A First Look at BISTRO Observations of the ρ Oph-A core

The BISTRO survey has recently begun to systematically investigate magnetic field structures in the dense cores using measurements of polarized dust emission, which is one of the most effective ways of probing the magnetic fields of such cores. Since POL-2 is newly commissioned, it is an important step to verify the consistency of our new data with those of previous studies. Therefore, we compare the POL-2 observations of ρ Oph-A with data from the SCUPOL polarimeter on the previous-generation submillimeter bolometer array on the JCMT, SCUBA (Greaves et al. 1999).

Figure 1 shows the 850 μm intensity map (Stokes I) obtained using the JCMT with SCUBA-2/POL-2, with well-known submillimeter and infrared sources labeled. The Stokes I image is consistent with previous deep submillimeter continuum images (e.g., Pattle et al. 2015).

Zoom In Zoom Out Reset image size

Figure 2 shows a comparison between the SCUBA-2/POL-2 data and the previous polarization data from SCUPOL (Greaves et al. 1999). Figures 2(a)–(c), respectively, show Stokes I, Q, and U images of the ρ Oph-A core region obtained from the JCMT with SCUBA-2/POL-2 (this work), and Figures 2(d)–(f), respectively, show Stokes I, Q, and U images of the ρ Oph-A core region obtained from the JCMT with SCUPOL (previous work; see also the SCUBA Polarimeter Legacy Catalogue; Matthews et al. 2009). The black boxes in Figures 2(a)–(c) show the regions covered by SCUPOL. As shown in Figure 2, our data are deeper and more clearly provide the morphology of the surrounding regions in all of the Stokes I, Q, and U images, although two have the same spatial resolution. However, it should be noted that the SCUPOL data are binned to generate 10'' polarization vectors.⁶⁰

Zoom In Zoom Out Reset image size

To compare the best intensity morphology with that from the previous data, we first introduce the detailed submillimeter morphology of the ρ Oph-A core and then present the polarimetric results.

Figure 1 shows the morphology of ρ Oph-A from our 850 μm Stokes I emission map. The region contains several subcores, which we outline below.

Oph-A SM1. Oph-A SM1 is a subcore located toward the peak of the 850 μm intensity (Ward-Thompson et al. 1989; also cf. Figure 1). It has the brightest submillimeter continuum in all of ρ Oph. The filamentary morphology in ρ Oph suggests that SM1 may be influenced by the B4 star Oph S1 (cf. Figure 1), which is a nearby young B-type star. Motte et al. (1998) reported that the total mass and dust temperature of Oph-A SM1 are 2 M_☉ and T ≈ 20 K, respectively.

VLA 1623. VLA 1623 is the prototypical Class 0 star (Andre et al. 1993). It drives a large-scale bipolar molecular outflow (Dent et al. 1995; Yu & Chernin 1997) and is embedded within a nearly spherical dust envelope (Andre et al. 1993). Bontemps & Andre (1997) found three emission clumps at centimeter wavelengths with the Very Large Array, which they interpreted as knots in the radio jet driving the large CO outflow (see also Chen et al. 2013). However, the position angles of the radio jet and the CO outflow differ by approximately 30°. Clump A was further resolved into two components at a high angular resolution (Chen et al. 2013) with the Submillimeter Array (SMA). VLA 1623 is also a binary system, with two components separated at high angular resolutions (Looney et al. 2000; Ward-Thompson et al. 2011). Since the POL-2 resolution is approximately 141 at 850 μm, we cannot separate these components, and we refer to them as a single source, VLA 1623, in this paper.

Other local structures. There are two filaments in the northern part of the ρ Oph-A core. These structures are consistent with not only the results obtained with SCUBA on the JCMT (Wilson et al. 1999) but also those seen in the map made with SCUBA-2 (Pattle et al. 2015) and IRAM (Motte et al. 1998) results. In addition to the filaments, Wilson et al. (1999) reported that there are two arcs of emission in the direction of the northwest extension of the VLA 1623 outflow. The outer arc appears relatively smooth at 850 μm, while the inner arc breaks up into a number of individual clumps, some of which are known protostars.

Polarized thermal emission from dust grains in clouds offers an ideal probe of the magnetic field structure on multiple scales, from protostellar disks to cores and clumps (e.g., Matthews et al. 2001; Crutcher et al. 2004).

