ASSESSING MOLECULAR OUTFLOWS AND TURBULENCE IN THE PROTOSTELLAR CLUSTER SERPENS SOUTH

Outflows and jets occur during the early stages of protostellar evolution, and they facilitate the accretion of material onto the protostar by carrying away excess angular momentum (Bachiller 1996). A molecular outflow consists of ambient gas that is entrained by a protostellar jet, expected to occur during the Class 0 and I stages of star formation. Class II sources also power jets and winds and may drive outflows; however, molecular outflows are less prominent as a source evolves and the surrounding molecular gas becomes depleted. Observationally, outflows serve as signposts that emanate from their driving protostellar sources and indicate that star formation is ongoing in a particular region. Further, the characteristics and morphology of outflows may indicate the evolution of the forming protostar and constrain the gas entrainment mechanism (Arce et al. 2007).

Carbon monoxide is the most common probe of molecular outflows, because after H₂ it is the most prevalent molecule in molecular clouds and its low-level rotational states are easily excited in these environments (e.g., Wilson et al. 2009). The CO molecular outflow components are generally characterized as standard high-velocity (SHV) with velocities of a few to about 20 km s⁻¹, and extremely high velocity (EHV) with even higher velocities, typically up to ∼100 km s⁻¹ (see Bachiller 1996). Typically the SHV outflows are more massive than EHV, and in this paper we conduct a census of mostly SHV components within the protostellar cluster Serpens South in order to determine the impact of outflows on the surrounding cloud environment.

Outflows may reach distances up to several parsecs, affecting the surrounding protostellar environment corresponding to these scales, including the core and cloud. Since outflows sweep up ambient gas, they influence the intimate relation between a protostar and its surroundings. Considering that the majority of clustered star formation occurs within clusters of 100 or more members, and the size scales of clusters are typically less than a few parsecs (Lada & Lada 2003), there is ample opportunity for outflows to encounter other nearby outflows and their respective protostars, as well as potentially disrupt the surrounding cloud environment.

It remains a challenge to quantify the cumulative impact of protostellar outflows on molecular clouds and subsequent star formation, but evidence of outflow-cloud interaction has been observed in a variety of star-forming regions (e.g., Arce et al. 2010; Nakamura et al. 2011a, 2011b; Duarte-Cabral et al. 2012). Specifically, outflows interact with the ambient cluster gas (Fuller & Ladd 2002; Arce & Sargent 2006), inject momentum and energy into the surrounding cloud (Arce et al. 2007; Swift & Welch 2008), and they may feed turbulent motions that affect ongoing star formation (see also Frank et al. 2014). Models of clustered star formation must incorporate outflows in order to drive turbulence and produce outcomes close to what is observed in star-forming regions, including star formation rates (SFRs) and efficiencies (e.g., Li & Nakamura 2006; Nakamura & Li 2007, 2011; Carroll et al. 2009; Wang et al. 2010; Krumholz et al. 2014). The cumulative effect of outflow activity, particularly in clustered star-forming regions, can be constrained by observations of clusters.

In the work presented here, we study outflow characteristics and assess their impact within Serpens South, which is the second case study in our survey of clustered star-forming regions ranging in evolutionary stage, beginning with NGC 1333 (Plunkett et al. 2013). NGC 1333 is a prototypical region that has been well studied, while Serpens South is more recently discovered (Gutermuth et al. 2008), more massive, and with a higher fraction of protostars in their earliest stages compared with NGC 1333. We describe Serpens South in this context in the following sub-section. The goals of the current paper are (a) measure the outflow activity in Serpens South, (b) investigate the sustained impact of outflows compared with turbulence and gravity, and (c) establish an empirical scenario of cluster evolution based on the comparison between Serpens South and NGC 1333.

The protostellar cluster Serpens South merits study for its youth, active star formation and corresponding outflows, the concentration of young stellar objects (YSOs) along a dense filamentary structure, and its relative proximity (429 pc, as discussed in the following paragraph). Serpens South was discovered by Gutermuth et al. (2008) in imaging of the Serpens-Aquila Rift with Spitzer IRAC, in which its dark filamentary structure appeared as absorption in the 8 μm band. Subsequent studies have confirmed the filamentary structure of this cloud using dust continuum maps (André et al. 2010; Maury et al. 2011) and molecular line emission maps (Kirk et al. 2013; Tanaka et al. 2013; Fernández-López et al. 2014).

Serpens South is one of only a few active star-forming regions known within a distance ≲1 kpc away. However, various distances have been adopted throughout the literature. Serpens South and the Serpens Main cluster (3° to the north) were found to have similar local standard of rest velocities (v_LSR), in addition to their close proximity on the plane of the sky, and therefore the distance to Serpens South is based on the distance determined for Serpens Main (Gutermuth et al. 2008). The majority of works since the discovery of Serpens South by Gutermuth et al. (2008) have assumed a distance to Serpens South of 260 ± 37 pc, which is the distance determined using photometry of Serpens Main by Straižys et al. (1996) and consistent with the distance determined for the Serpens-Aquila Rift by Straižys et al. (2003). Most recently, Dzib et al. (2011) used Very Long Baseline Array trigonometric parallax measurements to determine a distance of 429 ± 2 pc to the YSO EC 95 associated with Serpens Main, updated from the previous measurement of 415 pc by Dzib et al. (2010). Throughout this paper, we assume that Serpens South is part of the same complex as Serpens Main, and we adopt the distance of 429 ± 2 pc from Dzib et al. (2011).

Outflows in this region were observed by Nakamura et al. (2011a) at an angular resolution of $22^{\prime\prime}$ and presented in the context of the filamentary 1.1 mm dust continuum image by ASTE. Near-IR observations by Teixeira et al. (2012) show corresponding molecular hydrogen emission-line objects. Polarization observations by Sugitani et al. (2011) reveal the importance of the magnetic field in forming the main filament because the magnetic field is well ordered and oriented perpendicular to this filament.

The classification of candidate YSOs in Serpens South was first done by Gutermuth et al. (2008) following the data reduction pipeline presented by Evans et al. (2007), which was developed for the Spitzer c2d and Gould Belt projects, incorporating 2MASS and Spitzer 1.25–70 μm photometry. Within the $14^{\prime} \times 10^{\prime}$ region that they studied, Gutermuth et al. (2008) reported 54 Class I protostars (including flat-spectrum sources) and 37 Class II YSOs. Hence, about 60% of the YSOs are protostars, but at the cluster center this fraction rises to about 80%, which is where we focus our study (see also Section 4.4). Gutermuth et al. (2008) reported a mean YSO surface density of 430 pc⁻² at the distance of 260 pc, which corresponds to ∼160 pc⁻² at the farther distance of 429 pc that we assume here. The median projected distance between nearest-neighbor YSOs in this region is 132 (see Gutermuth et al. 2008, for discussion), or 0.03 pc at a distance of 429 pc. This nearest-neighbor spacing is less than the typical extent of outflows, and therefore this region provides an ideal case study where outflows likely interact with one another, the surrounding protostars, and their cluster environment.

The Herschel Gould Belt Survey (André et al. 2010; Bontemps et al. 2010) revealed several of the youngest Class 0 members of Serpens South, further suggesting the overall youth of the region. Maury et al. (2011) suggest that the high number of protostars compared with the total number of YSOs in Serpens South indicates that this cluster experienced a recent burst of star formation, particularly along a filamentary structure spanning from the southeast to northwest. They used complementary 1.2 mm continuum maps obtained with the MAx-Planck Millimeter BOlometer array (MAMBO) on the Institut de Radioastronomie Millimétrique (IRAM) 30 m telescope, along with Herschel and Spitzer maps, to perform a source extraction and classification over wavelengths from 2 μm to 1.2 mm. They determined that this cluster has formed 9 Class 0, 48 Class I, and 37 Class II objects within an area of $15^{\prime} \times 25^{\prime}$ that has a mass of $\sim 610$ M_⊙. They suggest that a high SFR and a relatively low star formation efficiency are evidence of the early evolutionary stage of this region.

Here we present the highest-resolution observations published to date of millimeter-wavelength molecular line and continuum emission observations made toward the most active central region of Serpens South. Specifically, we observed several isotopologues and energy level transitions of carbon monoxide, tracing the SHV outflow components. ¹²CO traces cool outflow emission, a probe of swept-up molecular gas. ¹³CO is less optically thick and was used to obtain reliable estimates of outflow mass. Together, observations of ¹²CO and ¹³CO provide a comprehensive view of the molecular gas kinematics in the region. Continuum emission was simultaneously observed to detect dense dust envelopes around protostars. Some of these protostars, along with the protostars observed in the previously mentioned studies, may be responsible for driving the outflows detected in line emission.

