the Infrared Array Camera (IRAC; Fazio et al. 2004) at 3.6, 4.5, 5.8, and 8.0 μm, and with the MIPS camera (Rieke et al. 2004) at 24 and 70 μm. The angular resolution of Spitzer is 2'', 6'', and 18'' at 8, 24, and 70 μm, respectively.

The MOPEX package (Makovoz et al. 2006) was used to build mosaics of the sky about 1 deg² wide from individual Basic Calibrated Data (BCD) frames. Details of the post-processing and BCD pipeline modifications are described in Meade et al. (2007) for GLIMPSE and in Mizuno et al. (2008) for MIPSGAL. The large Vulpecula mosaics presented in this paper are also built with MOPEX using the 1 deg² plates available through the Infrared Science Archive³ (IRSA).

The search for YSO candidates in Vulpecula relies on the Point Source Catalogs (PSCs) generated from the GLIMPSE and MIPSGAL surveys. For GLIMPSE, point sources are extracted using a modified version of the point-spread function (PSF) fitting program DAOPHOT (Stetson 1987). The GLIMPSE PSC we used is the enhanced data product v2.0 available on the IRSA Web site. For MIPSGAL, sources are extracted from the 24 μm maps using a Pixel-Response Function fitting method (S. Shenoy et al. 2010, in preparation) and the APEX software developed at the Spitzer Science Center.⁴ This catalog is then merged with the GLIMPSE PSC to include IRAC and Two Micron All Sky Survey (2MASS)⁵ fluxes if they are available. 70 μm sources are extracted from reprocessed 70 μm maps (R. Paladini et al. 2010, in preparation) using the StarFinder software (Diolaiti et al. 2000), and matched to the MIPSGAL PSC with a search radius of 9''.

We complement our data set with visible, submillimeter, millimeter, and radio data to obtain a broader view of the Vulpecula complex, and to investigate the morphology, star formation history, and possible relationship between the H ii regions. We make use of the VLA Galactic Plane Survey (VGPS) data, which consists of H i line and 21 cm continuum emission data with a resolution of 1'×1'×1.56 km s⁻¹ and 1', respectively (Stil et al. 2006). These data are analyzed in details in Section 3.1. We also use the Virginia Tech Spectral-line Survey (VTSS) which provides arcminute-resolution images of the 6563 Å Hα recombination line of atomic hydrogen in the Vulpecula region (Dennison et al. 1998). S. Bontemps (2009, private communication; see Schneider et al. 2006 for details) also provided our team with a 12-resolution extinction map of Vul OB1 generated from stellar counts and colors derived from the 2MASS PSC (see Figure 1). The visual extinction in Vulpecula ranges from 3 to 20 with a distribution peaking at Av ∼5. We also use the CO survey of our Galaxy published in Dame et al. (2001). However, the relatively low spatial resolution of ∼75 over Vul OB1 prevented a relevant detailed study of the CO emission. We find nonetheless a very good correlation between the CO emission and the visual extinction that both trace the cold and dusty molecular material (cf. Figure 1). In addition, we used the submillimeter data obtained during the second flight of the Balloon-borne Large Aperture Submillimeter Telescope (BLAST; Pascale et al. 2008), where a 4 deg² region was mapped around Sh2-86 at 250, 350, and 500 μm (Chapin et al. 2008). The angular resolution of the BLAST maps are 40'', 50'', and 60'' at 250, 350, and 500 μm, respectively. And finally, we exploit the recently released 1.1 mm data of the Galactic Plane Survey carried out with Caltech Submillimeter Observatory/Bolocam (J. Aguirre et al. 2010, in preparation) and providing an angular resolution of 31''.

Zoom In Zoom Out Reset image size

Sh2-86 is an extended H ii region about 40' wide (Sharpless 1959) located in the southern part of Vulpecula (Figures 1 and 2 and Figure 11 in the Appendix). It is excited by the open cluster NGC 6823, which is part of the OB association Vul OB1 (Massey et al. 1995; Bica et al. 2008). Pigulski et al. (2000) estimate the age of the star cluster at 3 ± 1 Myr old. In the atomic gas distribution derived from the VGPS data, Sh2-86 appears as an oblong hole at velocities between 26 and 31 km s⁻¹. We find that the dimension of the shell is approximately 60' along the Galactic plane and 25' across it. In the optical, the southeastern part of Sh2-86 is very contrasted. It exhibits silhouetted pillar-like structures pointing toward NGC 6823, emission and reflection nebulae as well as filamentary structures. The infrared emission is also contrasted in this region tracing the complex interplay between the UV radiation and the surrounding dusty ISM.

