SMA OBSERVATIONS OF CLASS 0 PROTOSTARS: A HIGH ANGULAR RESOLUTION SURVEY OF PROTOSTELLAR BINARY SYSTEMS

The sources in this study were collected from the SMA science archive. A total of 33 sources were selected by using the following three criteria: (1) available in the SMA archive,¹¹ (2) confirmed classification as a Class 0 object by other studies, and (3) known distance less than 500 pc. These sources were observed at the SMA between 2004 and 2009 with different science goals (e.g., for studying outflow, kinematics, or chemistry in the Class 0 protostars). Only for a few sources (i.e., IRAS 03282, NGC 1333 SVS 13, and CG 30), the proposed science goal was to study the formation of binary stars. The target list and their basic properties (i.e., bolometric luminosity L_bol, bolometric temperature T_bol, and envelope mass M_env) are summarized in Table 1. Most sources in Table 1 were cataloged by both André et al. (2000) and Froebrich (2005). We refer readers to these two papers and references therein for more details about the basic properties of these sources. Note that several sources may represent transition objects between the Class 0 and Class I phases (see Froebrich 2005 for more detailed discussion).

Distances to the different sources were obtained from the literature. Early estimates of the distance to Perseus indicated that this region is 350 pc away (e.g., Herbig & Jones 1983). However, more recent studies indicate that this region is significantly closer. Photometric distances toward the western part of Perseus place this cloud at 220 pc (Černis 1990; Černis & Straižys 2003), consistent with recent very long baseline interferometry (VLBI) observations that give a distance to NGC 1333 of 240 ± 20 pc (Hirota et al. 2008, 2011). We therefore adopt a distance of 240 pc for all objects in Perseus, in agreement with other recent studies of this cloud (e.g., Enoch et al. 2006; Evans et al. 2009; Arce et al. 2010, 2011). The basic properties (bolometric luminosity and envelope mass) of the Perseus sources listed in André et al. (2000) and Froebrich (2005) were modified using the new distance of 240 pc (see Table 1). For sources in the Taurus star-forming region, we adopt a distance of 140 pc (Elias 1978; Kenyon et al. 1994; Loinard et al. 2007b). For sources in the Orion region, we adopt a distance of 440 pc, obtained by the VLBI observations of Hirota et al. (2007). For sources in the Ophuichus cloud, we use a distance of 120 pc, derived from recent Very Long Baseline Array observations by Loinard et al. (2008). For the Lupus region, the distance is estimated to be 130 pc (Murphy et al. 1986). For isolated objects, the distance references are as follows: CB 17 (Launhardt et al. 2010), CG 30 (Knude et al. 1999), L483 (Dame & Thaddeus 1985), L723 (Goldsmith et al. 1984), B335 (Tomita et al. 1979), L1157 (Straižys et al. 1992), and L1251B (Kun & Prusti 1993).

Observations were carried out with the SMA¹² (Ho et al. 2004) between 2004 and 2009. Five different array configurations, Subcompact (Subcom), Compact (Com), Compact-North (Com-N; for several equatorial and southern sources), Extended (Ext), and Very Extended (V-Ext), were used in the observations, with baselines ranging from about 10 m to 500 m. Only one source, IRAS 16293–2422, was also observed with the Extended Submillimeter Array (eSMA¹³).

The 230 GHz and 345 GHz receivers were used in the observations. Both receivers have two sidebands (lower and upper), with 2 GHz bandwidth each (separated by 10 GHz). The receiver setups were typically tuned to the CO (2–1) line in the 230 GHz observations (see, e.g., Chen et al. 2008a; 2010) and to the CO (3–2) line in the 345 GHz observations (see, e.g., Jørgensen et al. 2007; Lee et al. 2009). The total continuum bandwidth, combining the line-free portions of the two sidebands, are about 3.0–3.8 GHz for the 1.3 mm and 850 μm dust continuum. We present only the continuum results in this paper. At 230 GHz, system temperatures generally range from 80 to 300 K (depending on elevation), with a typical value of 150 K. At 350 GHz, system temperatures range from 150 to 500 K, with a typical value of 250 K. The SMA primary beams are ∼55'' at 230 GHz and ∼36'' at 345 GHz.

The visibility data were reduced using the MIR package (Qi 2005). During the reduction, visibility points with clearly deviating phases and/or amplitudes were flagged. The passband (spectral response) was calibrated through observations of available planets (e.g., Saturn) and strong quasars (e.g., 3C 454.3) at the beginning and/or end of each track. Time variation of amplitude and phase was calibrated through frequent observations of quasars near to each source. The flux density scale was calibrated by observing planets such as Uranus and Ceres. For a few tracks, strong quasars (e.g., 3C 273) with stable flux densities during the observing dates were also used as flux calibrators. We estimate a typical flux accuracy of about 20%–30%. The calibrated visibility data were then imaged using the MIRIAD toolbox (Sault et al. 1995). Each continuum map was deconvolved down to the 1σ theoretical rms noise level using the MIRIAD clean routine. The synthesized beam sizes (with robust weighting 1.0) and rms noise levels are summarized in Tables 2 and 3. Figure 1 shows the beam sizes plotted as a function of the corresponding linear resolutions. The median angular resolution in this survey is 31 at λ1.3 mm and 10 at λ850 μm, corresponding to the median linear resolutions of approximately 640 AU and 220 AU, respectively. Combining the observations at the two wavelengths, the median angular and linear resolutions in this survey are 25 and 600 AU, respectively. Figure 2 shows the distribution of the rms noise levels. The median 1σ theoretical noise is 1.0 mJy beam⁻¹ at λ1.3 mm and 1.7 mJy beam⁻¹ at λ850 μm, while in the cleaned images, the median 1σ noise is 2.3 mJy beam⁻¹ and 5.5 mJy beam⁻¹, respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Before analyzing the binary/multiple frequency in the Class 0 protostellar phase, it is important to note how binary/multiple systems are defined in this work. Strictly speaking, only systems that show bound motions can be referred to as binary/multiple systems. However, compared with well-studied MS/PMS binaries (for which the orbital motion can be determined by direct imaging or spectroscopic monitoring at optical and near-infrared wavelengths; see Duchêne et al. 2007), these kinds of systems are extremely rare in the protostellar phase (see, e.g., Rodríguez 2004). The main reasons are: (1) the sample of protobinary systems is relatively small, (2) it is difficult to directly observe the protostellar phase, due to the limitation of angular resolution obtained at millimeter wavelengths, and (3) it is difficult to monitor bound motion in protobinary systems, as most known protobinary systems have large separations (100–1000 AU) and thus have long orbital periods. In this work, we use the following criterion to define binary/multiple systems: the separation or the closest separation (if there are more than two components in the systems) is less than 5000 AU. We explain below why we choose 5000 AU as the limit.