Interferometric observations were made with CARMA in the D-array configuration during the period 2011 April and May. We mosaicked a region of approximately $5^{\prime} \times 7^{\prime}$ (oriented with position angle $-30{}^\circ$; see Figure 1) within Serpens South. Our mosaic was centered at R.A. = 18^h30^m00^s, decl. = $-2{}^\circ 2^{\prime} 0^{\prime\prime}$ and consisted of 66 pointings in a hexagonal-packed pattern; the map achieved better than Nyquist sampling with horizontal spacing of $30^{\prime\prime}$ and vertical spacing of $25\buildrel{\prime\prime}\over{.} 8$. In total, we observed 33.8 hr, with 24.6 hr on source. Integrations were 30 s per mosaic position, and 3 minutes on the phase calibrator 1743-038, with 33 pointings observed in between each phase calibrator observation. Uranus was the primary flux calibrator.

We simultaneously observed the J = 1 − 0 transitions of ¹²CO (115.27 GHz) and ¹³CO (110.20 GHz) in the upper side band. The spectral windows of ¹²CO and ¹³CO had widths of 31 MHz ($\sim 80$ km s⁻¹) and 8 MHz ($\sim 20$ km s⁻¹), respectively. With 383 channels in each CO window, the velocity resolutions were 0.2 and 0.06 km s⁻¹, respectively. In addition, we simultaneously observed the continuum emission using eight spectral windows of 495 MHz each, resulting in a map with a total continuum bandwidth of 4 GHz that covers the same region as that of CO. The beam size of $\sim 5^{\prime\prime}$ corresponds to a physical resolution of 0.01 pc, or 2000 AU (at a distance of 429 pc).

Data were reduced using MIRIAD (Sault et al. 1995). Anomalous amplitudes and phases were flagged, as were any observations with system temperatures higher than 500 K. No shadowed data were used, nor were observations when the source was at elevations between 85° and 90°. Flux and gain calibrations were applied, and the data were Fourier transformed.

Images of the data were obtained using the task invert, specifying a cell size of 2'', and then deconvolved with the maximum entropy deconvolution mosmem. The clean images were then restored with a Gaussian beam. We subtracted the continuum image from the molecular line maps using avmaths in the image domain. A summary of the CARMA observations is given in Table 1.

We mapped the same region with the IRAM 30 m telescope in 2012 April. The center of the map is R.A. = ${{18}^{{\rm h}}}{{30}^{{\rm m}}}{{03}^{{\rm s}}}$, decl. = $-02{}^\circ 02^{\prime}$ to match the CARMA map coverage. We mapped the region using individual $2^{\prime} \times 2^{\prime}$ on-the-fly (OTF) maps, with each OTF map scanned twice (in orthogonal directions). The $2^{\prime} \times 2^{\prime}$ maps cover a contiguous region of 38 arcmin² outlined in Figure 1. We tested several off positions, with the best being located at R.A. = 18^h27^m3^s.11, decl. = −02°22'00'' (an offset of −2700'' in R.A. and +1200'' in decl. from the central map position).

Here we present data obtained using Eight MIxer Receiver (EMIR; Carter et al. 2012) with the E090 band and the backend FTS. We tuned to 110.5 GHz in the upper-inner sub-band to detect ¹³CO J = 1 − 0, and we simultaneously observed ¹²CO J = 1 − 0 in the upper-outer sub-band. The velocity resolution was ${\rm \ \Delta\ }v=0.53$ km s⁻¹, and bandwidth was 4000 MHz (or approximately 10,000 km s⁻¹). The beamsizes at the frequencies of ¹²CO J = 1 − 0 and ¹³CO J = 1 − 0 observed with the 30 m telescope are 225 and 235, respectively, corresponding to ∼0.05 pc at d = 429 pc.

We reduced the data using the GILDAS/CLASS software package (the Grenoble Image and Line Data Analysis System).⁶ For ¹²CO J = 1 − 0 we corrected for beam efficiency of $\eta =0.78$ and forward efficiency of 0.95, and for ¹³CO J = 1 − 0 we corrected for beam efficiency of η = 0.79 and forward efficiency of 0.95 (Kramer et al. 2013). We fit a first-order baseline to all spectra and combined all OTF pointings within 12''. The resulting rms noise levels of the final reduced maps are 0.2 and 0.05 K per velocity channel for ¹²CO and ¹³CO maps, respectively.

Deep observations (rms ≈ 0.1 K for ¹²CO, and 0.04 for ¹³CO) at the off position, using the frequency-switch mode, showed that there was in fact significant (∼2 K, with a width of $\sim 1$ km s⁻¹ for ¹²CO) extended emission in the off position at cloud velocities. We added the frequency-switched spectrum from the off-source position to correct the OTF maps at these velocities. Although the emission at the off position may marginally affect the spectra in our map at cloud velocities, at velocities beyond ±3 km s⁻¹ with respect to the ambient cloud velocity (the velocities we consider for outflows) we do not detect significant emission at the off position.

The J = 3 − 2 transition of ¹²CO was observed with APEX in 2009 October. The $\sim 6^{\prime} \times 6^{\prime}$ region centered at R.A. = ${{18}^{{\rm h}}}{{30}^{{\rm m}}}{{03}^{{\rm s}}},$ decl. = $-02{}^\circ 02^{\prime}$ (see Figures 1 and 3) was mapped using the OTF method. The velocity resolution of the observation was ${\rm \ \Delta\ }v=0.2$ km s⁻¹, with a bandwidth of 180 MHz, or 160 km s⁻¹. The beamsize at 345.8 GHz was $19\buildrel{\prime\prime}\over{.} 2$, and the data cube has a pixel size of 9''.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For the purpose of obtaining the opacity for each velocity channel (see Appendix A), the cube was smoothed to have a beamsize of $34\buildrel{\prime\prime}\over{.} 8$ and regridded to a pixel size of 11'' to match the CSO observations (see below). The rms of the smoothed data cube is 0.16 K per 0.2 km s⁻¹ channel.

These data were also used to determine excitation temperatures (see Appendix B) in conjunction with the IRAM ¹²CO J = 1 − 0 data, and for this purpose the APEX data were smoothed to a beamsize of $22\buildrel{\prime\prime}\over{.} 5$ and regridded to a pixel size of 11'' and velocity resolution of ${\rm \ \Delta\ }v=0.53$ km s⁻¹. The resultant rms following this smooth and regrid was 0.13 K per 0.53 km s⁻¹ channel.

The J = 3 − 2 transition of ¹³CO was observed with CSO in 2011 June. Four positions were observed, with the central pointing being R.A. = ${{18}^{{\rm h}}}{{30}^{{\rm m}}}{{03}^{{\rm s}}}$, decl. = $-02{}^\circ 02^{\prime}$ (see Figure 1). The additional three pointings had offsets of (+3'', −47''), (−705, +120''), (−33'', −20'') with respect to the central pointing, selected based on various features seen in ¹²CO emission. The beamsize at 330.6 GHz is 35''. Velocity resolution is ${\rm \ \Delta\ }v=0.06$ km s⁻¹, and the bandwidth was 500 MHz, or 450 km s⁻¹.

For the purpose of obtaining the opacity for each velocity channel, the cube was regridded to match the velocity resolution of the APEX observations, ${\rm \ \Delta\ }v=0.2$ km s⁻¹. The resulting rms is 0.3 K.

Combining interferometer (CARMA) and single-dish (IRAM 30 m) ¹²CO and ¹³CO J = 1 − 0 data recovers extended emission (up to the combined map size of about 0.8 pc) that was filtered out by the interferometry, and it is imperative for accurately determining mass, momentum, and energy of the outflows, presented in Section 3.3. The combination follows the same method as described in Plunkett et al. (2013), corresponding to the nonlinear joint deconvolution method of Stanimirovic (2002). This method is particularly useful for recovering flux in mosaics that cover regions with clumpy structure. The combination was done with MIRIAD. First, we regridded the interferometer map to the velocity resolution of the single-dish map, and then we regridded the pixels of the single-dish map using the interferometer map as the template.

The joint deconvolution was done with the MIRIAD task mosmem using the dirty interferometer map and single-dish (regridded) map as inputs, along with dirty beams for each. The beam sizes were $5\buildrel{\prime\prime}\over{.} 3\times 4\buildrel{\prime\prime}\over{.} 1$ for ¹²CO and $5\buildrel{\prime\prime}\over{.} 3\times 4\buildrel{\prime\prime}\over{.} 5$ for ¹³CO. We converted the single-dish maps to have units of Jy beam⁻¹, in order to match the units of the interferometer maps. Several other parameters were necessary inputs in MIRIAD so that the convolution process converged within a reasonable number of iterations. One such input, "rmsfac," specifies the ratios between the true noise rms (output by MIRIAD using imstat in several emission-free regions) and the theoretical rms (determined by mossen) for both the interferometer and single-dish maps. These factors are necessary in order for mosmem to reduce the residuals in the combined map (i.e., the difference between the dirty image and the model modified by the point-spread function) to have the same rms as the theoretical rms multiplied by the "rmsfac." We used "rmsfac" of 1.6 and 1.5 for the interferometer and single-dish maps, respectively.