Chapin et al. (2008) also report the detection of 49 compact submillimeter sources associated with Sh2-86 using BLAST observations. The presence of these clumps, with masses ranging from 14 to 700 M_☉, indicates the capability of Sh2-86 to form massive stars.

Sh2-87 and Sh2-88 are also H ii regions initially discovered by Sharpless (1959). These regions are active sites of star formation as indicated by the presence of H₂O maser line emissions and bipolar outflows (Barsony 1989; Deharveng et al. 2000). Bolocam data show bright sources at the location of Sh2-87 and Sh2-88 indicating the presence of cold core candidates. Our analysis of the VGPS data reveals that the morphology of the H i shell around Sh2-87 at velocities between 22 and 26 km s⁻¹ is very similar to that of the 24 μm emission. The H ii region appears as a hole in the H i distribution as the hydrogen is ionized by an embedded B0 star (Felli & Harten 1981). A compact source seen in absorption sits at the center of Sh2-87, its position is coincident with the location of the peak emission at 24 μm (l, b) = (60.88, − 0.13). Barsony (1989) and Xue & Wu (2008) give a detailed description of this H ii region.

Sh2-88 is a diffuse nebula of diameter ∼20' located at (l, b) = (61.45, 0.34), which is excited by the O8 star BD +25°3952 (Cappa et al. 2002). The star formation activity taking place in Sh2-88 occurs in a couple of nebular knots identified by Lortet-Zuckermann (1974), Sh2-88A and Sh2-88B, located 15' southeast of Sh2-88. These knots are responsible for the bright 24 μm emission visible in Figure 2 at (l, b) = (61.47, 0.1). Sh2-88B is actually the brightest of the three Sharpless objects at mid-IR wavelengths. Figure 10 in the Appendix shows a composite image of Sh-88B in IRAC bands. It consists of a compact cometary H ii region and an ultracompact (UC) H ii region. Deharveng et al. (2000) present a detailed analysis of Sh2-88B and its stellar content. At radio wavelength, the association of Sh2-88 with a H i shell is somewhat more difficult than in the other two cases as the edges of the shell are not as well defined. We detect, however, a faint hole at velocities between 22 and 26 km s⁻¹ as well as a compact source seen in absorption in the center of Sh2-88B.

We have discovered several pillar-like structures on either side of the Galactic equator between Sh2-86 and Sh2-87 (595 < l < 61°). These objects are similar to the archetypal pillars of creation found in M16 (Hester et al. 1996; Urquhart et al. 2003). Such pillars are usually associated with recent episodes of star formation as the winds and radiation emanating from young massive stars are responsible for sculpting these elongated elephant trunks out of the surrounding molecular material. An accepted mechanism for the formation of the pillars is the slow photoevaporation of a pre-existing molecular clump shadowing a tail of more diffuse gas, but other mechanisms have been proposed based on hydrodynamical instabilities for instance. Spitzer (1954), Bertoldi (1989), Lefloch & Lazareff (1994), and Carlqvist et al. (2003) present analytical models for the formation of pillar-like structures, and Miao et al. (2006), Mizuta et al. (2006), and Gritschneder et al. (2009) carry out numerical simulations of the formation and evolution of such objects. According to these studies, the pressure at the surface of the pillars due to the strong external radiation appears to trigger the formation of new stars, which is confirmed by recent observations (e.g., Sugitani et al. 2002; Reach et al. 2004, 2009; Bowler et al. 2009).

Figure 3 presents the 24 μm image of the pillars found in Vulpecula, and gives the nomenclature to identify individual objects. The three pillars VulP12-13-14 were already known prior to these observations (Chapin et al. 2008); they are located northeast of NCG 6823 and are obviously associated with NGC 6823 as they point toward the star cluster. However, the 11 pillars VulP1 to VulP11 have never been reported in the literature before this study. Their association with Vul OB1 is not as straightforward as in the case of NGC 6823 since no obvious source could be identified as their sculptor. The OB stars mentioned previously are certainly good candidates, for instance, the star HD 186746 (spectral type B8Ia) is located right above the tip of the pillar VulP1 in projection (Figures 2 and 3). Nevertheless, 10 of the discovered pillars, out of 11, seem to point toward the same faint diffuse nebulosity located at (l, b) = (60.37, − 0.04), which is not coincident with any known OB stars. We argue that the chance for 10 randomly distributed pillars to point toward the same object is negligible;⁶ thus the discovered pillars VulP1–VulP10 are most certainly associated with each other. The central nebulosity might host the object responsible for the formation of the pillars. In addition, a diffuse faint circularly symmetric structure centered on the same nebulosity is discernible in the background diffuse emission. Bright objects at 24 μm also seem to delineate this faint structure at a radius of ∼15' centered at (l, b) = (60.37, − 0.04). These morphological features suggest that a single event in the past may have molded the pillars region.