As introduced in Section 1, through initial fragmentation, large-scale (∼0.1–1.0 pc) molecular clumps (or filaments) fragment into individual dense cores, which subsequently collapse to form individual stellar systems (i.e., fragmented components from initial clump fragmentation are not bound stellar systems but prestellar cores). The fragmented components from initial fragmentation generally have separations of more than 5000 AU, consistent with the typical Jeans length in large-scale molecular clumps/filaments (4000–5000 AU at T = 10 K and $n_{\rm H_2}$ = 10⁵–10⁶ cm⁻³). This is also in agreement with the results from the high angular resolution survey by Looney et al. (2000) that systems with separations ⩽6000 AU have common envelopes while wider systems have separated envelopes. Interestingly, by analyzing observational data in the Taurus region (e.g., Ghez et al. 1993; Gomez et al. 1993), Larson (1995) also found the existence of an "intrinsic" scale (∼0.04 pc or 8000 AU) in binary star formation process, and this scale is found to be consistent with the Jeans length in large molecular clouds. Therefore, we consider that components with separations >5000 AU are likely not relevant for the formation of bound stellar systems. This is also the reason why we list several well-known sources, e.g., NGC 1333 IRAS 2A and 2B (separation ∼7500 AU) and IRAS 4A and 4B (separation ∼7200 AU), as separate targets in this work (see Table 1).

We use the densities of YSOs and dense cores in the Perseus molecular cloud to estimate the likelihood of identifying a background or foreground source as a possible binary (or multiple) companion in this study. Many of the targets (14 of 33, or 42%) in this sample are located in the Perseus cloud, and most of the rest of the sources are in regions with lower protostellar densities. In general, MS and evolved stars in the foreground and background of the cloud have very small amounts of (or no) circumstellar dust, and should be undetectable by this interferometric submillimeter and millimeter continuum survey. Therefore, the only possible "contaminating" sources in this study should be young stars with dusty circumstellar envelopes associated with the cloud. In Perseus, there are a total of 400 YSOs in an area of 3.86 deg² (see Jørgensen et al. 2006). Assuming that these YSOs all have detectable dust continuum emission at the SMA, the density of potential dust continuum sources is 103.6 YSOs deg⁻². We therefore estimate the probability of there being a contaminating YSO within a 5000 AU (21'' at the distance of Perseus) radius that could mistakenly be identified as a companion to be 0.01. Another estimate of the contamination likelihood can be obtained using the density of dense cores detected in previous single-dish dust continuum surveys. Hatchell et al. (2005; SCUBA 850 μm) and Enoch et al. (2006; Bolocam 1.1 mm) found 91 (within an area of 3 deg²) and 122 (within an area of 7.5 deg²) dense cores in Perseus, respectively. From these two large-scale surveys, the probabilities are estimated to be 3.2 × 10⁻³ and 1.7 × 10⁻³, respectively. Hence, it is very unlikely that any of the identified companions in this work are background or foreground sources.

Nevertheless, we want to note that further high angular resolution kinematical data are important for complementing the separation criterion used here. For example, for relatively wide protobinary systems (>1000 AU), molecular line observations are needed to verify whether the components therein have similar systematic velocities, in order to insure that the components resolved therein are physically associated.¹⁶ For close systems (on the scale of 100 AU), high angular resolution line (e.g., CO or H₂) observations are needed to distinguish whether each component drives its own outflow, in order to be sure that each component is associated with a protostellar object (and not a dense, starless, condensation).

In this work, we adopt the following terminology and definitions to describe the complexities of binary/multiple stars:

$$\begin{equation} {\rm multiplicity\ frequency\ (MF)} = \frac{B+T+Q}{S+B+T+Q}, \end{equation} \tag{ 4 }$$

$$\begin{equation} {\rm companion\ star\ fraction\ (CSF)} = \frac{B+2T+3Q}{S+B+T+Q}, \end{equation} \tag{ 5 }$$

where S, B, T, and Q are the numbers of single, binary, triple, and quadruple systems, respectively. The "multiplicity frequency," suggested by Reipurth & Zinnecker (1993), represents the probability that any system is a binary/multiple system. The "companion star fraction," on the other hand, looks at the individual stars and counts the number of companions relative to the sample size. It should be noted that many previous works used "companion star fraction" rather than "multiplicity frequency" in their analysis (see, e.g., Haisch et al. 2004; Duchêne et al. 2004; Connelley et al. 2008b). Actually, as noted by Reipurth & Zinnecker (1993), the most precise description of multiplicity consists of explicitly listing the number of singles, binaries, triples, and quadruples, like S:B:T:Q, although this description (as well as others) does not account for hierarchicality.

In this SMA survey toward 33 protostars, 22 sources show signatures of binarity or multiplicity (see the Appendix for detailed descriptions of the individual sources). Table 6 lists the names of these sources, as well as their basic observed parameters (separation and circumstellar mass ratio). If we adopt 5000 AU as the limit, source CB 17 (214 or 5350 AU) is beyond this limit and will not be regarded as a "bound" binary system in this work. Therefore, the numbers of singles,¹⁷ binaries,¹⁸ triples,¹⁹ and quadruples²⁰ are considered to be 12, 14, 5, and 2, respectively, which is S:B:T:Q = 0.86:1.00:0.36:0.14 when normalized to the number of binaries. The overall MF and CSF derived in the sample, without correcting for completeness (see below), is 0.64 ± 0.08 and 0.91 ± 0.05, respectively, with the separations in the range between 50 AU (which is the smallest separation we detect between two components) and 5000 AU (the limit set in this work). The statistical standard errors are estimated as [MF(1−MF)/N]^1/2 and [CSF(1−CSF)/N]^1/2 (where N is the total number of sources in the sample), respectively.

It is important to note that the angular resolution is not uniform in this SMA survey (see Tables 2 and 3). The median angular resolution in the survey is 25, corresponding to a median linear resolution of 600 AU in nearby clouds (see Figure 1). There are very likely more binary/multiple systems with separations less than 600 AU unresolved in this survey. Therefore, this survey is incomplete across the observed separation range. However, it is difficult to correct for this incompleteness at present because we would have to adopt assumptions regarding the separation distributions of Class 0 protobinary systems, which are unfortunately unknown yet (see further discussion in Section 4.2). Consequently, the MF and CSF derived from this work could be considered as lower limits over the separation range observed here.

Recently, Maury et al. (2010) argued that binary systems are relatively rare in the Class 0 phase, as only one object shows a possible binary detection in their small PdBI survey toward five Class 0 objects. However, two protobinary systems detected in the SMA survey presented here, L1448C (see Appendix A.1) and IRAM 04191 (see Appendix A.8), are missing in their PdBI A-configuration (the most extended configuration) observations due to the low flux-ratio sensitivity (only 0.6–0.9; see Maury et al. 2010). Therefore, taking both L1448C and IRAM 04191 into account, the re-calculated MF in their sample is actually 0.60, which is consistent with the MF derived in this SMA survey (0.64).

Figure 19 shows the distribution of separations for the Class 0 binary/multiple systems observed in this work (see Table 6). It must be noted that these separations, measured in the SMA (sub-)millimeter images, are the projected separations between the components in individual systems. However, as discussed by Reipurth & Zinnecker (1993), using reasonable statistical assumptions, one could convert the mean projected separation (l) to the mean semimajor axis (a) with the ratio of l/a ∼ 0.79–0.85 (see Reipurth & Zinnecker 1993, and references therein). We therefore consider that there is no significant error by assuming that the overall statistical distribution of observed projected binary separations is representative of the actual distribution of the semimajor axes. In this work, we do not use these ratios to convert observed projected separations to the "semimajor" axes, and instead show the original results from the observations.