Another parameter is the flux calibration factor, or the number by which the single-dish data are multiplied in order to convert to the same scale as the interferometer data. We determined this factor using the task immerge, choosing uv-data in the range 12–25 m (which is the range of "baselines" that the interferometer and single-dish data have in common), and specifying channels where emission structure is comparable in the interferometer and singe-dish maps. We used a factor of 1.1 for the ¹²CO deconvolution and 1.3 for the ¹³CO deconvolution. Using these parameters, the deconvolution converged within 100 iterations for each channel. The final "CLEAN" maps were generated with the task restor. The resulting mean rms sensitivity of the combined CARMA and IRAM maps of ¹²CO and ¹³CO was 0.16 Jy/beam/channel and 0.11 Jy/beam/channel, respectively, with 0.5 km s⁻¹ channel width and beamsize $5\buildrel{\prime\prime}\over{.} 3\times 4\buildrel{\prime\prime}\over{.} 6$ (corresponding to ∼0.01 pc at d = 429 pc). The zeroth-order moment map of the combined CARMA+IRAM ¹²CO data is shown in Figure 3 (without including the ambient velocity channels, and where the blue- and redshifted wing emission are shown in contours), and channel maps of ¹²CO and ¹³CO are shown in Figures 4 and 5, respectively.

Zoom In Zoom Out Reset image size

Using our CARMA continuum observations, we detected seven distinct continuum sources with flux greater than $4\sigma$, and we labeled them in numerical order with increasing R.A. The sources are shown in Figure 2 and given in Table 2. These sources were fit as 2D Gaussians using the MIRIAD task imfit. Using the integrated flux measured with imfit, we calculated masses of the continuum sources according to the following relation (see Schnee et al. 2010):

$$\begin{eqnarray}&&M=\frac{{{d}^{2}}{{S}_{\nu }}}{{{B}_{\nu }}({{T}_{D}}){{\kappa }_{\nu }}},\end{eqnarray} \tag{ 1 }$$

where d is distance to the source, ${{S}_{\nu }}$ is the 2.7 mm continuum flux, and ${{B}_{\nu }}({{T}_{D}})$ is the Planck function. We assumed a dust temperature of T_D = 30 K (Chen et al. 2013) and a dust opacity of ${{\kappa }_{\nu }}=0.1{{(\nu /1200\;{\rm GHz})}^{\beta }}$ cm² g⁻¹, with β = 1, which corresponds to ${{\kappa }_{\nu }}=0.0092$ cm² g⁻¹ at $\lambda =2.7$ mm (Looney et al. 2000). Using the dust opacities extrapolated from Ossenkopf & Henning (1994, Column 6 of Table 1), assuming a gas-to-dust ratio of 100 and β = 1, then $\kappa _{\nu }^{{\rm OH}}(\beta =1)=0.0043$ cm² g⁻¹; extrapolating instead using $\beta =2.0$, then $\kappa _{\nu }^{{\rm OH}}(\beta =2.0)=0.0021$ cm² g⁻¹. In these cases, continuum masses would be about 2–4 times less than what we report, respectively. We note these specifically to remind that without further detailed modeling, our simple mass calculation is uncertain by a factor of a few depending on choices of κ and its spectral index β. Schnee et al. (2014) found a median emissivity spectral index of $\beta =0.9\pm 0.3$, and so our assumption of β = 1 is within their uncertainty.

Given that our continuum map has an rms of 1 mJy, we are able to detect envelopes that have masses greater than 0.04 M_⊙ (at better than a $4\sigma$ level). Of the continuum sources in our map, the sources CARMA-3 and CARMA-6/7 were previously detected in the MAMBO 1.2 mm dust continuum map of Maury et al. (2011), and they also appear to coincide with the central clump detected in the ASTE/AzTEC 1.1 mm continuum emission map (see e.g., Kirk et al. 2013; Tanaka et al. 2013). Maury et al. (2011) classified these sources (which they call SerpS-MM16 and SerpS-MM18) as Class 0/I and Class 0, respectively, based on the envelope mass versus bolometric luminosity (${{M}_{{\rm env}}}-{{L}_{{\rm bol}}}$) evolutionary diagram that they present. Further, Maury et al. (2011) noted that these particular sources are each multiples, although they report quantities that are integrated over each system.

The extended morphologies shown in our 2.7 mm continuum map (see Figure 2) reinforce the notion that the sources CARMA-3 and CARMA-6/7 are each multiples. The source CARMA-3 is seen as an elongated structure with a peak of emission in the southwest and extended emission to the northeast. According to a double-Gaussian fit, the centers of two components in this system are separated by $\sim 7^{\prime\prime}$, or ∼3000 AU not taking into account inclination effects. The total integrated flux of the sources is 39 mJy, with the southwest (CARMA-3 SW) and northeast (CARMA-3 NE) components representing 75% and 25% of the flux, respectively. These components likely share a circumbinary envelope, and since there is no absolute minimum in emission marking a clear delineation between the two components, in Table 2 we also present a single-Gaussian fit and the corresponding total mass for this system. A better understanding of the individual components merits further modeling.

For the source CARMA-6/7, although the multiple components show some overlapping extended emission, we see two distinct peaks of emission, with the northeastern peak (CARMA-7) being the strongest continuum emission in our map. The peaks of emission are separated by $\sim 16^{\prime\prime}$ (∼6500 AU) on the plane of the sky, but dust emission between the sources may pertain to a shared circumbinary envelope. Since we detect two distinct peaks, we fit two Gaussians with imfit, and we determine the masses of the components CARMA-6 and CARMA-7 to be 0.43 and 1.21 M_⊙, respectively.

Two distinct Spitzer MIPS 24 μm sources are coincident with the two peaks of mm continuum emission associated with CARMA-6/7, and the MIPS sources were classified as "red" following Evans et al. (2007). The label "red" indicates that the flux density at 24 μm is at least three times the flux density of the nearest available IRAC band detection in Gutermuth et al. (2008). We suggest that since the MIPS 24 μm sources coincide with the continuum sources, they are likely deeply embedded YSO candidates, even though these sources lack detection in one of the IRAC bands and therefore are not classified as YSO candidates according to the criteria of Evans et al. (2007). CARMA-3 lies about 10'' north of several Spitzer sources, and at least one of those is particularly bright in 4 and 8 μm emission, as can be seen in Figure 2. Upon close inspection of the location of this continuum source, we find weak 4–24 μm emission at the position of CARMA-3, but it is not clear whether the 4–24 μm emission is associated with a stronger source to the south. This may have detracted from detecting a source at the location of CARMA-3 according to the criteria of Evans et al. (2007), as emission from a stronger source could have dominated an underlying weaker source nearby. Additionally, a source seen in mm continuum emission but not 24 μm emission may signify that this source is still IR-dark and therefore at a very early stage of evolution.

Three continuum sources in our map have not been previously published as mm continuum sources, but do coincide with MIPS 24 μm sources. These sources we name CARMA-1, CARMA-4, and CARMA-5 (see Table 2). CARMA-1 and CARMA-5 were classified as YSOc and YSOc_red (Evans et al. 2007), respectively, indicating YSO candidates ("red" is described above). CARMA-5 is located only $\sim 15^{\prime\prime}$ (0.03 pc) from CARMA-3, making it difficult to resolve with previous single-dish observations. Among the continuum sources in this region, CARMA-1 has the largest corresponding MIPS 24 μm flux. The combination of large 24 μm flux and small mass (based on 2.7 mm continuum flux) suggest that this source is among the more evolved mm sources in the map.

Several of the continuum sources that we have discussed here have been referred to by various names in previous literature, and we refer the reader to the right column of Table 2 for these labels. In addition, Teixeira et al. (2012) indicate an IR source in the region to the northeast that they call P4 and they claim drives an outflow, but we detect no continuum counterpart for this source. To our knowledge, the source CARMA-2 has not been identified in previous observations. This source only has two pixels with continuum flux greater than $4\sigma$, but these are surrounded by pixels with flux of 2–3σ (hence, only marginally detected above the rms noise). If this source is indeed a YSO in Serpens South, we calculate a mass of 0.07 M_⊙.

Finally, we also detect unresolved continuum emission associated with the source CARMA-4, for which we cannot confirm whether it is a YSO candidate or a background galaxy detection. First of all, it was classified as a galaxy candidate ("Galc_red") according to Evans et al. (2007). This classification was due to its weak MIPS 24 μm detection of 9.91 ± 1.05 mJy, which is about a factor of at least 10 weaker than the other MIPS sources previously described in this region. However, the coincident continuum emission may be evidence that this is a YSO candidate. If we assume that this source is indeed a YSO in Serpens South, the continuum emission implies an inner envelope/disk mass of 0.07 M_⊙ (see Table 2).