Figure 3 shows our submillimeter polarization vector maps of the ρ Oph-A region observed with SCUBA-2/POL-2. Since it is worthwhile to directly compare our data with the previous submillimeter polarization vector map, Figure 3 is prepared with the same criteria, I > 0, P/δP > 2, and δP < 4%, as were used in the previous results by SCUPOL (see Figure 44 of Matthews et al. 2009). The selection criteria here are mainly for the purpose of the comparison with the SCUPOL data; however, we have found this to be fairly reasonable to see the magnetic field structure in this region by changing various P/δP or δP selections. In addition, Figure 3 suggests that the vector maps with both P/δP > 2 and P/δP > 3 are almost the same, if we use the additional criterion of δP < 4%. Without this δP criterion, the vector map with P/δP > 2 has many more vectors but also has rms noise values of δP and δθ too high to allow interpretation of the magnetic field behavior. Therefore, we use the criteria of I > 0, P/δP > 2, and δP < 4% in the following discussion to maximize the number of polarization vectors that can be used for our discussion below on the magnetic field directions. The angle errors (δθ) of approximately 15° in the 2σ case are acceptable for such discussions. Therefore, we show both the 2σ and 3σ cases in several figures and use 2σ data for the discussion on magnetic field directions.

Zoom In Zoom Out Reset image size

Our data are more sensitive than those obtained by SCUPOL, as shown in Section 4.1. The new submillimeter polarization vectors inside the dense regions agree well with the results by Matthews et al. (2009), especially in the bright region near SM1. The dominant submillimeter polarization position angle in the bright region is approximately 130° (as discussed in the following section). We have also checked whether our new data are consistent with the JCMT 800 μm aperture polarimetric data of Holland et al. (1996). The measured positions are not exactly the same, but both the P and θ values are consistent with each other between the two studies.

Note that there are clear inconsistencies between the SCUBA-2/POL-2 and SCUPOL data in the outer parts of the Oph-A core region. To the southeast and south of the core, the SCUPOL data show more numerous polarization vectors, even at a very low intensity level, while to the northwest and northeast, the SCUBA-2/POL-2 data reveal more vectors. Since our data have a higher S/N, as shown in Figure 2, we believe that our SCUBA-2/POL-2 data are more reliable in the faint outer regions down to approximately 30 mJy beam⁻¹, while care is necessary when using the SCUPOL vectors in the outer regions. The reason why the SCUPOL data have more vectors in some outer core regions is not clear. However, we note that the SCUPOL maps were made by chopping, while POL-2 was in a scanning mode. Therefore, the chopping effect cannot be excluded in the SCUPOL data, which were taken at different times by different observers.

Based on the robustness toward the fainter regions mentioned above, our data clearly show the polarizations in the fainter regions surrounding the core, and the degrees of polarization are much higher (>5%) in the outer envelope. This trend is clear in our polarization map, whose vector length is proportional to the degree of polarization (see also Figure 7).

Figure 4 shows the degree of polarization errors (δP) versus the inverse intensity (${I}^{-1}$); the polarization uncertainty increases steadily with decreasing intensity. For δP > 4%, we see significant scatter in this relation, whereas the data with δP < 4% are fairly well correlated. There are five vectors with δP < 4% that show substantial scatter from the main trend and are labeled in Figure 4 (cf. Table 1). Aside from these five cases, vectors with δP < 4% appear to be robust. All five anomalous positions (ID: 164, 237, 238, 239, 240) are located near the map boundary where the noise levels are higher. Vector 164 is located in the east of Oph-A, vector 237 is located between the northwest filamentary structure and GSS 26, and vectors, 238, 239, and 240 are located in the upper part of the northeast filamentary structure. Note that including these five vectors does not affect our results. Figure 4 also shows that our polarization data present a large scatter when the intensity levels are less than approximately 30 mJy beam⁻¹, which corresponds to N(H₂) ∼ 4 × 10²¹ cm⁻², assuming a temperature of 10 K (Kauffmann 2007).

Zoom In Zoom Out Reset image size

The ρ Oph molecular cloud has been observed with 1.3 mm continuum mapping (Motte et al. 1998) and line mapping (Umemoto et al. 1999; see also White et al. 2015), and the ρ Oph-A core region is one of the most obvious sources. Matthews et al. (2009) presented a bulk analysis of SCUPOL 850 μm polarization vector maps, which include the ρ Oph-A core. The submillimeter polarization position angle is about 130°, on average (measured east of north), which indicates a magnetic field direction of approximately 40° (by rotating the submillimeter polarization vectors by 90°). This angle is consistent with the well-known 50° component determined via infrared polarimetry observations (Sato et al. 1988; Kwon et al. 2015). Therefore, the magnetic field seems largely consistent between the outer low-density cloud and the high-density cores.

To investigate magnetic field structures in this region in more detail, we use the POL-2 polarization vectors rotated by 90°, as shown in Figure 5. Figure 6 shows the inferred morphology of the magnetic field in the ρ Oph-A core region. In this figure, the vector maps are shown in two ways: one selected with the polarization S/N (P/δP > 2 or 3), and the other selected with the intensity S/N (I/δI > 20). The latter intensity selection is shown because selecting by S/N in P will tend to bias the polarization data sample to high P values (especially toward regions of low intensity), so a comparison with a sample selected by I/δI is made to show that without this bias, the polarization fraction is still larger on average for cloud sightlines in the envelope. Figure 7 demonstrates that this correlation is robust, as also seen from the negative correlation between the degrees of polarization and intensities in both the P and I selection data. We find that P ∝ I^γ, where γ ∼ −0.8 for the P selection and −0.7 for the I selection. Note that there is a larger dispersion at the low-intensity regions in the I selection because not only high P data but also several low P data exist in the I selection. This trend might be due to a combination of several factors, such grain alignment and magnetic field geometry. A detailed discussion will be presented elsewhere.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We have found by eye that there are at least 10 distinct magnetic field components in the core region, and we refer to them as "components a–j" (see Figure 8). Please note that our division of these components does not mean that these field components are always independent, but all or some of them could be smoothly connected with each other. The aim of the region division here is mainly to identify the change of directions and degrees of the polarization vectors and to compare them with the near-infrared polarization data. A summary of these components is as follows.