Zoom In Zoom Out Reset image size

As shown in the top of Figure 19, the distribution of separations for Class 0 protostars shows a general trend in which the CSF increases from wide separations to close separations. Quantitatively, the CSF is 0.42 ± 0.09 for separations from 50 AU to 1700 AU, 0.27 ± 0.08 from 1700 AU to 3350 AU, and 0.21 ± 0.07 from 3350 AU to 5000 AU. The median separation in this survey is 1800 AU. Previous large-scale surveys toward MS and PMS stars found that the distributions of separations could be well fitted with log-normal functions with peaks at less than 100 AU (see, e.g., Duquennoy & Mayor 1991 and Patience et al. 2002). As shown in the bottom of Figure 19, the distribution of separations for Class 0 protostars does not exhibit a log-normal-like distribution, and rather shows a trend where most companions are found with separations around 1800 AU. We ascribe this distribution to the limited angular resolution of the SMA observations (i.e., the completeness limit of the survey, see below), which allowed us to mostly detect binary separations on the order of 1000 AU. Further higher angular resolution large-scale surveys toward Class 0 protostars are needed, in order to achieve a wider distribution of separations (from ∼0.1 AU to ∼10³ AU) for protobinary systems.

To better understand the evolution of binary properties, it is important to compare the separation distribution of Class 0 protobinary systems with those derived for Class I, PMS, and MS binaries. Yet, it must be noted in advance that MS and PMS binaries, as well as Class I binaries, have been well studied over the past two decades thanks to the availability of both large samples and high angular resolution obtained with large optical/near-infrared telescopes. The sample size and angular resolution in this SMA survey are an order of magnitude less than the optical and near-infrared surveys. Therefore, large uncertainties may remain in our comparisons due to the small sample size and limited angular resolution in this work.

In a large-scale survey toward Class I YSOs, Connelley et al. (2008a, 2008b) found the number of single, binary, triple, and quadruple stars is S:B:T:Q = 122:51:12:4, with separations ranging from about 50 AU to 5000 AU. Based on the large-scale survey of Duquennoy & Mayor (1991), with a similar separations range, the corresponding number found in solar-type MS stars is S:B:T:Q = 131:32:1:0 (see Connelley et al. 2008b). These two studies, which summarize the binary properties for Class I YSOs and MS stars, will be mainly used here for comparison, as their separation ranges basically match the observed separation range in this SMA survey. From these two large-scale surveys, the MF and CSF are derived to be 0.35 ± 0.03 and 0.46 ± 0.04 for Class I YSOs, and 0.20 ± 0.03 and 0.21 ± 0.03 for MS stars, respectively. Despite the incompleteness in the SMA survey and the difference between the sample sizes, the MF (0.64 ± 0.08) and CSF (0.91 ± 0.05) found in Class 0 protostars are 1.8 (for MF) and 2.0 (for CSF) times larger than the values found in Class I YSOs, and 3.2 (for MF) and 4.3 (for CSF) times larger than the values found in MS stars (see Figure 20). Furthermore, the observed fraction of high-order multiple systems to binary systems in Class 0 protostars (7:14 or 0.50 ± 0.09) is also larger than the fractions found in Class I YSOs (0.31 ± 0.07) and MS stars (0.03 ± 0.01).²¹ To estimate how different the distribution of separations for Class 0 protostars is from those distributions for Class I YSOs and MS stars, we use the equation given by Brandeker et al. (2006) in their Appendix B2 to calculate the probability that the results in a given separation range from two studies are consistent with each other. Using this method, the probabilities that Class 0 protostars have the same distribution of separations as Class I YSOs and MS stars are estimated to be 1.9 × 10⁻⁶ and 4.1 × 10⁻¹⁵, respectively.

Zoom In Zoom Out Reset image size

The distributions of separations derived for PMS binaries are in general similar to those for Class I binaries (see, e.g., Connelley et al. 2008b). Therefore, we consider that the comparisons between Class 0 and PMS binaries will likely hold similar results to those derived from the comparisons between Class 0 and Class I binaries. Indeed, for example, in the survey toward PMS stars, Ghez et al. (1997) found a CSF of 0.21 ± 0.04 with separations from 150 AU to 1800 AU; with the same separation range, the CSF found by Reipurth & Zinnecker (1993) is 0.16 ± 0.02. The CSF derived from this survey with the same separation range is 0.39 ± 0.08, approximately two times higher, comparable to the ratios estimated between the Class 0 and Class I binaries.

As we indicated in Section 4.1, this SMA survey is incomplete across the observed separation range, and the derived MF and CSF in this work should be considered as lower limits. Therefore, the ratios estimated above should also be considered as lower limits. Considering that the median separation in this SMA survey is 1800 AU (larger than the lowest linear resolution in the survey of 1500 AU), the survey can be regarded to be complete for protobinary systems with separations from 1800 AU to 5000 AU. Since, to our knowledge, this work is the first large-scale observational survey of multiplicity in Class 0 protostars, the distribution of protobinary systems with smaller separations is not yet known. Hence, it is difficult to correct for our incompleteness to derive an accurate MF or CSF for Class 0 protobinary systems with separations between 50 AU and 1800 AU. Nevertheless, we may adopt a simple method to "correct" for completeness in this work. For separations smaller than 1800 AU, we assume that the completeness in a separation range is given by the averaged fraction of the sources in the total sample that have linear resolutions intermediate between the higher and lower limits of the range. For example, 55% of the sample has a resolution better than the median resolution of 600 AU, then we assume that we only detect 55% + 100}/2 of the total number of multiple systems in the range between 600 AU and 1800 AU. Using this method, the MF and CSF are derived to be 0.33 ± 0.08 and 0.42 ± 0.08 for systems with separations from 1800 AU to 5000 AU (100% completeness), 0.15 ± 0.06 and 0.24 ± 0.07 for systems with separations from 600 AU to 1800 AU (78% completeness), while 0.15 ± 0.06 and 0.24 ± 0.07 for systems with separations from 50 AU to 600 AU (32% completeness). After correcting for completeness, the overall MF and CSF in this SMA survey are 1.0 ± 0.1 and 1.5 ± 0.2, respectively, over the separation range from 50 AU to 5000 AU. Compared with the data derived from Class I YSO survey (Connelley et al. 2008a, 2008b), these "corrected" values result in a difference, in both the MF and the CSF, of a factor of three between Class 0 protostars and Class I YSOs, which is indeed larger than the factor of two estimated above. However, it must be noted that these "corrected" values are uncertain. Further high angular resolution large-scale surveys toward Class 0 protostars are needed, in order to draw a real statistically significant separation distribution for close protobinary systems.