In order to compare with sources detected (or not detected) in previous mm-continuum observations of this region by Maury et al. (2011), we scaled the continuum emission fluxes detected at 2.7 mm (110 GHz) to the expected flux at 1.2 mm (240 GHz). In that case, we expect thermal emission flux to increase by a factor of 9–20 for β = 1–2, respectively. For CARMA-3 and CARMA-6/7—the sources that we detect in common with Maury et al. (2011)—we measure integrated fluxes at 2.7 mm that, when scaled to 1.2 mm assuming β = 1, correspond to 80% and 90%, respectively, of the fluxes measured by Maury et al. (2011). This agreement within uncertainties of the 2.7 and 1.2 mm continuum fluxes, made with interferometer and single-dish telescope observations, respectively, suggests that (1) the majority of emission for these sources arises from thermal dust emission on the Rayleigh–Jeans tail (also described in the Orion Molecular Cloud by Schnee et al. 2014), and (2) the sources are compact sources without significant extended emission that would be filtered out by the interferometer observations. Conversely, assuming β = 2 in the scaling from 2.7 to 1.2 mm results in expected 1.2 mm integrated fluxes that are greater by about a factor of two than those reported by Maury et al. (2011). Without a more detailed study to determine the actual spectral emissivity of these sources, we cannot rule out the possibility that up to half of the 2.7 mm flux that we detect is contributed from non-thermal emission, such as free–free emission.

We also investigate the remaining sources that were not detected in 1.2 mm observations by Maury et al. (2011). Specifically, we detected the sources CARMA-1, CARMA-2, CARMA-4, and CARMA-5 with a peak intensity $\leqslant 7\sigma$, and we scaled the 2.7 mm fluxes to expected 1.2 mm fluxes for comparison. Further, accounting for beam dilution in an 11'' beam, the corresponding flux detected by Maury et al. (2011) would be reduced by at least a factor of about 5 (assuming that an unresolved source fully fills our $5^{\prime\prime}$ beam, which we consider an upper limit for what are likely more compact sources). Therefore, we find that a $1\sigma$ detection in our map corresponds to ∼2–5 mJy/11''-beam at 1.2 mm, whereas Maury et al. (2011) report a $1\sigma$ rms noise of 15 mJy/11''-beam. Therefore, our 2.7 mm continuum source detections that are not in common with those of Maury et al. (2011) are likely due to differences in sensitivity and beamsize of the observations; even a $7\sigma$ detection in our 2.7 mm map would correspond to $\sim 1\sigma$ detection in a 1.2 mm map with an 11'' beam.

In the following section, we refer to several of these continuum sources as potential outflow-driving sources, and in Figure 3 we show continuum emission along with zeroth- moment CO maps. In that figure (also Figures 2 and 8), we overlay the locations of protostars in the region identified in the Spitzer Gould Belt survey (Evans et al. 2007; Gutermuth et al. 2008). We note that while we focus on the continuum sources in the context of our discussion of possible outflow-driving sources, there are many (∼45) densely packed protostars in the mapped region. Specifically, within a radius of 30'' from a point centered between the sources CARMA-3, CARMA-5, CARMA-6, and CARMA-7 (the location that appears to be the approximate apex of the majority of outflow emission) reside 10 protostars identified with Spitzer, and 18 protostars reside within a radius of 1'. Whether some Spitzer sources correspond to weak continuum sources below our detection limit, or do not correspond to continuum sources at all, they nonetheless may contribute to the outflow activity described in Section 3.2. In a future study, we plan to present observations with higher resolution and sensitivity in the central region that will likely give us a better grasp of specific outflow-driving sources, and in that case we plan to discuss the Spitzer and continuum sources further.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Our suite of high spatial resolution mm-wavelength ¹²CO and continuum maps, along with archival Spitzer IRAC images, allows us to scrutinize the outflow morphologies in the densest regions, assess the impact of these outflows on their surrounding environment, suggest probable driving sources where possible, and compare with previous identifications from the literature. Along with the dense concentration of YSOs in the central cluster previously mentioned, the resolution of our observations and the breadth of the outflow emission make it a challenge to identify individual driving sources. In this section we describe the CO outflow morphologies that we observed at particular velocity ranges, and we present the quantitative characteristics of these outflows in Section 3.3.

We accompany the following discussion with references to several CO maps in Figures 3–9, beginning with Figure 3 showing zeroth-moment maps of ¹²CO J = 1 − 0 and J = 3 − 2 emission. In Figure 4, we show ¹²CO J = 1 − 0 velocity-binned maps, summing channels in variable-sized bins so that near the cloud velocity we sum over two consecutive channels (${\rm \ \Delta\ }v=1.1$ km s⁻¹), and for high outflow velocities we sum over 10 channels (${\rm \ \Delta\ }v=5.8$ km s⁻¹). Additionally, ¹³CO emission is shown in a similar manner in Figure 5, while featuring especially the lower-velocity channels where ¹³CO emission is detected. Figure 6 shows ¹³CO overlaid with ¹²CO contours for certain velocity ranges. Figures 7–8 are moment maps of low-, mid-, and high-velocity outflow emission probed by ¹²CO J = 1 − 0 CARMA+IRAM observations. Velocity ranges shown in those maps are asymmetric about the cloud velocity in order to reveal particular features that were identified by close inspection of the data cube, and those features are discussed specifically in the text in the following sections. An outflow candidate is shown in a similar way (but at lower outflow velocities) in Figure 9 and discussed in Section 3.2.3.

Zoom In Zoom Out Reset image size

Throughout, we use the notation of v_LSR to indicate channel velocity (relative to local standard of rest), v_cloud for cloud velocity, and ${{v}_{{\rm out}}}={{v}_{{\rm LSR}}}-{{v}_{{\rm cloud}}}$ for outflow velocity (along the line of sight). We adopt a systemic cloud velocity of ${{v}_{{\rm cloud}}}$ = 8 km s⁻¹ pertaining to the region we mapped. Previous works (e.g., Nakamura et al. 2011a; Kirk et al. 2013) adopted a cloud-wide velocity centroid of ${{v}_{{\rm cloud}}}=7.5$ km s⁻¹, but Figure 4 of Kirk et al. (2013) shows that in the region we mapped (centered at R.A. = 18^h30^m, decl. = $-2{}^\circ 2^{\prime}$), there is a predominance of a larger centroid velocity of approximately ${{v}_{{\rm cloud}}}=8$ km s⁻¹.

3.2.1. Blueshifted Outflows

3.2.2. Redshifted Outflows

3.2.3. Outflow Emission near CARMA-1

The mass, momentum, and energy of outflowing gas were calculated in order to assess the overall contribution of outflow activity within the mapped region, and we compared these with turbulent momentum and energy and gravitational potential energy. First, we calculated H₂ column density (${{N}_{{{{\rm H}}_{2}}}}$) for each pixel ($x,y$) and velocity channel (with channel width ${\rm \ \Delta\ }v$) as a function of excitation temperature (T_ex) according to the following relations (following Dunham et al. 2014):

$$\begin{eqnarray}&&{{N}_{{{{\rm H}}_{2}}}}(x,y,v,{{T}_{{\rm ex}}})=f(J,{{T}_{{\rm ex}}},{{X}_{{\rm CO}}})T_{{\rm mb},12}^{*}{\rm \ \Delta\ }v\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&f(J,{{T}_{{\rm ex}}},{{X}_{{\rm CO}}})=X_{{\rm CO},12}^{-1}\frac{3k}{8{{\pi }^{3}}\nu {{\mu }^{2}}}\frac{2J+1}{J+1}\frac{Q({{T}_{{\rm ex}}})}{{{g}_{L}}}{{e}^{\frac{{{E}_{J+1}}}{k{{T}_{ex}}}}}.\end{eqnarray} \tag{ 3 }$$

The opacity correction is explained in Appendix A, with the main-beam temperature of ¹²CO corrected for optical depth, $T_{{\rm mb},12}^{*}$, given by Equation (A4). Appendix B describes the method(s) to solve for excitation temperature, T_ex. In the above equations, J = 0 is the lower energy level of the rotational transition, and ${{g}_{J}}=2J+1=1$ is the degeneracy of the lower state. For X_CO we assume a standard abundance ratio ${{[}^{12}}{\rm CO}/{{{\rm H}}_{2}}]={{10}^{-4}}$ (Frerking et al. 1982). For ¹²CO J = 1 − 0, $\nu =115.27$ GHz, and μ is the ¹²CO dipole moment with a value 0.11 Debye. $Q({{T}_{{\rm ex}}})$ is the partition function, which here is approximated as $Q({{T}_{{\rm ex}}})=k{{T}_{{\rm ex}}}/hB$, and B = 57.635 GHz is the rotation constant for CO. ${{E}_{J+1}}=7.64\times {{10}^{-16}}$ erg is the energy of the upper level such that the energy of the transition is $E={{E}_{J+1}}-{{E}_{J}}$, and ${{E}_{J+1}}/k=5.53$ K.