Zoom In Zoom Out Reset image size

We analyze the radio data in the pillars region to look for a possible association between these structures and the OB association. Out of the 15 pillars identified in Figure 3, only three are detected in the VGPS H i line data. The pillar VulP10 presents a deep absorption feature at 27–33 km s⁻¹ as well as smaller features centered at 8, 13, and 20 km s⁻¹. These H i features are marginally resolved into two close compact sources located at the base of the pillar. Since these features are quite different from the IR morphology of VulP10, we cannot exclude a fortuitous association with a coincident radio source seen in projection.

VulP4 and VulP5 appear to have similar characteristics; they present comparable morphologies in the H i line data and in the Spitzer bands. They both emerge from the molecular cloud to be exposed to the ionizing radiation. These pillars are located at the edge of H i shells, and their silhouette is clearly identified at velocities 21–23 km s⁻¹, which places them at roughly the same distance as the three Sharpless objects for which V_LSR = 22–31km s⁻¹.

VGPS continuum emission shows a bright two-lobed compact source located at the base of the pillar VulP10, consistent with the two blobs observed in the H i line data. No other pillars are detected in the 21 cm continuum.

Tables 3 and 4 in the Appendix summarize the geometrical and morphological information extracted from the Spitzer observations for individual pillars. Of particular interest are the bright (∼2 Jy) compact sources found at 70 μm in the core of the pillars VulP1 and VulP3 (see Figure 4). These embedded sources are faint at 24 μm (∼1 mJy) and have no counterparts in IRAC and 2MASS bands. Even though star formation is expected to occur preferentially at the tip of such structures, these red sources are likely sites of massive star formation. Longer wavelength⁷ observations would be necessary to identify the nature of these sources. Figure 4 shows the pillars VulP5 and VulP6 and their associated red sources. YSO candidates are located at the tip of each pillar, and also in the pedestal of VulP5 along some infrared dark clouds (IRDCs). Submillimeter BLAST sources are also located at the tip of the pillars and along the same IRDCs. Two Bolocam sources are also located along one of the IRDCs.

Zoom In Zoom Out Reset image size

We first compile a catalog of IRAC point sources based on the highly reliable PSC available from the IRSA Web site as a GLIMPSE I v2.0 enhanced data product.⁹ All sources found in this catalog with galactic coordinates in the range 586 < l < 622 and −10 < b < 08 are considered (∼3.28 × 10⁵ sources). Fluxes are converted to magnitudes using the zero points given in Table 1. To ensure good photometry and avoid flux stealing between adjacent sources as noted by Robitaille et al. (2008), we require that valid point sources have magnitude errors lower than 0.2 mag, and that their close source flag¹⁰ (csf) is zero. We also apply specific cuts on IRAC magnitudes to achieve high detection reliability; the magnitude limits we adopt are 14.2, 14.1, 11.9, and 10.8 mag at 3.6, 4.5, 5.8, and 8.0 μm, respectively. According to the GLIMPSE I Assurance Quality Document,¹¹ these limits ensure a detection reliability above 98% in all four IRAC bands. The reliability of the faintest sources however might vary across the Vul OB1 region as it depends on the background brightness structure and the local source density.

This intermediate catalog is then merged with the MIPSGAL 24 μm PSC (S. Shenoy et al. 2010, in preparation) with the constraint that the maximum separation between two matching sources is 3''. Once again to ensure a good photometry, we require that the magnitude of 24 μm sources and their associated errors be constrained. These constraints, however, are applied at a later stage of the selection procedure (see Section 4.2).

Out of nearly 3.3 × 10⁵ sources present in the GLIMPSE catalog within our search area, only ∼1.47 × 10⁵ bypassed the initial magnitude cuts. Most of the exclusions were due to noisy detections in bands 3 and 4 of IRAC.

In the end, the initial catalog from which we search for YSO candidates contains point sources with well measured photometry and with at least one detection in IRAC bands, plus the associated 2MASS and MIPSGAL 24 μm fluxes if they exist.