The decrease in both the MF and CSF, as well as the fraction of high-order multiple systems, suggests that binary properties change as stars evolve from the Class 0 to Class I/PMS, all the way to the MS phases. As reviewed by Goodwin et al. (2007), the evolution of binary properties has been ascribed to two main mechanisms: (1) the rapid dynamical decay of young small-N systems (2 < N < 12; see, e.g., Reipurth 2000; Reipurth & Clarke 2001; Goodwin & Kroupa 2005); and (2) dynamical destruction of loose binary systems in dense clusters (see, e.g., Kroupa 1995). Here, we consider that small-N decay is likely the main process for the evolution of binary properties because (1) most sources in this sample, as well as those in the Class I, PMS, and MS surveys discussed above, are not in dense cluster environments, and (2) the timescale for the decrease in binary fraction caused by destruction in dense cluster (a few 10 Myr) is much greater than the lifetimes of Class 0 protostars (∼10⁴–10⁵ yr) and Class I YSOs (∼10⁵ yr). As estimated by Goodwin et al. (2007), a triple system is unstable to decay with a general half-life of a few 10⁴ yr, which is consistent with the lifetime estimated for Class 0 protostars. Thus, most of the decays/ejections are expected to happen during the Class 0 phase (e.g., Reipurth 2000). This small-N decay will modify the binary properties by reducing the overall MF and CSF, and reducing the fraction of high-order multiple systems to binary systems. All have been evidenced by the comparisons between this work and previous large-scale surveys toward Class I and MS/PMS binaries.

In short, we find that approximately two-thirds of Class 0 protostars are binary or multiple systems in this SMA survey, and this could only be a lower limit to the fraction of Class 0 protostars formed in such systems. The results derived from this work, in concert with the comparison between this survey and large surveys toward Class I and MS/PMS binaries, is consistent with the possibility that most, if not all, stars are formed in binary or multiple systems and that single stars result from the decay of multiple systems, as suggested by Larson (1972).

We have derived the circumstellar gas masses for the sources in the sample using the (sub)millimeter dust continuum emission (see Section 3 and Tables 4 and 5). We note that these values are uncertain and generally cannot be compared among the sources in the sample because different sources were observed differently (i.e., with different angular resolution, uv-coverage, and different resulting sensitivity), which probe different size scales and amounts of material. Relative masses between components within individual systems are more reliable. Assuming, for simplicity, that sources in a multiple system have similar temperatures, then we can obtain the circumstellar mass ratio from the continuum emission flux ratio of the sources (see Equations (1)–(3)). Table 6 lists the circumstellar mass ratios of the binary/multiple systems observed in this work, under the assumption that the circumstellar mass is proportional to the (sub)millimeter continuum flux. It must be noted that this circumstellar mass ratio does not necessarily reflect the mass ratio of the hydrostatic cores nor that of the final stars, but still provides a valuable comparison between the components in these systems. Figure 21 shows the distribution of circumstellar mass ratios for the protobinary systems in this work. The distribution appears to be flat like that of more evolved long-period PMS and MS binary stars (see, e.g., Reipurth & Zinnecker 1993; Halbwachs et al. 2003). We find that 67% ± 8% of protobinary systems have circumstellar mass ratios below 0.5, i.e., unequal-mass systems are more common than equal-mass systems in the sample. This result suggests that unequal-mass systems may be preferred in the process of (wide) binary star formation (separations on the order of 1000 AU).

Zoom In Zoom Out Reset image size

Figure 22 shows the distribution of circumstellar mass ratios versus separations. The distribution is somewhat scattered, similar to the distribution of PMS binaries (see, e.g., Reipurth & Zinnecker 1993). Numerical simulations predict that close binary systems (separations on the order of 1 AU) are likely to have mass ratios near unity (see Bate 2000), which is indirectly supported by statistical studies of MS binary systems (see, e.g., Halbwachs et al. 2003). However, we do not find this trend in Figure 22. It is probably because (1) the sample size in this work is still small, and (2) we cannot resolve close binary systems in which components may prefer similar masses.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Different physical processes at different evolutionary stages may trigger different fragmentation processes (e.g., Tohline 2002; Machida et al. 2008). At different evolutionary stages, the varying properties of a collapsing core (e.g., core temperature and density) will result in different fragmentation scales. As introduced in Section 1, depending on the evolution of a collapsing core, fragmentation models can be divided into initial fragmentation, prompt (isothermal) fragmentation, adiabatic fragmentation, and secondary fragmentation. These models suggested that the fragmentation of a core could take place sequentially during the core's different collapsing phases, which would result in a hierarchical system eventually. However, it is yet unclear whether the fragmentation at one specific phase is dominant over other phases, and whether most collapsing cores go though all these fragmentation processes. Based on the extensive single-dish and interferometric observations, we show in Figure 23 a suggested sequential fragmentation picture. This is a rough outline and further efforts are needed to collect corresponding samples and to derive their kinematical properties, in order to understand the fragmentation mechanisms at different core's collapse phases in detail.

The initial fragmentation of a large-scale clump or filament (∼0.1–1.0 pc) occurs before the collapse of individual prestellar cores, and results in cores with separations of about 10³–10⁴ AU. This large-scale fragmentation has been frequently observed with single-dish millimeter telescopes (see, e.g., Kauffmann et al. 2008; Launhardt et al. 2010), and it is likely controlled by the combined effects of gravity and turbulence. Figure 23 shows a possible example of the initial clump fragmentation, CB 246 (see Launhardt et al. 2010 for more details), where the extended large-scale clump has fragmented into separated cores. These cores are invisible in the Spitzer Space Telescope (Spitzer) infrared images and may represent the prestellar cores, which will further collapse and fragment.

Most of the protobinary systems that have been observed with current millimeter arrays have separations in scales of 1000 AU (e.g., Looney et al. 2000, and this work). These systems appear to support the scenario of prompt fragmentation, which occurs at the end of the isothermal collapse phase (see Tohline 2002). Nevertheless, these observed systems are mostly made of Class 0 objects that have already gone through the fragmentation process, and thus these systems do not exhibit the initial conditions of a fragmenting core. An observational example representing the prompt fragmentation may be the prestellar core R CrA SMM 1A (see Figure 23), in which multiple faint condensations were discovered with separations between 1000 AU and 2000 AU. These condensations are invisible in the deep infrared images and have extremely low bolometric luminosities (<0.1 L_☉) and temperatures (<20 K), indicating that these are young sources have yet to form protostars, and therefore represent the earliest phase of core fragmentation observed (see Chen & Arce 2010 for more details). More recently, based on the kinematic data collected at different interferometric arrays, Chen et al. (2012a) found that most protostellar cores with binary systems (separations scale ∼1000 AU) formed therein have ratios of rotational energy to gravitational energy β_rot > 1%, which is consistent with theoretical simulations (see, e.g., Boss 1999). This suggests that the level of rotational energy in a dense core plays an important role in the prompt fragmentation process.

Unfortunately, there is a lack of observations of protobinary systems with separations between 10 AU and 100 AU, which could be used to study the adiabatic fragmentation scenario (see Machida et al. 2008). This can be easily explained by the fact that current millimeter interferometers (e.g., SMA and PdBI) normally reach 1''–2'' angular resolution under general conditions, and thus mostly resolve protobinary systems with separations of 100–200 AU or larger in nearby star-forming clouds (e.g., Perseus and Ophuichus). In Figure 23, we show the eSMA 850 μm image of source IRAS 16293A, which represents the best angular resolution that can be achieved at the SMA. In IRAS 16293 A, a close binary system with a separation of 50 AU is revealed (see also Appendix A.19). We speculate that the close binary system in IRAS 16293A was formed through adiabatic fragmentation, which took place after the initial clump fragmentation (that formed the IRAS 16293 and IRAS 16293 E cores; see, e.g., Figure 8 in Jørgensen et al. 2008) and the prompt (isothermal) fragmentation (that formed the IRAS 16293 A-B system; see Figure 15(d)). We note that the protobinary system VLA 1623 (see Appendix A.17 and Figures 15(a) and (b)) may show another example of adiabatic fragmentation. Here, the VLA 1623 system appears to have fragmented from the Oph-A core through initial clump fragmentation and have later fragmented into VLA 1623 West and East (through prompt fragmentation); VLA1623 East seems to have subsequently gone through adiabatic fragmentation to form VLA 1623 East A and B. As we discussed in Appendix A.17, source VLA 1623 B could also be a binary system, which needs to be verified by further higher angular resolution observations.