Having measured column density for each pixel, mass is then determined to be

$$\begin{eqnarray}&&M(x,y,v)=2.72{{m}_{{\rm H}}}{{N}_{{{{\rm H}}_{2}}}}(x,y,v)A,\end{eqnarray} \tag{ 4 }$$

where m_H is the mass of a hydrogen atom, such that 2.72m_H is the mean molecular weight of the gas (Bourke et al. 1997), and A is the area of a pixel. We sum over the cube in spatial and spectral dimensions including pixels that have signal greater than 3σ to get the total outflow mass, momentum, and energy:

$$\begin{eqnarray}&&{{M}_{{\rm out}}}={{{\rm \ \Sigma\ }}_{{\rm area}}}{{{\rm \ \Sigma\ }}_{{\rm vel}}}M(x,y,v)\end{eqnarray} \tag{ 5 }$$

$$\begin{eqnarray}&&{{P}_{{\rm out}}}={{{\rm \ \Sigma\ }}_{{\rm area}}}{{{\rm \ \Sigma\ }}_{{\rm vel}}}M(x,y,v)|v-{{v}_{{\rm cloud}}}|\end{eqnarray} \tag{ 6 }$$

$$\begin{eqnarray}&&{{E}_{{\rm out}}}=\frac{1}{2}{{{\rm \ \Sigma\ }}_{{\rm area}}}{{{\rm \ \Sigma\ }}_{{\rm vel}}}M(x,y,v)|v-{{v}_{{\rm cloud}}}{{|}^{2}},\end{eqnarray} \tag{ 7 }$$

where v is the velocity corresponding to a given channel, and ${{v}_{{\rm cloud}}}$ = 8 km s⁻¹ is the cloud velocity. Mass, momentum, and energy for velocity channels with $3\lt |{{v}_{{\rm out}}}|\lt 31$ km s⁻¹ and the region we mapped covering ∼45 arcmin² (see Figure 1) are given in Table 3. The lower limit of this velocity range was chosen by inspecting low-velocity channels by eye to exclude those channels dominated by ambient cloud emission, and the upper limit was dictated by the spectral coverage of our observations. We measure a total outflow mass of 2.8 M_⊙, of which blueshifted and redshifted channels contribute 43% and 57%, respectively. These masses were calculated using the ${{T}_{{\rm ex}}}(x,y,v)$ values determined with the pixel-by-pixel + channel-by-channel (velocity-dependent) method as described in Appendix B (a.k.a. method 1). Using excitation temperatures determined according to the alternative methods 2–5 (these methods either spatially average over pixels in each channel or assume a constant excitation temperature, as described in Appendix B) results in total outflow masses ranging from 2.2 to 7.2 M_⊙, listed in Table 3.

Table 3.  Outflow Mass, Momentum, Energy

Method	Mass			Momentum	Energy
	All (M $_{\odot }$)	Bluea (M $_{\odot }$)	Redb (M $_{\odot }$)	(M $_{\odot }$ km s−1)	(×1045 erg)
(1) Tex(x, y, v), pixel-by-pixel + channel-by-channelc	2.8	1.2	1.6	13.6	1.0
(2) Tex(v), channel-by-channelc	2.9	1.2	1.7	13.9	1.0
(3) Tex = 10 K	2.2	1.1	1.2	10.8	0.8
(4) Tex = 30 K	4.6	2.3	2.4	22.4	1.7
(5) Tex = 50 K	7.2	3.5	3.7	34.7	2.6

^aBlue channels correspond to $|{{v}_{{\rm out}}}|$ = 3.3–30.9 km s⁻¹. ^bRed channels correspond to $|{{v}_{{\rm out}}}|$ = 3.0–30.6 km s⁻¹. ^cMeans: $\overline{{{T}_{{\rm ex}}}}(x,y,v)=17\pm 8$ K for (1), and $\overline{{{T}_{{\rm ex}}}}(v)=15\pm 4$ K for (2).

Download table as:  ASCIITypeset image

In Figure 10 we show the outflow mass distribution per velocity bin for Serpens South (panels a and b). For the blue- and redshifted mass spectra, we fit power laws of the form $m(v)\propto {{v}^{-\gamma }}$ for the velocity range $3\lt |{{v}_{{\rm out}}}|\lt 15$ km s⁻¹, and we found slopes of $\gamma =3.5\pm 0.1$ and $\gamma =3.3\pm 0.1$, respectively. The slope may be an evolutionary indicator of mass ejection in a cloud, such that the slope steepens with time (Smith et al. 1997). We performed a similar analysis of the outflow mass within the slightly more evolved region NGC 1333 (Plunkett et al. 2013), assuming ${{T}_{{\rm ex}}}=17$ K, which is the mean T_ex in Serpens South, and the blue- and redshifted mass spectra for that region can be fit with power laws with slopes of $\gamma =5.1\pm 0.2$ and $\gamma =4.9\pm 0.1$ (see Figure 10, panels a and c). These two regions are therefore presented as examples for which the shallower (steeper) slope pertains to the less (more) evolved region.

Zoom In Zoom Out Reset image size

It has been previously suggested that a break in γ at $\sim 10$ km s⁻¹can be interpreted as two distinct outflow velocity components (e.g., Richer et al. 2000; Su et al. 2004). In that case, the high-velocity gas corresponds to a recently accelerated component and steeper slope (γ ∼ 3–4), and the low-velocity gas is a slower, coasting component with a less steep slope ($\gamma \sim 2$). Neither Serpens South nor NGC 1333 showed a break in their power spectra according to our primary analysis. We note that our analysis included a velocity-dependent correction for opacity of the ¹²CO line, which especially accounted for more "hidden" mass at low velocities than in the case of the optically thin assumption. When assuming that the ¹²CO line is optically thin ($\tau \ll 1$), breaks in the power spectra are apparent at outflow velocities of ∼5–10 km s⁻¹ (see Figures 10(b)–(c)). Our findings are in line with the suggestion by Bally et al. (1999) that a broken power-law index for outflows may not be a physical distinction, but rather an artifact of the optically thin assumption.

Comparing blue- and redshifted mass at each outflow velocity, the ratio ${{M}_{{\rm out},{\rm blue}}}/{{M}_{{\rm out},{\rm red}}}(|{{v}_{{\rm out}}}|)$ ranges from 0.1 to 1.5, with a mean of 0.5. While we expect outflows to be bipolar, and hence the blue- and redshifted gas should total approximately equal amounts, in this map it appears that slightly more gas is being entrained by redshifted lobes than by blueshifted lobes, especially in velocity channels with $|{{v}_{{\rm out}}}|$ ∼ 5–10 km s⁻¹. Possibly some outflow gas has escaped the cloud more readily in the foreground than in the background, or extends beyond the map edges, depending on the locations of the driving protostars within their nascent gas cloud. This mass ratio may also entail that the surrounding gas environment was not homogeneous in mass and/or density at the time when protostars began to drive outflows. We present maps of T_ex in Appendix B, and the highest T_ex values are found in redshifted velocity channels.

In addition to the mass calculation, we calculate outflow momentum to be ${{P}_{{\rm out}}}$ = 13.6 M_⊙ km s⁻¹ and energy to be ${{E}_{{\rm out}}}\sim 1.0\times {{10}^{45}}$ erg. Table 3 lists these diagnostics using each T_ex method, and we consider these values to be lower limits because they are uncorrected for inclination angle of the outflows. Assuming an average inclination angle with respect to the line of sight $\xi =57\buildrel{\circ}\over{.} 3$ (e.g., Nakamura et al. 2011a; Dunham et al. 2014), and assuming that all outflow motion is along the jet axis (see Downes & Cabrit 2007), results in greater momentum and energy by factors of 1.9 and 3.4, respectively. Hence, momentum and energy from method 1 are ${{P}_{{\rm out}}}$ = 25.3 M_⊙ km s⁻¹ and ${{E}_{{\rm out}}}=3.3\times {{10}^{45}}$ erg, after correction for an average inclination angle of $\xi =57\buildrel{\circ}\over{.} 3$.

We performed a comparison of the cumulative mass and energy of outflows with those determined by Nakamura et al. (2011a) using ¹²CO J = 3 − 2 assuming (1) constant T_ex = 30 K, (2) distance of 260 pc, and (3) including only the velocity channels between $-15$ to $4$ km s⁻¹ and $11$ to $30$ km s⁻¹, with a cloud velocity ${{v}_{c}}=7.5$ km s⁻¹. We found the cumulative mass and energy of outflows to be 1.8 and 1.1 times the values presented by Nakamura et al. (2011a), respectively. In addition to using a different energy level transition, this discrepancy may also be attributable to the correction we make for opacity of CO, whereas Nakamura et al. (2011a) assume that ¹²CO J = 3 − 2 is optically thin. On the other hand, we also note that the map coverages are not identical, and specifically the blue features B7, B8, and B9 from Nakamura et al. (2011a) are not covered entirely by our map. Nonetheless, we consider these masses and energies to agree within the uncertainties that we cannot account for here.