We start our search for YSO candidates in the initial catalog by removing extragalactic contaminants that might be misidentified as YSOs. Using the selection criteria presented in the Appendices of Gutermuth et al. (2008), which are based on a complementary analysis of the Bootes field IRAC data (Stern et al. 2005), we identify and reject six sources dominated by polycyclic aromatic hydrocarbons (PAHs) feature emission, likely star-forming galaxies or weak-line active galactic nuclei (AGNs). No sources matching the colors and magnitudes criteria for broad-line AGNs were found. Considering a search area of 6.5 deg², we find a density of extragalactic sources in Vulpecula slightly lower than 1 per deg². This is comparable to the density of ∼0.5 galaxy per deg² found in the zone of avoidance by Marleau et al. (2008) using GLIMPSE and MIPSGAL data. We further filter the catalog according to Gutermuth et al. criteria and reject eight sources that have a large 4.5 μm excess emission consistent with molecular hydrogen line emission found in regions where high-velocity outflows interact with the cold molecular cloud. The remaining sources are presumably of stellar origin.

All sources with the following color constraints are considered likely YSOs:

$$\begin{equation*} \begin{array}{c} {[}4.5]-[8.0] > 0.5,\\ {[}3.6]-[5.8] > 0.35,\\ {[}3.6]-[5.8] \le 3.5\times ([4.5]-[8.0])-1.25, \end{array} \end{equation*}$$

Out of the ∼2.38 × 10⁴ sources possessing all four IRAC magnitudes, we identified 820 YSO candidates based on their IR-excess emission. We further classify the selected objects according to their infrared spectral index α_IR = ∂log(λF_λ)/∂log(λ) as defined by Lada (1987). We specifically compute the spectral index α_IRAC as the slope of the spectral energy distribution (SED) measured from 3.6 to 8.0 μm. Objects with α_IRAC > − 0.3 have a flatish or rising SED indicating the presence of a cold dusty envelop infalling onto a central protostar; these are designated class 0/I YSOs. Objects with −0.3>α_IRAC > − 1.6 are classified as class II YSOs; these are pre-main-sequence stars with warm optically thick dusty disks orbiting around them. We define class III objects as having −1.6>α_IRAC > − 2.56: these stars have cleared most of their circumstellar environment; their near-IR emission is mostly photospheric but they may present some excess emission above 20 μm. Such objects are dubbed anemic disks by Lada et al. (2006). Finally, objects with α_IRAC < −2.56 are stars with photospheric emission only; this spectral index corresponds to the slope of the stellar photosphere SED in the Rayleigh–Jeans domain.

Note that this YSO classification is based on the classification scheme proposed by Greene et al. (1994), except that the Flat Class (0.3>α> − 0.3) is included into the class 0/I which represents the population of protostars with infalling envelopes. Calvet et al. (1994) indeed showed that Flat Class YSOs could be interpreted as infalling envelopes. Figure 5 presents a color–color diagram based on IRAC data only; the different symbols indicate the classes associated with each YSO candidate identified in our analysis.

Zoom In Zoom Out Reset image size

The next step is to exploit the extra information contained at longer wavelengths to confirm, or reclassify, the YSO candidates identified solely from their IRAC colors. We first check that the SEDs of all YSO candidates continue to rise from 8 to 24 μm. We also look for objects previously classified as photospheric sources that exhibit bright 24 μm fluxes ([5.8] − [24]>1.5). These sources are thought to be transition disks, i.e., class II YSOs with significant dust clearing from their inner disks (Calvet et al. 2002; D'Alessio et al. 2005), these objects are of particular importance for understanding the evolution of circumstellar disks around young stars. We also check for protostar misclassifications due to extreme visual extinction levels. Gutermuth et al. (2008) suggest that if a protostar that has MIPS detection does not meet the criterion [5.8] − [24]>4 (if they possess [5.8] photometry) or [4.5] − [24]>4, then it is likely a highly reddened class II YSO. Finally, we require that any sources lacking detections in some IRAC bands yet having bright 24 μm fluxes ([24] < 7 and [X] − [24]>4.5 mag, where [X] is the photometry for any IRAC detection available in our catalog) have to be added to the list of likely YSOs, and to be classified as highly embedded protostars. We apply the 24 μm based color constraints on YSO candidates only if their MIPS detection reliability is >95% (S. Shenoy 2009, private communication); thus we require that, for the MIPS photometry to be relevant for the YSO selection, the 24 μm magnitude is [24] < 8.6 mag and that its associated error is σ_[24] < 0.2 mag.