We believe that routine observations at angular resolution better than 01 will reveal more multiple protobinary systems with smaller separations, which will then allow us to study in detail a statistically significant number of protostellar binary/multiple systems with a wide range of separations (from 1 AU to 10⁴ AU). The Atacama Large Millimeter/submillimeter Array (ALMA), capable of attaining an angular resolution down to 10 mas (when fully completed), will provide a breakthrough in our knowledge of binary star formation. It will certainly help us find closer protobinary systems with separations in the scale of 1–10 AU, and will even allow us to test the secondary fragmentation scenario (see, e.g., Machida et al. 2008), by which stellar systems with separations of 0.01–0.1 AU can be formed.

L1448 is a dark globule located on the west end of the Perseus cloud, which has been observed extensively in the past two decades (see, e.g., Bally et al. 2008, and references therein). Figure 3(a) shows the SCUBA 850 μm image of the L1448 region. The region contains three well-known Class 0 protostars: L1448 IRS2, L1448 N, and L1448C (see, e.g., Barsony et al. 1998; O'Linger et al. 2006).

L1448 IRS2 was suggested to be a protobinary candidate (with a 10'' separation) by Wolf-Chase et al. (2000) based on the NRAO 12 m CO (1–0) observations that mapped two distinct molecular outflows from IRS2. Volgenau et al. (2002) also claim the detection of a binary system in IRS2 using BIMA, but no results (e.g., spatial separation or image) have been presented yet. However, the SMA 1.3 mm dust continuum observations with an angular resolution of 34 × 25 (see Figure 3(c)), as well as Spitzer infrared observations (e.g., Chen et al. 2010), do not find any companions around IRS2 (see also O'Linger et al. 2006). We therefore conclude that L1448 IRS2 is a single protostar.

L1448 N is also referred to as L1448 IRS3 by several authors. However, we prefer using the name L1448 N because the IRAS source IRS3 in fact consists of two protostars, L1448 N and L1448C (see O'Linger et al. 2006). L1448 N is the brightest source at infrared and millimeter wavelengths in L1448, and has been resolved into three distinct sources by previous interferometric observations (e.g., Looney et al. 2000). The three sources are named L1448 NA, L1448 NB, and L1448 NW (O'Linger et al. 2006). The SMA 1.3 mm dust continuum image of L1448 N shows these three distinct sources (see Figure 3(b)). The dust continuum emission is dominated by source NB, which is located 74 ± 03 south of source NA. Source L1448 NW, located approximately 20'' northwest of the NA–NB pair, has relatively weaker dust continuum emission in the interferometric observations.

L1448C, also known as L1448-mm, is one of the best-studied Class 0 protostars. L1448C is well known for driving a young, highly collimated, and high-velocity bipolar outflow (e.g., Bachiller et al. 1990; Hirano et al. 2010). Spitzer infrared observations resolved L1448C into two separated components, with an angular separation of about 8'' (see Jørgensen et al. 2006). The SMA observations at both 850 μm and 1.3 mm continuum bands detect these two components (see Figures 3(d) and (e)). The dust continuum emission is dominated by the northern component, and only weak dust continuum emission is found from the southern component (5σ detection). The northern component is the driving source of the well-known high-velocity bipolar outflow (Jørgensen et al. 2007; Hirano et al. 2010). Maury et al. (2010) argued that the southern component is not a protostar but a dusty clump associated with the outflow driven by the northern component. However, we note that a relatively low velocity bipolar outflow associated with the southern component has been found in the SMA CO (2–1) observations (see Jørgensen et al. 2007) and CO (3–2) observations (see Hirano et al. 2010), which clearly indicates that the southern component is a protostellar object.

The protostellar core PER 065 was discovered by the dust continuum survey in the Perseus cloud, and was classified as a Class 0 protostar by Hatchell et al. (2007) and Enoch et al. (2009). Because of its low luminosity (0.03–0.22 L_☉), it was also cataloged as one of the very low luminosity objects in Perseus (No. 065; see Dunham et al. 2008). The SMA 1.3 mm dust continuum observations resolve, for the first time, the faint core into three components, named A, B, and C, respectively (see Figure 4). The separations between components A and B and between B and C are 39 ± 04 and 75 ± 04, respectively.

NCG 1333 is a well-studied protostellar cluster located in the western part of the Perseus cloud (see, e.g., a review by Walawender et al. 2008 and references therein, and Arce et al. 2010). Submillimeter and millimeter dust continuum observations have revealed several protostellar cores within the region (see, e.g., Sandell & Knee 2001), of which three cores, SVS 13, IRAS 2, and IRAS 4, are the most well known and studied (see Figure 5(a)).

Figure 5(b) shows the SMA 1.3 mm continuum image of SVS 13, in which three distinct sources, referred to as A, B, and C, are found. The SMA results are consistent with the results found in other high angular resolution observations (e.g., Bachiller et al. 1998; Looney et al. 2000; Chen et al. 2009). Another weak continuum source, which is spatially coincident with the radio source VLA 3 (Rodríguez et al. 1997) and the 3 mm dust continuum source detected by Looney et al. (2000) and Chen et al. (2009), is found in the SMA 1.3 mm dust continuum image (see Figure 5(c)). Also note that, with the VLA centimeter observations, Anglada et al. (2000, 2004) resolved source A into a close binary system with an angular separation of 03.

The SMA continuum images of IRAS 2 and IRAS 4 are shown in Figures 5(d) and (e) and Figures 5(f) and (g), respectively. The continuum results of the two sources derived independently in this work are comparable with those in Jørgensen et al. (2007). We therefore refer readers to that work for more details about the SMA results. We note that IRAS 2A is resolved into two separated components in the high angular resolution 850 μm image (see Figure 5(e)), with a faint companion 15 ± 02 to the north of the main dust peak. In the CO observations, a quadrupolar outflow centered at IRAS 2A was detected (see, e.g., Jørgensen et al. 2007). This quadrupolar outflow is likely two different outflows driven by the two binary components in IRAS 2A. Further high angular resolution observations are needed to confirm the tentative detection of the faint secondary source in IRAS 2A and to distinguish which source drives which outflow.

IRAS 03282+3035 (IRAS 03282) is a young Class 0 protostar located in the western part of the Perseus cloud, which has been studied by various groups using different molecular line transitions (see, e.g., Bachiller et al. 1991, 1994; Arce & Sargent 2006; Chen et al. 2007; Tobin et al. 2011).