As previously mentioned (see Section 3.3 for more discussion), we have assumed a minimum outflow velocity of $|{{v}_{{\rm out}}}|=3$ km s⁻¹, so as not to confuse emission from the ambient cloud. However, as Dunham et al. (2014) suggest, typical escape velocities from individual sources are less than this, and therefore our mass calculation potentially omits some outflowing gas. Making an admittedly simple extrapolation to velocities as low as ${{v}_{{\rm out}}}=0.3$ km s⁻¹, based on the spectral fit shown in Figure 10, we find that the total mass may increase by an order of magnitude or more, which was also shown by Downes & Cabrit (2007), Offner et al. (2011), and Dunham et al. (2014) for low-velocity outflows. When assessing the impact that outflows have on the cloud environment, the momentum and energy are more appropriate diagnostics than mass, as they will be less affected by the low-velocity channels (i.e., a factor of a few when extrapolating based on Figure 10) so that the main conclusions that we propose below will hold qualitatively.

Conversely, the high-velocity outflows contribute more significantly to the momentum and energy budgets, rather than mass. The velocity range that we probe here, with $|{{v}_{{\rm out}}}|\lt 31$ km s⁻¹, is mostly representative of the SHV outflow component. However, EHV components have been shown to be driven by some low-mass protostars (e.g., Tafalla et al. 2004), and if present they can account for 2–4 times as much momentum and energy as the SHV components. In the following sections we compare the relative contributions of momentum and energy from outflows, turbulence, and gravity of the cloud. Although the magnitude of the relative contribution by outflows may change if in fact there are EHV components associated with these sources, our discussion and conclusions (below) that outflows significantly impact the surrounding environment hold. Nonetheless, given that we do not probe the entire range of low- and high-velocity outflow gas, it should be kept in mind that we likely underestimate the mass and subsequent mass-dependent kinematic properties expressed in Sections 4.1–4.3.

In order to characterize the outflows in this region, we present several common outflow properties in Table 4. We list the dependence of each property on assumed inclination in column 2, where ξ is the angle between the outflow axis and the line of sight (i.e., $\xi =90{}^\circ$ corresponds to an edge-on system). For each property, we present values corresponding to the lower and upper limits of the characteristic length (columns 3–4); we also calculate the values corresponding to the median characteristic length or dynamical time, both uncorrected for inclination angle (column 5) and assuming an average inclination angle of $\langle \xi \rangle =57\buildrel{\circ}\over{.} 3$ (column 6). The respective units are shown in column 7.

Table 4.  Outflow Properties

Property	Inclination	Limitsa		Mediansb		Units
	Dependence	Lower	Upper	Uncorrected	$\langle \xi \rangle =57\buildrel{\circ}\over{.} 3$
Characteristic length (${{R}_{{\rm char},{\rm out}}}$)	$1/{\rm sin} \xi$	0.1	0.4	0.19	0.22	(pc)
Characteristic velocity (${{V}_{{\rm char},{\rm out}}}$)	$1/{\rm cos} \xi$	⋯	⋯	4.8	8.9	( km s−1)
Dynamical time (tdyn)	${\rm cos} \xi /{\rm sin} \xi$	1.9	8.9	3.8	2.4	(×104 yr)
Mass-loss rate ($\dot{M}$)	${\rm sin} \xi /{\rm cos} \xi$	1.5	0.3	0.7	1.2	($\times {{10}^{-4}}$ M⊙ yr−1)
Outflow force (Fout)	${\rm sin} \xi /{{{\rm cos} }^{2}}\xi$	7.2	1.5	3.6	10.3	($\times {{10}^{-4}}$ M⊙ km s−1 yr−1)
Outflow luminosity (Lout)	${\rm sin} \xi /{{{\rm cos} }^{3}}\xi$	42.2	9.0	21.1	112.6	($\times {{10}^{-2}}{{L}_{\odot }}$)

^a"Lower" and "upper" correspond to outflow property listed in first column obtained using the lower and upper limits of R_char (row 1), respectively, and uncorrected for inclination angle. ^bCorrespond to the outflow properties listed in first column obtained using the median R_char value (row 1), with "uncorrected" indicating raw values without correcting for inclination angle, and $\langle \xi \rangle =57\buildrel{\circ}\over{.} 3$ is the assumed average outflow inclination angle with respect to the line of sight.

Download table as:  ASCIITypeset image

The first outflow property is the characteristic outflow length (${{R}_{{\rm char},{\rm out}}}$), and this value propagates in subsequent calculations. In Section 3.2, we discussed several coherent outflow features that have lengths of $45^{\prime\prime}$–$210^{\prime\prime}$ (${{R}_{{\rm char},{\rm out}}}=0.1$ pc to 0.4 pc, respectively) and a median of $90^{\prime\prime}$ (${{R}_{{\rm char},{\rm out}}}=0.19$ pc, or 0.22 pc for $\xi =57\buildrel{\circ}\over{.} 3$). Based on the morphology of the emission detected here, we suggest that the majority of the outflows that we discussed originate from the YSOs near the dense cluster center in the map, and we assume that the previously mentioned values for ${{R}_{{\rm char},{\rm out}}}$ are the limits and median lengths representative of outflows found in this region. We use these values in further diagnostic calculations, although it has also been shown that outflow lengths vary by at least an order of magnitude in other regions (e.g., Nakamura et al. 2011b; Plunkett et al. 2013), so more extreme values may be feasible. Nonetheless, our median value and range that we present in Table 4 are consistent with the literature, especially Nakamura et al. (2011a), who estimated the mean length of outflow lobes in Serpens South to be 0.1 –0.2 pc, or 0.17–0.33 pc at the assumed distance of 429 pc.

Next, we calculated the characteristic velocity (${{V}_{{\rm char},{\rm out}}}$), which is the total outflow momentum divided by outflow mass in the region, and we found that ${{V}_{{\rm char},{\rm out}}}={{P}_{{\rm out}}}/{{M}_{{\rm out}}}=4.8$ km s⁻¹. Correcting for an inclination angle $\xi =57\buildrel{\circ}\over{.} 3$ resulted in ${{V}_{{\rm char},{\rm out}}}=8.9$ km s⁻¹. The resulting median dynamical time defined as ${{t}_{{\rm dyn}}}={{R}_{{\rm char},{\rm out}}}/{{V}_{{\rm char},{\rm out}}}$ is $3.8\times {{10}^{4}}$ yr ($2.4\times {{10}^{4}}$ yr for $\xi =57\buildrel{\circ}\over{.} 3$), with a range of (1.9–8.9) × ${{10}^{4}}$ yr for the ${{R}_{{\rm char},{\rm out}}}$ range mentioned previously. The remainder of diagnostics in Table 4 pertain to outflow mass, momentum, and energy, and we explain those diagnostics in the following sections.

In the model of Nakamura & Li (2007), outflows balance infall motions and accretion, sustain the cloud in quasi-static equilibrium, and lead to a slow SFR. Without outflows to inject additional turbulent motions in this model, the turbulence would dissipate in less than several free-fall times. Therefore, the following calculations aim to quantify momentum and energy injection by outflows, specifically in the context of infall and turbulence in the same region. The calculations of mass-loss rate, outflow force, and luminosity (see rows 4–6 of Table 4) are inversely proportional to the dynamical time, which is a source of uncertainty of a factor of a few, as explained in the previous section. We report the median values in the following text, but Table 4 also includes a range that corresponds to the lower and upper limits for t_dyn (uncorrected for inclination angle), as well as median values, both uncorrected and corrected for an average inclination angle of $\langle \xi \rangle =57\buildrel{\circ}\over{.} 3$.

Reipurth & Bally (2001) explained that the "unified" outflow model relates outflows with jets and/or winds, such that the same primary engine powers a YSO's high-velocity jet/wind as well as lower-velocity lobes of an outflow traced by CO emission. Hence, the cumulative mass-loss rate via outflows and the mass-loss rate via winds can be expressed according to ${{\dot{M}}_{{\rm out}}}{{V}_{{\rm char},{\rm out}}}={{\dot{M}}_{{\rm wind}}}{{V}_{{\rm wind}}}$. From our CO observations, we found the cumulative mass-loss rate via molecular outflows (${{\dot{M}}_{{\rm out}}}={{M}_{{\rm out}}}/{{t}_{{\rm dyn}}}$) to be $1.2\times {{10}^{-4}}$ M_⊙ yr⁻¹ (assuming $\xi =57\buildrel{\circ}\over{.} 3$). It follows that the cumulative wind mass-loss rate is ${{\dot{M}}_{{\rm wind}}}=1.0\times {{10}^{-5}}$ M_⊙ yr⁻¹, assuming ${{V}_{{\rm wind}}}=100$ km s⁻¹(with a likely range of ${{V}_{{\rm wind}}}=100$–200 km s⁻¹, see Arce et al. 2007; Bally et al. 2007, and references therein). Further, it has been shown that mass-loss rates are tied to accretion rates (Hartigan et al. 2000). Based on the wind mass-loss rate, and assuming the fraction of stellar mass launched in a wind to be ${{f}_{w}}$ ∼ 0.1–0.3 (Shu et al. 1988; Pelletier & Pudritz 1992; Lee et al. 2007a, 2007b), we expect the total mass accretion rate onto the stars powering outflows in this region to be (∼0.3–1.0) × ${{10}^{-4}}$ M_⊙ yr⁻¹ (or lower by a factor of 2 if we assume the larger ${{V}_{{\rm wind}}}=200$ km s⁻¹).