We tested the ability of the method at finding genuine YSOs on a well studied star-forming region, RCW79 (Zavagno et al. 2006), and it proved to be adequately reliable and efficient (A. Zavagno 2009, private communication). Furthermore, this method has been successfully applied to study embedded stellar clusters in NGC 1333 (Gutermuth et al. 2008), the star formation activity in the H ii region W5 (Koenig et al. 2008) or in the giant molecular cloud G216 − 2.5 (Megeath et al. 2009). Note that other methods, also based on Spitzer colors and magnitudes, have been developed to find YSO candidates (e.g., Hartmann et al. 2005; Harvey et al. 2007; Robitaille et al. 2008; Chavarría et al. 2008).

We finally find 856 YSO candidates in Vul OB1: 239 are likely protostars with infalling envelops (class 0/I), 464 are disk-bearing stars (class II), and 153 are class III objects with very little circumstellar material. Among the class I objects, 15 YSO candidates are likely deeply embedded protostars, and 89 have a spectral index consistent with the Flat Class. Among the class II objects, 85 are highly reddened class II YSOs that were misclassified as protostars based on IRAC criteria only. Further 21 sources are classified as likely transition disk class II objects.

Except for a few cases, all YSO candidates have detections in all four IRAC bands, about half of them have 2MASS detections in all three JHK bands, 75% have 24 μm counterparts (it reaches 100% for class 0/I objects) and 8% have 70 μm detections. Table 2 gives the coordinates and fluxes of all YSO candidates. Table 5 in the Appendix gives the coordinates, fluxes, and type of the contaminants identified in Vul OB1.

To test the robustness of our analysis relative to the interstellar extinction, we have used the visual extinction map of Vul OB1 and the formulae provided by Flaherty (2007) to deredden IRAC fluxes. We run the exact same procedure as described above, and we find that the YSO census is only marginally different from the uncorrected case (except for the class III population that is reduced by 35%). This justifies the choice of [5.8] − [8.0] as an extinction-free indicator for finding YSOs.

In the present study, we make the deliberate choice to favor the highest possible detection reliability, at the expense of a modest completeness figure. This choice was largely motivated by the need to automate the search for YSOs over such a large area of the sky.

As mentioned in Section 4.1, we achieve a detection reliability better than 98% for the set of IRAC magnitude cuts chosen for populating our initial catalog. However, the actual detection reliability depends to a certain extent on the level of the background diffuse emission. An adequate assessment of the completeness of our YSOs catalog would require to carry out a large spectroscopic survey of the red objects in Vul OB1 in order to identify genuine YSOs, and then compare the results to our catalog. This is impractical though, given the large number of (faint) objects in Vul OB1.

Considering the stringent criteria used to reach a reliability over 98%, we expect a noticeable number of YSOs to be excluded from our initial PSC. For instance, any saturated sources, that are either bright sources on faint background or faint sources on bright background, are not considered in our analysis. There are very few saturated data in IRAC images so that the YSOs selection is basically unaffected by this effect. Still, we expect that a fraction of YSO candidates will not have 24 μm counterparts because of saturation issues (especially around Sh2-88B and Sh2-87 where the background is high). Extended sources are also excluded based on the selection procedure used in this work (see Section 4), but their actual number is expected to be fairly small at 2.3 kpc. Point sources with too large a magnitude error are not considered either. We set the limit for bad photometry to be σ_mag > 0.2 mag. IRAC bands 3 and 4 being the noisiest bands, the selection criterion on σ_mag mostly impacts [5.6] and [8.0] detections.

The completeness of our YSO catalog is also dependent on the evolutionary stage of the object. For instance, the wavelength band 2–24 μm is well suited for detecting and identifying disks around young stars (class II), but longer wavelengths are more appropriate for observing early-stage objects (class 0/I). We believe that our sample of IRAC-selected YSO candidates may lack a significant fraction of these protostars. For instance, we looked for a possible correlation between the spectral index α_IRAC and the visual extinction of the YSO candidates expecting to find a population of protostars with high values of α_IRAC, i.e., the youngest objects, to be the most extinct/embedded objects. However, we did not find such correlation in our sample which suggests that we might be missing the most embedded phases of star formation based on IRAC and MIPSGAL 24 μm data only. Similarly, the population of class III YSOs which lacks near-IR excess emission is strongly underestimated in our study, even if they are genuine YSOs possessing Hα or X-ray emission. For all these reasons, it is very difficult to assess the completeness of our YSO catalog with confidence. We use the evolutionary models of Baraffe et al. (1998) to compute the mass of a 2 Myr old star at 2.3 kpc with no IR-excess emission having a 3.6 μm magnitude of 14 mag, and we find a mass of ∼1 M_☉. Considering that the 8 μm channel of IRAC is the less sensitive channel (limit magnitude of 10.8 mag), and that most YSOs have significant IR-excess emission, we estimate our YSO catalog to be complete down to a few solar masses.