The OVRO 1.3 mm dust continuum observations (at 09 × 07 resolution) revealed two millimeter sources in IRAS 03282, with an angular separation of 15 (see Launhardt 2004). Although the SMA 1.3 mm observations (at 25 × 22) do not resolve the two components, the SMA 850 μm observations (at 09 × 08) show two separated components (see Figure 6). We refer to the main continuum source as MMS 1 and the faint source as MMS 2. In the SMA images, the angular separation between the two sources is 16 ± 02, similar to the result found in the OVRO 1.3 mm images.

Barnard 1 (B1) is a dark cloud in the western part of Perseus. Single-dish dust continuum observations have revealed four protostellar cores in this cloud (see, e.g., Matthews & Wilson 2002; Hatchell et al. 2005, and Enoch et al. 2006), which are commonly referred to as B1-a, b, c, and d (see Figure 7(a)).

The Class 0 protostar B1-c was studied with BIMA by Matthews et al. (2006). The BIMA 3.3 mm dust continuum observations (at 64 × 49 resolution) revealed a protostellar envelope with radius of ∼2400 AU and mass of 2.1–2.9 M_☉. The SMA 1.3 mm dust continuum observations, with higher angular resolution (34 × 29), resolve an inner envelope with radius of 700 ± 100 AU and mass of 0.24 ± 0.05 M_☉ (see Figure 7(b)).

The B1-b protostellar core is elongated in the north–south direction in the SCUBA images (see Figure 7(a)). The 3 mm dust continuum observations at the Nobeyama Millimeter Array found two separate sources in the B1-b core, referred to as B1-bN and B1-bS (Hirano et al. 1999). The SMA 1.3 mm observations show similar results to those found by Hirano et al. (1999): the two sources are separated by 174 ± 05 and have similar circumstellar masses (mass ratio 0.7 ± 0.2).

IC 348 MMS is a Class 0 protostar discovered in the southwestern part of the IC 348 cluster (see Eislöffel et al. 2003). It drives a collimated molecular outflow, traced by H₂ emission (Eislöffel et al. 2003) and CO emission (Tafalla et al. 2006). The SMA 1.3 mm dust continuum images of IC 348 MMS reveal, for the first time, two sources embedded in the common envelope detected by single-dish dust continuum observations (see Figure 8(b)). The angular separation between the two sources is 98 ± 04, and the mass ratio is 0.25 ± 0.10. The main continuum source MMS 1 is the driving source of the collimated molecular outflow.

The Class 0 protostar HH 211 MMS is well known for driving a highly collimated bipolar outflow, which has a narrow bipolar jet-like component in the high-velocity CO/SiO emission and a relatively wide-angle bipolar cavity in the low-velocity CO emission (see, e.g., Gueth & Guilloteau 1999; Hirano et al. 2006; Lee et al. 2007b).

Based on very high angular resolution SMA observations (at 020 × 015), Lee et al. (2009) found that HH 211 MMS actually contains two dust continuum sources, SMM 1 and SMM 2, with an angular separation of 031 ± 002 (see Figure 8(c)). Source SMM 1 appears to be the driving source of the collimated jet/outflow, while the nature of source SMM 2, which has a mass of only 1.5–4 M_Jup, is still uncertain. More interestingly, based on the reflection-symmetric wiggle of the HH 211 jet, Lee et al. (2010) suggested that source SMM 1 itself could be a very low mass protobinary with a separation of ∼4.6 AU (the putative 4.6 AU companion was not directly detected, but inferred from the morphology of the outflow). Further higher angular resolution observations are needed to confirm this interesting tentative detection. More details about HH 211 MMS and its surrounding envelope can be found in Gueth & Guilloteau (1999), Lee et al. (2007b, 2009, 2010), and Tanner & Arce (2011).

CB 17 (also known as L1389) is a small, isolated, and slightly cometary-shaped dark cloud, which was classified as a Bok globule by Clemens & Barvainis (1988). CB 17 has been studied by various groups using different molecular line transitions (e.g., Pavlyuchenkov et al. 2006) and multi-wavelength dust continuum emission (e.g., Launhardt et al. 2010).

The SMA 1.3 mm dust continuum observations reveal two sources within CB 17, which are separated by 214 ± 05 in the northwest–southeast direction (see Figure 9(a)). The northwestern dust continuum source is associated with an infrared source seen in the Spitzer images and is referred to as CB 17 IRS (see Chen et al. 2012b). No compact infrared emission is detected from the southeastern source in the Spitzer bands from 3.6 to 70 μm, and we refer to this source as CB 17 MMS. Interestingly, the SMA CO (2–1) observations suggest that source CB 17 MMS, an object with an extremely low bolometric luminosity (⩽0.04 L_☉), is driving a low-velocity (∼2.5 km s⁻¹) molecular outflow. These characteristics have lent this source to be classified as a candidate of the so-called first hydrostatic cores (see Chen et al. 2012b for more details).

IRAM 04191+1522 (IRAM 04191) is a well-studied Class 0 protostar located in the southern part of the Taurus molecular cloud (André et al. 1999). IRAM 04191 drives a large bipolar molecular outflow, features bright molecular line and dust continuum emission, and shows evidences of supersonic infall, fast rotation, depletion, and deuteration (see, e.g., André et al. 1999; Belloche et al. 2002; Belloche & André 2004; Lee et al. 2005).

The SMA 1.3 mm dust continuum observations reveal two distinct continuum sources with an angular separation of 80 ± 04, embedded within the elongated core of IRAM 04191 (see Figure 9(b)). The discovery of this binary system was reported by Chen et al. (2012a). The southeastern source, associated with an infrared source seen in the Spitzer images (see Dunham et al. 2006), represents the well-known Class 0 protostar and is referred to as IRAM 04191 IRS here. In contrast, the northwestern source has no infrared emission detected in the Spitzer images, and is referred to as IRAM 04191 MMS (see Chen et al. 2012a for more details).

L1521F is a well-studied dense core in Taurus. This core shows high central density, infall asymmetry, molecular depletion, and enhanced deuterium fractionation, and was suggested as a highly evolved starless core by Crapsi et al. (2004). However, Spitzer observations found that L1521F harbors a very low luminosity protostar, referred to as L1521F-IRS (see Bourke et al. 2006), which is in the early stage of the Class 0 phase (Shinnaga et al. 2009; Terebey et al. 2009). The SMA 1.3 mm dust continuum observations of L1521F-IRS show a faint source (see Figure 9(c)). More than 95% of the flux around L1521F-IRS is resolved out by the SMA observations, compared to the flux detected in the IRAM 30 m 1.2 mm maps (600 ± 150 mJy; Bourke et al. 2006). We also note that L1521F-IRS was observed at higher angular resolution with the IRAM-PdBI by Maury et al. (2010). In the PdBI 1.3 mm dust continuum images, there is also only one object detected.

L1527-IRS is another protostar in Taurus, which is associated with a far-infrared source and is probably more evolved than L1521F-IRS (see, e.g., Terebey et al. 2009). Based on the JCMT 800 μm continuum observations, Fuller et al. (1996) suggested L1527-IRS is a binary system with the secondary at about 20'' from the main protostar. However, no evidence is found of this secondary source in the SMA 1.3 mm dust continuum image (see Figure 9(d)). We therefore conclude that L1527-IRS is a single protostar (see also Maury et al. 2010).