Kirk et al. (2013) estimated the infall rate of material onto the cluster-forming clump to be about ${{\dot{M}}_{{\rm infall}}}\sim 2.2\times {{10}^{-4}}$ M_⊙ yr⁻¹ (assuming filament inclination angle that they find to be 12°, and scaling to a distance of 429 pc). We can relate this quantity with the total accretion rate onto stars within the cluster using ${{\dot{M}}_{{\rm acc}}}={{f}_{{\rm SFE}}}{{\dot{M}}_{{\rm infall}}}$, with f_SFE typically in the range of 0.1–0.3 (Lada & Lada 2003), since not all the material that is added to the clump through gravitational infall will be used to make protostars (and disks). Assuming f_w = 0.1 and ${{f}_{{\rm SFE}}}=0.3$, we then estimated the infall rate for this region to be $\sim 3\times {{10}^{-4}}$ M_⊙ yr⁻¹, which is comparable to the infall rate onto the cluster found by Kirk et al. (2013). Hence, the cumulative mass-loss rate via molecular outflows that we measured and the observed infall rate appear to be consistent in Serpens South.

In addition to assessing the outflowing mass over time, we calculated the total outflow momentum per unit time, known as the cumulative outflow force (${{F}_{{\rm out}}}={{P}_{{\rm out}}}/{{t}_{{\rm dyn}}}$), to be $10.3\times {{10}^{-4}}$ M_⊙ km s⁻¹ yr⁻¹ (assuming $\xi =57\buildrel{\circ}\over{.} 3$). Further, the cumulative outflow luminosity (${{L}_{{\rm out}}}={{E}_{{\rm out}}}/{{t}_{{\rm dyn}}}$) is $1.13\ {{L}_{\odot }}$. For comparison, Nakamura et al. (2011a) reported a range ${{L}_{{\rm out}}}=0.8$–$1.7\ {{L}_{\odot }}$ (corrected for a distance of 429 pc), assuming an inclination angle of $\xi =57\buildrel{\circ}\over{.} 3$ and a range of characteristic lengths ${{R}_{{\rm out}}}=0.1$–$0.2$ pc.

Comparing the turbulence energy dissipation rate and the outflow luminosity quantifies the contribution of outflows and their impact on the cloud turbulence. In order to determine this, first we solved for turbulent energy, which is defined as ${{E}_{{\rm turb}}}=\frac{1}{2}{{M}_{{\rm cl}}}\sigma _{3{\rm D}}^{2}$. Throughout, we adopted a cloud mass of ${{M}_{{\rm cl}}}$ = 220 M_⊙ and a radius of ${{R}_{{\rm cl}}}=0.17$ pc, based on observations of HCO⁺ ($J=4-3$) by Nakamura et al. (2011a) and adjusting for a distance of 429 pc; the mass is consistent within $\sim 20\%$ of that estimated by Kirk et al. (2013) for the central cluster. Also, 3D and 1D velocity dispersions were related according to the following relations: ${{\sigma }_{3{\rm D}}}=\sqrt{3}{{\sigma }_{v}}$ and ${{\sigma }_{v}}={\rm \ \Delta\ }{{v}_{{\rm turb}}}/(2\sqrt{2{\rm ln} 2})$. The velocity dispersion ${\rm \ \Delta\ }{{v}_{{\rm turb}}}=1.36$ km s⁻¹ (M. Fernández-López, 2014, private communication) was determined from the FWHM of N₂H⁺, which traces mostly dense and quiescent gas. The resulting turbulent energy within the mapped region was found to be $2.2\times {{10}^{45}}$ erg.

The turbulence energy dissipation rate is given by ${{L}_{{\rm turb}}}={{E}_{{\rm turb}}}/{{t}_{{\rm diss}}}$. The dissipation time is defined as ${{t}_{{\rm diss}}}={{t}_{{\rm ff}}}(3.9\kappa /{{M}_{{\rm rms}}})$ (see Mac Low 1999), where free-fall time is ${{t}_{{\rm ff}}}=\sqrt{3\pi /(32G\rho )}$, and $\kappa ={{\lambda }_{d}}/{{\lambda }_{J}}\sim 1$ assuming that the turbulence driving length (${{\lambda }_{d}}$) is equal to the Jean's length (${{\lambda }_{J}}$) of the clump. We determined the gas density of the cloud ($\rho =3{{M}_{{\rm cl}}}/4\pi R_{{\rm cl}}^{3}$) to be $\rho =8\times {{10}^{-19}}$ g cm⁻³. The Mach number is ${{M}_{{\rm rms}}}={{\sigma }_{3{\rm D}}}\sqrt{2.72{{m}_{{\rm H}}}/(kT)}=5.5$, with T = 10.8 K (Friesen et al. 2013). The resulting turbulent energy dissipation rate is ${{L}_{{\rm turb}}}=0.3\ {{L}_{\odot }}$. The ratio ${{r}_{L}}={{L}_{{\rm out}}}/{{L}_{{\rm turb}}}$ ∼ 0.6–3 (the range corresponds to L_out uncorrected for inclination angle, and assuming $\xi =57\buildrel{\circ}\over{.} 3$) quantifies the potential for outflows to counteract turbulence dissipation in the region. A value of at least 1 indicates that outflows drive energy comparable to or greater than the turbulent energy in the region, as is the case here after correcting for inclination angle.

Another way to assess the potential impact of outflows on the cloud turbulence is to compare the total outflow force and the turbulence momentum dissipation rate. In the outflow-regulated cluster formation model, the outflow force should be comparable to or greater than the turbulence momentum dissipation rate in order for outflow feedback to replenish internal supersonic turbulence and maintain the clump close to quasi-virial equilibrium (Li & Nakamura 2006; Nakamura & Li 2014). Turbulence momentum dissipation rate of the internal turbulent motion is defined as $d{{P}_{{\rm turb}}}/dt=-0.21{{M}_{{\rm cl}}}{{\sigma }_{3{\rm D}}}/{{t}_{{\rm diss}}}$ (Equation (4) of Nakamura & Li 2014). We found $d{{P}_{{\rm turb}}}/dt=-2.8\times {{10}^{-4}}$ M_⊙ km s⁻¹ ${\rm y}{{{\rm r}}^{-1}}$, which is approximately 25% of the median outflow force (assuming $\xi =57\buildrel{\circ}\over{.} 3$). Even accounting for a factor of a few uncertainty in the outflow force that we mention above, the ratio of $d{{P}_{{\rm turb}}}/dt$ to F_out likely remains less than 1. Assuming that outflows can transfer momentum and energy to a substantial fraction of the gas in the immediate vicinity of the dense cluster of young protostars, the ratio of outflow force to turbulence momentum dissipation along with the ratio r_L (see above) suggest that outflows are an important source of turbulence in this region.

We compared the outflow energy with the gravitational binding energy of the clump to investigate whether the outflows could potentially unbind the cloud. Using the previously mentioned cloud mass and radius, the gravitational binding energy ($W=-GM_{{\rm cl}}^{2}/{{R}_{{\rm cl}}}$) of the central region is $-2.5\times {{10}^{46}}$ erg. The cloud mass is about 50 times greater than the outflow mass, while the gravitational binding energy is about 7 times greater than the outflow energy, assuming an inclination angle of $\xi =57\buildrel{\circ}\over{.} 3$ (see Table 3). In order to quantify whether outflows have sufficient energy to overcome gravity and disperse the surrounding gas, Nakamura & Li (2014) utilized the non-dimensional parameter ${{\eta }_{{\rm out}}}\equiv -2{{E}_{{\rm out}}}/W$. We found that ${{\eta }_{{\rm out}}}=0.3$ (assuming $\xi =57\buildrel{\circ}\over{.} 3$), which implies that outflow feedback is not sufficient to provide enough kinetic energy to unbind the entire cloud.

Serpens South is the second region observed as part of a study of clustered star formation; in Plunkett et al. (2013) we mapped the prototypical protostellar cluster NGC 1333 utilizing CARMA and FCRAO observations of CO J = 1 − 0. Given the consistency among the observations, here we compare Serpens South and the more evolved NGC 1333, and we present an empirical scenario of outflow-cluster interaction at different evolutionary stages. The comparisons that we discuss below are also summarized in Table 5.

Table 5.  Comparison of Serpens South and NGC 1333

Property	Serpens Southa	NGC 1333a	Units
Protostellar number density	130	110	(pc−2)
Protostellar fraction	80	50	(%)
Outflow mass per area	9.7	26	(M⊙ pc−2)
Outflow emission prevalenceb	90	60	(%)
${{r}_{L}}\;(={{L}_{{\rm out}}}/{{L}_{{\rm turb}}})$	0.6–3c	6–24c	⋯
${{\eta }_{{\rm out}}}\ (=-2{{E}_{{\rm out}}}/W)$	0.1–0.3c	0.2–0.7c	⋯

^aPertains to area enclosed in white rectangle in Figure 2 for Serpens South, and region mapped by Plunkett et al. (2013) for NGC 1333. ^bOutflow emission coverage (pixels projected onto the plane of the sky) with respect to a mapped area of 0.3 pc². ^cThe given lower and upper limits of ranges correspond to (1) raw values without correcting for inclination angle, and (2) assuming $\langle \xi \rangle =57\buildrel{\circ}\over{.} 3$ as the average outflow inclination angle with respect to the line of sight, respectively.