Lastly, we have to account for the possible misidentification of YSO candidates with evolved stars. The circumstellar environment of planetary nebulae (PNe) or asymptotic giant branch (AGB) stars are rich in warm dust grains emitting in the near- mid-IR such that these objects occupy the same locus in color–color diagrams as YSOs (Hora et al. 2008; Srinivasan et al. 2009). Following the analysis of Robitaille et al. (2008), we estimate the fraction of evolved stars in our catalog of YSO candidates to be about 25%, of which most are AGB stars. We will argue in Section 5.2 that AGB stars could be further excluded from the sample of YSO candidates based on clustering criteria.

We estimate the current star formation efficiency (SFE) in Vulpecula by comparing the mass of the gas reservoir, M_cloud, with the mass that has turned into stars during the last few million years, M_YSO. The SFE is then derived as follows:

$$\begin{equation} \rm {SFE}=\frac{{\it M}_{\rm YSO}}{{\it M}_{\rm YSO}+{\it M}_{\rm cloud}}. \end{equation} \tag{ 1 }$$

We compute the mass of the cloud using our extinction map of Vulpecula and the formula relating the column density and the visual extinction N_H/A_V = 1.37 × 10²¹ cm⁻² mag⁻¹ assuming a value of R_V = 5.5 typical for molecular clouds as suggested in Evans et al. (2009). We use a contour of A_V = 7 mag to delineate the active star-forming region on the A_V map of Vulpecula. We find an average mass surface density of 16.2 M_☉ pc⁻². We integrate the surface density over the 2.7 × 10³ pc² enclosing the A_V > 7 region, and we find a cloud mass M_cloud = 4.5 × 10⁴M_☉.

Since our YSO sample represents only the high end of the initial mass function (IMF), as we estimated in Section 4.4, we need to account for the missing mass of the total YSO population in order to derive a relevant M_YSO. We use the IMF of Kroupa (2001) to compute the number of stars with masses ranging from 0.01 to 50 M_☉. We normalize the IMF assuming 510 stars have masses >1 M_☉ within the A_V > 7 region, and we find that Vul OB1 should contain 2 × 10⁴ YSOs with a total mass of M_YSO = 4200M_☉. We thus obtain a SFE of ∼8%, similar to other star-forming regions in nearby molecular clouds (Evans et al. 2009).

The spatial distribution of YSO candidates in Vul OB1 is presented in Figure 6. YSO candidates are not distributed randomly in the field. We rather find several coherent structures, or groups of YSO candidates, which in most cases are reminiscent of the mid-infrared morphology of the complex. For instance, the highest densities of YSO candidates are located close to, or within the three H ii regions. Other YSOs appear to line up along IRDCs, e.g., at (l, b) = (59.98, 0.06) at the pedestal of the pillar VulP5 (see Figure 4); or along bright contrasted structures of the extended 24 μm emission, e.g., at (l, b) = (61.82, 0.33) or (l, b) = (59.80, 0.64). YSO candidates have also been identified at the tip of most newly discovered pillar structures (see Table 4 in the Appendix and Sections 3.2 and 5.4.2).

Zoom In Zoom Out Reset image size

If we assume the IMF parameters presented in Section 5.1, we estimate that the average surface density of YSO candidates is 7.4 YSO pc⁻². This value is consistent with the typical surface density found in other star-forming regions, e.g., 13.0 YSO pc⁻² in Serpens (Harvey et al. 2007) or 3.3 YSO pc⁻² in Lupus (Merín et al. 2008). We build a surface density map of YSO candidates following the definition of Chavarría et al. (2008). At each point of a 5'' grid, we compute the surface density of YSOs as

$$\begin{equation*} \sigma =\frac{N}{\pi r_N^2}, \end{equation*}$$

where r_N is the distance to the N = 5 nearest neighbor. This image is then convolved with a Gaussian of FWHM = 1' to obtain a smooth contour map of the surface density. Figure 6 shows the surface density contours over a 70 μm image of Vul OB1. It clearly delineates the mid-IR-bright H ii regions, plus other groups of YSO candidates associated with fainter compact mid-IR sources, e.g., at (l, b) = (60.60, − 0.70). The surface density peaks at ∼50 YSO pc⁻² at the position (l, b) = (59.80, 0.064) which is coincident with the BLAST source V30, also known as IRAS 19410+2336. This source is a candidate hypercompact (HC) H ii region according to Chapin et al. (2008), and about 15 of our YSO candidates appear to be grouped around it (see Section 5.4.2).