HH 114 MMS is a Class 0 protostar located in the western part of the λ Orionis molecular shell, which drives a highly collimated bipolar outflow (see Arce & Sargent 2006). In the SMA 1.3 mm and 880 μm dust continuum images (see Figure 10), source HH 114 MMS remains as a single object at the angular resolution of 22 × 19, which is consistent with the OVRO 2.7 mm and 3.4 mm continuum observations (Arce & Sargent 2006).

The MMS 6 core is the brightest source in submillimeter and millimeter wavelengths in the Orion molecular cloud-3 (OMC-3) region (Chini et al. 1997; Lis et al. 1998; Johnstone & Bally 1999). The SMA 850 μm continuum images show two sources in the MMS 6 core, referred to as SMM 1 and SMM 2, with an angular separation of 108 ± 02 (see Figure 11(a)). The northern continuum source SMM 1 is spatially coincident with a continuum source detected in early interferometric observations (see Matthews et al. 2005; Takahashi et al. 2009). The southern continuum source SMM 2 is associated with a continuum source detected in the BIMA 1.3 mm observations by Matthews et al. (2005), but it was not detected in the multi-wavelength observations by Takahashi et al. (2009). We note that source MMS 6 was also observed in the SMA 0.9 mm continuum by Takahashi et al. (2009), but the sensitivity in their image was about two times worse than that in the SMA 850 μm image presented here.

FIR 5 is a Class 0 protostar located in the NGC 2024 cloud, the most active star-forming region in the Orion B cloud (see, e.g., a review by Meyer et al. 2008, and references therein). With a high-velocity, collimated molecular outflow extending over 5' south of the core (Richer et al. 1992), FIR 5 is the brightest and also probably most evolved protostellar core among the dense cores in NGC 2024 (Mezger et al. 1992).

In the SMA 850 μm continuum images, the FIR 5 core is resolved into two main components, referred to as SMM 1 and SMM 2, with an angular separation of 41 ± 02 (see Figure 11(b)). The SMA results are consistent with the results derived from the PdBI 3 mm continuum observations, in which the FIR 5 core was also resolved into two main components (see Wiesemeyer et al. 1997). Nevertheless, in the BIMA 1.3 mm deep observations, Lai et al. (2002) also found several weaker dust condensations around the two main components, which are not detected in the SMA images. We note that the SMA results of FIR 5 (including also dust polarization and CO emission results) were independently published by Alves et al. (2011), and similar dust continuum results were derived there.

HH 212 is a remarkable jet in the Orion L1630 cloud, discovered by Zinnecker et al. (1998). The jet is associated with a bipolar molecular outflow seen in CO lines (see, e.g., Lee et al. 2000, 2006a, 2007a) and SiO lines (see, e.g., Codella et al. 2007; Lee et al. 2008). The driving source of the jet/outflow, referred to as HH 212 MMS, is a low-luminosity Class 0 protostar associated with a cold IRAS source (IRAS 05413−0104; Zinnecker et al. 1992).

The SMA 1.3 mm and 850 μm dust continuum images of HH 212 MMS were shown by Lee et al. (2006a, 2007a, 2008). By re-analyzing the high angular resolution 850 μm images in Lee et al. (2008; angular resolution 035 × 032), a faint continuum component is found to the west of source HH 212 MMS (see Figure 12(c)). Hereafter, we refer to the main dust continuum component as MMS 1 and to the faint continuum component as MMS 2. The separation between the two components is measured to be 053 ± 005. We also retrieved the SMA 230 GHz data of HH 212 MMS obtained in the Extended and Very Extended configurations (the 230 GHz data published by Lee et al. 2006a were obtained only in the Compact configuration). The high angular resolution 1.3 mm dust continuum images (angular resolution 08 × 07) show that source HH 212 MMS is extended in the east–west direction and a component is tentatively detected in the 1.3 mm images, which is spatially associated with the new component MMS 2 seen in the 850 μm images (see Figure 12(d)).

We also note that another dust continuum source was tentatively detected in the PdBI 1.4 mm continuum observations, which is about 17 to the southeast of the main dust continuum source MMS 1 (see Codella et al. 2007). In the SMA 850 μm images, a weak dust continuum source (∼7σ detection) is seen near this PdBI 1.4 mm source but 06 closer to the main continuum source MMS 1. We referred to this weak continuum source as MMS 3 (see Figure 12(c)). Altogether, there may be three dust continuum sources found within the elongated envelope of HH 212 MMS. Further high angular observations are needed to confirm this tentative multiple system detection.

The protostellar jet HH 25 is located in the northern part of the Orion B dark cloud. The jet is associated with a compact but collimated molecular outflow (e.g., Gibb & Davis 1998; Gibb et al. 2004). The driving source of the jet/outflow, referred to as HH 25 MMS, was revealed by a VLA 3.6 cm survey in this region (Bontemps et al. 1995), and was classified as a Class 0 protostar with a luminosity of ∼6 L_☉ (Gibb & Davis 1998).

In the single-dish dust continuum maps (IRAM 30 m, Lis et al. 1999; JCMT/SCUBA, Phillips et al. 2001), HH 25 MMS is elongated in the north–south direction, with the VLA source located at the northern tip of the extension. In the SMA 870 μm dust continuum image, three distinct, previously unknown sources are revealed within the large-scale elongated core, approximately aligned in the north–south direction (see Figure 13(a)). The three continuum sources, from north to south, are referred to as SMM 1, SMM 2, and SMM 3. The angular separations between sources SMM 1 and SMM 2 and between sources SMM 2 and SMM 3 are measured to be 127 ± 03 and 109 ± 03, respectively. Source SMM 1 is spatially coincident with the VLA 3.6 cm source, and is the driving source of the HH 25 jet/outflow.

CG 30 (also known as BHR 12; Bourke et al. 1995) is a bright-rimmed cometary globule located in the Gum Nebula region. In the SCUBA submillimeter (Henning et al. 2001) and ATCA 3 mm (Chen et al. 2008b) continuum observations, this globule was resolved into a wide protobinary system with an angular separation of ∼22''.

The SMA 1.3 mm dust continuum image of CG 30 shows two compact sources (see Figure 13(b)). The angular separation between the two sources is measured to be 218 ± 05, consistent with the ATCA result. Following Chen et al. (2008b), we refer to the northern source as CG 30N (which is spatially coincident with IRAS 08076−3556) and to the southern source as CG 30S. The two sources each drive their own bipolar protostellar jets/outflows, which are roughly perpendicular to each other (see Chen et al. 2008a, 2008b). Further discussion of the gas kinematics (from ATCA N₂H⁺ and SMA CO observations) and infrared emission (from Spitzer observations) in CG 30 can be found in Chen et al. (2008a, 2008b).

B228 is a Class 0 protostar located in the Lupus I cloud (Shirley et al. 2000). It drives a compact bipolar molecular outflow that extends in the northeast–southwest direction (see van Kempen et al. 2009). In the SMA dust continuum images, an elongated source is seen at both 1.3 mm and 1.1 mm (see Figure 14). The source is extended in the southwest direction, roughly following the direction of the blueshifted outflow lobe, suggesting that the outflow has a strong impact on the dust envelope, similar to L1157 (see, e.g., Gueth et al. 2003).