Download table as:  ASCIITypeset image

Gutermuth et al. (2008) suggested that the high percentage of YSOs that are protostars is evidence of Serpens South's youth. Here we present a tally of YSOs in Serpens South using the most recent classifications following Evans et al. (2007) and Maury et al. (2011), allowing a consistent comparison with NGC 1333. We count Class 0, Class I, and flat-SED sources as protostellar sources, and these along with Class II/III sources comprise YSOs. We count 45 protostellar sources and a total of 55 YSOs within the ∼22 arcmin² (0.3 pc²) region of Serpens South that we show outlined with a white rectangle in Figure 2, of which 32 protostars and a total of 37 YSOs reside within the ∼8 arcmin² (0.1 pc²) inner most cluster region outlined by the red rectangle in Figure 2. We compare this to the 26 protostellar sources out of a total of 55 YSOs in the 0.23 pc² region of NGC 1333 mapped by Plunkett et al. (2013).

The protostellar number densities (number of protostars per unit area mapped) are 130 and 110 pc⁻², respectively, for Serpens South and NGC 1333, showing that both of the regions we mapped are very densely populated with young members but about 20% more so in Serpens South. Further, in Serpens South the protostellar fraction (the ratio of protostellar sources to YSOs) is about 80%, whereas in NGC 1333 this percentage is 50%. We also note that within the central region, outlined by the red rectangle in Figure 2(b), the protostellar fraction of Serpens South rises to 90%; the central region corresponding to the same physical area in NGC 1333 maintains a protostellar fraction of ∼50%. The higher percentage of young sources in Serpens South suggests that the majority of sources in this region have only recently begun to drive outflows that impact the surrounding cloud environment, compared with NGC 1333.

In NGC 1333, Plunkett et al. (2013) identified outflows associated with 6 of 11 Class 0 sources and 2 of 11 Class I sources, but no outflows were associated with more evolved sources. This corresponds to 8 of 26 protostellar sources driving identifiable outflows, and at least four candidate outflows had unidentifiable driving sources. It is worth noting that NGC 1333 is closer (235 pc) than Serpens South (429 pc, see Section 1.2), and its outflow features are more clearly distinguishable such that we were able to associate outflows with their driving sources. This is not the case for Serpens South, which is farther and has a much more complex web of interspersed outflows, as seen in Figure 3 and elsewhere, so our comparisons here are in terms of cumulative outflow activity. In total, Plunkett et al. (2013) measured 6 M_⊙ of gas associated with outflow lobes or outflow candidates in NGC 1333 within a region of ∼0.23 pc², corresponding to ∼26 M_⊙ pc⁻². In Serpens South, we measure ∼2.8 M_⊙ of CO gas within a region of ∼0.34 pc², or ∼10 M_⊙ pc⁻². Granted, a comparison of mass entrained by outflows relies on the assumption that the initial conditions of the cluster before star formation began are also comparable.

Spatially, outflow emission is very prevalent in both regions. To quantify the fraction of the mapped area with outflow emission, we measured how many pixels (projected onto the plane of the sky) have ¹²CO signal-to-noise ratio greater than 3 in at least three velocity channels ("outflow emission prevalence" in Table 5). In the Serpens South map, 80% of the mapped region has emission in at least three velocity channels. For comparison, we scaled emission in the NGC 1333 map from Plunkett et al. (2013) to a distance of 429 pc and the same rms noise threshold as the Serpens South data. In a region corresponding to the same physical area of the Serpens South map we measured emission (again in at least three velocity channels) spanning about 60% of the mapped area on the plane of the sky. The simplest explanation for this is that the number density of protostellar sources, assumed to drive outflows, is higher in the central-most region of Serpens South than in NGC 1333. A more detailed study of individual outflows in Serpens South is needed to develop this argument. Considering that NGC 1333 is a more evolved protostellar region, we expect some of the outflowing material to have escaped that region, whereas in Serpens South the material has had less time to travel beyond the central cluster that we mapped. This interpretation of outflow emission prevalence compared between the two regions assumes that outflows are randomly oriented in both regions. Alternatively, if large-scale filaments and gas kinematics cause preferential orientations of the outflows and/or their driving sources, then the spatial prevalence projected on the plane of the sky would depend on the inclination angle of the majority of outflows. Since Serpens South is known to have a concentration of YSOs along a dense filamentary structure (Gutermuth et al. 2008), then this structure potentially affects the spatial distribution of outflow emission.

Two previously mentioned ratios, r_L and ${{\eta }_{{\rm out}}}$, quantify the relative contributions of energies from outflows and turbulence, and from outflows and gravity (respectively) in these regions. For the slightly more evolved region NGC 1333, these parameters are ${{r}_{L}}$ ∼ 6–24 and ${{\eta }_{{\rm out}}}$ = 0.2–0.7 (for ξ = 0–573), whereas in Serpens South we found these values to be ${{r}_{L}}$ ∼ 0.6–3 and ${{\eta }_{{\rm out}}}$ = 0.1–0.3 (for ξ = 0–573), respectively, which are about 10% – 50% the values for NGC 1333. The outflows in both regions contribute sufficiently to drive turbulence, although to a greater extent in NGC 1333, and in NGC 1333 the outflows may nearly counter the gravitational energy of the clump. It follows that as a cluster evolves from the very young stage of Serpens South to the slightly more evolved stage of NGC 1333, the outflows sustain turbulence in a similar proportion, but destruction via outflows is more imminent as a region evolves.

Finally, we explore an evolutionary scenario in which Serpens South may come to resemble NGC 1333. First, we calculated the relative protostellar fraction within Serpens South as it progresses from its current, young age ($\sim {\mkern 1mu} (2\pm 1)\times {{10}^{5}}$ yr, Friesen et al. 2013) to the slightly more evolved state of NGC 1333 over its expected age of $\sim 1-2$ Myr (Lada et al. 1996). We assumed lifetimes of 0.10, 0.34, and 0.35 Myr spent as Class 0, Class I, and Flat SED classes, respectively (Evans et al. 2009), and a uniform distribution of ages within each of these classes. In the simplest (unlikely) case, if no additional protostars form, then the protostellar fraction will reach that of NGC 1333 ($\sim 50\%$) in 0.4 Myr. In order to account for the ongoing formation of additional, young protostars, then incorporating also the SFR of 90 M_⊙ Myr⁻¹ (Gutermuth et al. 2008), and an average YSO mass of 0.5 M_⊙ (c.f. Evans et al. 2009), then we add one more protostar every 0.006 Myr, and in that case the protostellar fraction reaches 50% in 1.3 Myr. However, the SFR in the more evolved region NGC 1333 is about half of that in Serpens South, suggesting that the SFR likely decreases as a cluster evolves. If instead we assume the SFR over the entire time to be 45 M_⊙ Myr⁻¹ (as for NGC 1333, Lada et al. 1996), then the protostellar fraction reaches 50% in 0.9 Myr. Additional work should be done to determine how the SFR changes over time for a cluster, but even based on these simple calculations using representative values for the SFR, the protostellar fraction of Serpens South will likely decrease to be comparable to that of NGC 1333 in ∼1 Myr.

Also, as Serpens South evolves to the age of NGC 1333 during the next ∼1 Myr, we expect mass to escape from the cloud, possibly also related to the respective power-law slopes of the mass spectra in Figure 10. We calculated the escape velocity of Serpens South to be ${{v}_{{\rm esc}}}=3.4$ km s⁻¹, which is higher than that of NGC 1333, where ${{v}_{{\rm esc}}}=2.2$ km s⁻¹(Arce et al. 2010), but nonetheless most of the outflow mass we calculate in both regions pertains to gas with outflow velocities greater than the escape velocity. Assuming ages of Serpens South and NGC 1333 as above, and assuming a constant outflow rate, then the current mass-loss rate of 70–120 M_⊙ Myr⁻¹ implies that ∼100 M_⊙ may be lost before Serpens South reaches the evolutionary stage of NGC 1333. At that time, with a cloud mass that is about half of what we assumed here for the calculations of gravitational energy of the clump, the value of ${{\eta }_{{\rm out}}}$ in Serpens South will approximate that of NGC 1333. If cumulative outflow energy decreases with time, then Serpens South would come to resemble NGC 1333 sooner. Although this is a simple order-of-magnitude calculation with admittedly simple assumptions, we consider it valuable to place Serpens South and NGC 1333 within a relative evolutionary scenario. More thorough treatment of the assumptions we made here, including accretion onto the filament (Kirk et al. 2013) and varying (and overall decreasing) mass ejection rate over time, as well as further detailed analysis of additional clusters spanning a range of evolutionary stages, will constrain this scenario.