In addition to this population of clustered YSO candidates, there is a population of distributed IR-excess sources. Robitaille et al. (2008) argue that a fraction of these isolated red objects could be AGB stars mistaken for YSO candidates in our selection process (they occupy the same locus in the C–C diagram; cf. Section 4). However, Koenig et al. (2008) argue that the isolated YSO candidates could be genuine YSOs that formed from isolated events or that were ejected¹² from their birthplace due to gravitational interactions with other members of their native star cluster.

We now focus our analysis on the population of clustered YSO candidates since they are more reliable tracers of triggered star formation events. Various methods can be used to discriminate between distributed/clustered populations. For instance, the two-point correlation function was used by Karr & Martin (2003) on W5 and by Indebetouw et al. (2007) on M16, while the minimum spanning tree technique was used on W5 by Koenig et al. (2008) and on NGC 1333 by Gutermuth et al. (2008). Both methods provide valuable qualitative results but dynamical measurements are absolutely necessary to assign a definitive membership to a given YSO. Since such data are not available for Vul OB1 we opt for the simplest filtering, namely the nearest neighbor, to gain insight into the clustering properties of our YSO sample. In practice, we consider all YSO candidates and keep only those that have their nearest neighbor YSO within 3' (2 pc at 2.3 kpc). The result of this selection is a much more clustered distribution of YSO candidates that traces mid-IR morphology even more closely (192 YSO candidates are marked as isolated objects).

We also look for trends in the environment of YSO candidates compared to field stars. The top panel of Figure 7 shows that YSO candidates sit preferentially on regions of bright 24 μm background emission compared to the more evolved stars. Since the 24 μm emission is mostly due to the thermal emission of small dust grains excited by UV radiation, the trend for YSO candidates to sit on bright mid-IR background implies that the identified YSO candidates are largely associated with photon-dominated regions (PDRs) and H ii regions. This is consistent with known scenarios of triggered star formation mechanisms and with previous studies of star-forming regions (e.g., Koenig et al. 2008). We also compare the visual extinction associated with the YSO population to the entire population of point sources found in our initial catalog (see the bottom panel of Figure 7). We find that YSO candidates are mostly located in regions of high extinction compared to field stars. This is again consistent with the idea that infant stars still live in the dense and dusty cloud from which they were born. Remarkably, the environmental trends mentioned above are even more pronounced for the population of clustered YSO candidates (see Figure 7).

Zoom In Zoom Out Reset image size

Note that Chapin et al. (2008) identified a few BLAST sources in the direction of Vul OB1 that are actually located at 8.5 kpc from the Sun (radial velocity around −5 km s⁻¹), i.e., in the Perseus arm. A dozen of our YSO candidates seem to be associated with these distant BLAST sources so that these might belong to the Perseus arm.

We have built SEDs for all of our YSO candidates. Figure 8 presents a small SED sample with the Class designation associated with each object.

Zoom In Zoom Out Reset image size

We used the SED fitter tool developed by Robitaille et al. (2007) to extract the physical parameters of our YSO candidates. However, the limited wavelength coverage (near- to mid-IR) resulted in strong degeneracies over the output parameters, and we were unable to gain reliable information on the physical properties of our YSO candidates.

We looked for possible associations in the BLAST field to extend the wavelength coverage to the submillimeter, but in most cases several YSO candidates fell within the BLAST beam rendering the association ambiguous. For the YSO candidates that could be uniquely associated with a BLAST source, Chapin et al. (2008) provide physical parameters derived from SED fitting. They find clump masses ranging from 40 to 400 M_☉ and temperatures from 19 to 28 K. We also use Molinari et al. (2008a) diagnostic diagrams based on [24 − 70] and [250 − 500] colors to identify the most massive YSOs in Vul OB1. Molinari et al. show that the high-mass analogs of the low- or intermediate-mass class 0 objects have distinctive MIPS colors. From our sample of YSO candidates, we find six objects with [24 − 70]>3, which is indicative of an SED peaking longward of 70 μm presumably due to a massive infalling envelop. Three of these red MIPS sources are located next to the brightest BLAST source V30 and are barely discernible on the map, two have no BLAST counterparts, and one might have a BLAST counterpart but the association is ambiguous as another YSO candidate falls within the BLAST beam.