VLA 1623 is the prototypical Class 0 protostar (André et al. 1993), which drives a large-scale bipolar outflow (Dent et al. 1995; Yu & Chernin 1997). With VLA centimeter continuum observations, Bontemps & André (1997) found a series of emission clumps, named A, B, and C. At high angular resolution, clump A was further resolved into two components (Bontemps & André 1997; Looney et al. 2000).

In the SMA high angular resolution images (resolution 06 × 03; observed in 2007), three distinct dust continuum sources are detected (see Figure 15(a)). The three sources, aligning roughly from east to west, are referred to as VLA 1623 A, VLA 1623 B, and VLA 1623 West, respectively. Sources A and B, the east pair with an angular separation of 11 ± 01 (see Figure 15(b)), are spatially coincident with the BIMA 2.7 mm sources A and B in Looney et al. (2000) and the two VLA centimeter sources in clump A in Bontemps & André (1997). The mass ratio between sources A and B is close to 1.0 ± 0.3, similar to the result found by Looney et al. (2000). Source VLA 1623 West, located approximately 10'' to the east of the A–B pair, is spatially coincident with clump B in Bontemps & André (1997).

More recently, Maury et al. (2012) presented high angular resolution SMA data of VLA 1623 (observed in 2009) and argued that sources VLA 1623 B and VLA 1623 West are two dusty knots due to shocks along the outflow cavity wall of source VLA 1623 A (although they did not show any evidence that VLA 1623 A is the driving source of the VLA 1623 outflow). However, in the Spitzer mid-infrared images, source VLA 1623 West is associated with a point-like infrared source, strongly suggesting that it is a YSO (X. Chen et al. 2013, in preparation). Furthermore, the SMA CO (2–1) data show a quadrupolar outflow centered at source VLA 1623 B, indicating that this source is not only a protostellar object but also a potential close binary system unresolved in the current SMA dust continuum observations (X. Chen et al. 2013, in preparation).

Oph-MMS 126 is a Class 0 protostar discovered in the SEST 1.2 mm continuum survey toward the ρ Ophiuchi molecular cloud, which is associated with a faint, cold IRAS source (IRAS 16253–2429) and drives a low-velocity bipolar molecular outflow (see Stanke et al. 2006). In the SMA images, a faint dust continuum source is seen (about 5σ detection; see Figure 15(c)), which is associated with an infrared source seen in the Spitzer images (see Barsony et al. 2010). The total flux was measured to be 11 ± 2 mJy in the SMA 1.3 mm dust continuum images. Compared with the total flux obtained in the SEST observations (411 mJy; Stanke et al. 2006), more than 97% of the dust continuum emission flux from the large envelope was resolved out by the SMA.

IRAS 16293–2422 (IRAS 16293) is a well-studied Class 0 protostar located in the ρ Ophiuchi cloud. Millimeter interferometric observations resolved this source into a binary system, referred to as IRAS 16293 A and B, with a separation of 52 (see, e.g., Mundy et al. 1992; Looney et al. 2000). In the high angular resolution VLA radio continuum observations, source A is further resolved into a chain of radio components (see Loinard et al. 2007a; Pech et al. 2010), which are suggested to be related to a jet from source A. Recent ALMA Science verification data revealed an inverse P-Cygni profile toward source B which is direct evidence of infalling material close to the central protostar (Pineda et al. 2012).

Figure 15(d) shows the SMA 1.3 mm continuum image of IRAS 16293, in which two continuum sources, A and B, are detected with an angular separation of 53 ± 02. In the higher angular resolution eSMA 850 μm images, source A is further resolved into two close continuum sources (see Figure 15(e)). The separation between the two sources is 042 ± 005. Following Chandler et al. (2005), the two sources are referred to as Aa and Ab.

L483 is an isolated dense core, associated with a strong infrared source (IRAS 18148−0440). It was classified as a Class 0/I transition object by Tafalla et al. (2000). High angular resolution molecular line observations found a cold, quiescent core in L483 (see Jørgensen 2004). In the SMA 850 μm dust continuum observations, source L483 remains as single object at the angular resolution of 26 × 19 (see Figure 16), which is consistent with previous interferometric millimeter continuum observations (see, e.g., Jørgensen 2004).

L723 is a small isolated dark cloud. At the center of the cloud core is an IRAS point source (IRAS 19156+1906), which was classified as a Class 0 protostar with a bolometric luminosity L_bol ∼ 3.0 L_☉ and a circumstellar mass M_env ∼1.2 M_☉ (Shirley et al. 2000). Anglada et al. (1991) detected two radio continuum sources (VLA 1 and VLA 2) with 15'' separation, both located within the error ellipse of the IRAS position. However, only VLA 2 was found to be associated with dense gas in the cloud core (Girart et al. 1997).

Figure 17 (a) shows the SMA 1.3 mm dust continuum images of L723 VLA 2, in which two separated compact sources (labeled MMS 1 and MMS 2) are revealed. The angular separation (35 ± 03) and mass ratio (1.0 ± 0.3) between the two sources derived in the SMA images are consistent with the results derived earlier in the OVRO 3 mm continuum observations (see Launhardt 2004). The two close compact sources are embedded in a common envelope seen in the SCUBA 850 μm observations. We note that the SMA data of L723 VLA 2 (including molecular lines results) have been published by Girart et al. (2009). Also note that VLA 7 mm continuum observations discovered two components in source MMS 2 with an angular separation of ∼1'' (see Carrasco-González et al. 2008), which would be unresolved by the SMA observations presented here.

B335 is a well-studied, isolated Class 0 protostar, and is widely accepted as one of the best protostellar collapse candidates (see, e.g., Zhou et al. 1993; Evans et al. 2005). It drives a large-scale bipolar outflow extending in the east–west direction (see, e.g., Jørgensen et al. 2007). L1157 is also a well-studied, isolated Class 0 protostar, and is well known for its prominent outflow. This outflow shows rich chemistry in various molecular lines observations (see, e.g., Bachiller et al. 2001; Arce et al. 2008), and is commonly considered the prototypical chemically active outflow. In the SMA 1.3 mm dust continuum observations, both B335 and L1157 remain as single sources (see Figures 17(b) and (c)), which is consistent with previous high angular resolution continuum observations toward the two sources (see, e.g., Harvey et al. 2003 for B335 and Chiang et al. 2010 for L1157).

L1251B is a small protostellar group located at 300 ± 50 pc (Kun & Prusti 1993). It is associated with an IRAS point source (IRAS 22376+7455), and was found to be associated with a CO outflow by Sato & Fukui (1989). Using OVRO and SMA (as well as single-dish telescopes), Lee et al. (2006b, 2007c) presented multiple-line observations toward L1251B, which shows very complex kinematics, including infall, rotation, and outflow motions.

The SMA 1.3 mm dust continuum observations show four distinct sources embedded in the common envelope seen in the single-dish observations. Following Lee et al. (2007c), the four sources are named L1251B IRS1, IRS2, SMA-N, and SMA-S (see Figure 17(d)). Both IRS1 and IRS2 are associated with the infrared sources seen in the Spitzer images, and were classified as Class 0/I transition objects by Lee et al. (2006b). Sources SMA-N and SMA-S have weak 1.3 mm dust continuum emission (see Figure 17(d)) and are not detected in the Spitzer images, and are likely still in the early stages of the Class 0 phase (see Lee et al. 2007c for more details).