Finally, we searched the IRSA for Bolocam sources, and we found 24 sources in Vul OB1. Among them we find isolated sources not associated with any YSO candidates, these could be YSOs in their earliest evolutionary phases, i.e., starless cores emitting in the millimeter only. The other Bolocam sources are usually located close to the peaks of the YSO surface density and are associated with groups of YSO candidates. Note that almost all Bolocam sources have BLAST counterparts over the Vulpecula BLAST field.

5.4.1. The Case of SNR G59.5+0.1

5.4.2. Photoevaporation at the Tip of the Pillars

5.4.3. Embedded Star Cluster

5.4.4. Large-scale Propagating Star Formation

Based on radio observations of H i supershells and on energetics arguments, Ehlerová et al. (2001) estimate that the H i shell GS061+00+51, which encompasses Sh2-87 and Sh2-88, is 6–7 times larger and 3–4 Myr older than the shell GS59.7–0.4+44 that hosts Sh2-86. Ehlerová et al. attribute these age and size differences to the delayed expansion of H ii regions, and they suggest that star formation might be propagating from Sh2-88 to Sh2-86. Previously, Turner (1986) suggested that the morphology of the three H ii regions was shaped by a single supernova shock wave associated with the fossil H ii region Lynds 792 (l, b) = (60.81, 2.24). Since Lynds 792 is approximately equidistant from the three H ii regions, Turner's scenario implies that star formation would be triggered simultaneously, as opposed to sequentially in the case of Ehlerová et al. We now test the validity of these scenarios using the supplemental information gathered in this work.

Figure 9 shows the evolution of the number of YSO candidates as a function of their Galactic longitude. The increasing number of YSO candidates toward Sh2-86 seems to favor the scenario of Ehlerová et al. since the youngest star-forming region should indeed possess the largest number of YSOs, assuming identical IMFs and instantaneous star formation. Nevertheless, the shape of this histogram could also be interpreted as Sh2-86 being a more active star-forming region than the other two Sharpless objects, creating more YSOs from a larger molecular reservoir, which is unrelated to any process of propagating star formation. For a proper interpretation of the Figure 9 histogram, we need to compare the typical lifetime of a YSO and the timescale necessary for star formation to propagate in a molecular cloud.

Zoom In Zoom Out Reset image size

Nomura & Kamaya (2001) have studied self-propagating star formation using numerical simulations and found that the time delay of sequential star formation sites against the original one, Δt, correlates with their physical separation, Δx, as Δt ∼ 50 Myr[Δx/(0.5 kpc)]^0.5. Efremov & Elmegreen (1998) have derived a similar expression based on observations of star clusters in the Large Magellanic Cloud (Δt ∼ 26 Myr[Δx/(0.5 kpc)]^0.4). Assuming Sh2-88 and Sh2-86 are separated by ∼80 pc, we find that star formation would take ∼10–20 Myr to propagate across Vul OB1. Besides, Evans et al. (2009) estimate the lifetime of a YSO to be of the order of 1–3 Myr based on a large statistical sample of YSOs identified as part of the c2d legacy survey.¹³ If star formation was indeed propagating along the Galactic plane at the pace derived above, YSOs would only be found in a narrow slab of longitudes since they would age, and fade away in the IR, faster than the triggering shock needs to cross the molecular complex. We can already rule out this scenario based on the presence of YSOs in all three H ii regions of Vul OB1 (see Figure 6). It is possible however that star formation is propagating faster than previously estimated due to the pronounced inhomogeneities of the propagation medium (cf. distribution of CO in Figure 1). If the propagation timescale was of the same order of the YSOs lifetime, then YSOs would be found all across Vul OB1 with a gradient in the evolutionary phases of the YSO population as a function of longitude.

We use the ratio of class II–III to class 0–I as an indicator of the aging of the YSO population. Figure 9 shows that this ratio does not present a gradient with respect to longitudes. This indicates that if star formation was once triggered in Vul OB1, then the triggering agent would have been independent of Galactic longitude, and this definitely rules out the suggestion of Ehlerová et al. Nonetheless, the histogram of Figure 9 is still consistent with Turner's scenario in which star formation occurs as an instantaneous burst in the three H ii regions.