• (a)  
  Small P (<3%) and ∼50° component at SM1 and VLA 1623 around the center of the observed field of view.
• (b)  
  Large P and ∼40° component near A-MM 7 to the east of A-MM 5.
• (c)  
  Large P and ∼20° component near A-MM 5.
• (d)  
  Large P and ∼100° component at A-MM 4.
• (e)  
  Small P (<3%) and ∼100° component between A-MM 5 and SM1N.
• (f)  
  Large P and ∼80° component to the west of SM1.
• (g)  
  Large P and ∼70° component to the east of SM1 and SM1N.
• (h)  
  Large P and ∼80° component at A-MM 3.
• (i)  
  Small P (<3%) and ∼80° component between SM2 and A-MM 8.
• (j)  
  Large P and ∼120° component between SM2 and A-MM 8.

Zoom In Zoom Out Reset image size

Figure 8 illustrates that these components differ from each other either in polarization position angle or degree of polarization (see also Table 2). Components a, c, and i are already seen in and consistent with the SCUPOL data (Tamura 1999). Components b, d, e, f, g, h, and j are additionally identified in our SCUBA-2/POL-2 data. One can also see the polarization vectors associated with components b, d, e, g, and j in Matthews et al. (2009). We also note that our results suggest that the magnetic field is mostly well organized (rather than disordered due to turbulence; see Section 5.2).

In the central region around SM1 (component a), the vectors are well aligned with the 50° magnetic field component observed in the surrounding medium on various scales (see Section 5.4). Although the average direction of the main component is approximately 50°, the magnetic field tends to be locally perpendicular (approximately 100°–110°) to the arcstructure (south part of region f). Between SM1N and A-MM 6, the magnetic field direction is almost east–west (component e), while the arc extends to the northeast or northwest, and the magnetic field directions extend toward LFAM 1 and GSS 30–1 (component f). A perpendicular field relative to the core shape (i.e., the elongation of the arcstructure between the northeast and northwest filaments) is important for the formation and growth of this core. Such orthogonal fields are often seen in the densest parts of the cloud or cloud cores (Tamura et al. 1987, 1988; Nagai et al. 1998; André et al. 2013; Palmeirim et al. 2013; Matthews et al. 2014; Fissel et al. 2016; Planck Collaboration et al. 2016).

There are other local structures besides the 50° component. To the north of A-MM 6 (component c), the magnetic field direction is almost north–south, and to the north of A-MM 7, the magnetic field direction is almost north–east (component b), which is the same as the direction of the northeast filament. Notable are the low degree of polarization near SM1N and some deviation in magnetic field direction near VLA 1623 and its outflow region.

In this paper, we have assumed that the 850 μm emission measured by SCUBA-2 is dominated by the thermal dust continuum emission. However, the continuum emission can be contaminated by CO (3–2) line emission (Drabek et al. 2012). Figure 9 shows an overlay of the CO (3–2) line emission from White et al. (2015) on the 850 μm continuum map. In the dense center of Oph-A, the CO contamination fraction is typically <1%. However, in the brightest regions of CO emission from the outflow from VLA 1623, the contamination fraction can be much higher (Pattle et al. 2015). The regions that have high CO contamination have very low column density values and are mostly along the jet axis between Oph-B and Oph-C/E/F, outside our field of view. In the dense center of Oph-A, the fractional contribution of CO is <1% and generally does not exceed 10% anywhere on source. The ring-shaped region seen in our Stokes I image to the west of Oph-A is dominated by the CO emission rather than the thermal dust emission. Since the CO emission is weak toward the bright dust emission (<5%), even if it is polarized by the Goldreich–Kylafis effect, it will contribute minimally to our results.

Zoom In Zoom Out Reset image size

Polarization arising from dust grains, which are aligned with their major axes perpendicular to the magnetic field (e.g., Hoang & Lazarian 2009), allows us to estimate the magnetic field direction. However, the present uncertainties in theories of dust grain alignment limit the ability of current techniques to trace magnetic fields without ambiguities (see Lazarian 2007; Lazarian et al. 2015 for a review).

The most common method to infer the magnetic field strength from polarized dust emission is the Davis–Chandrasekhar–Fermi method (more commonly referred as the Chandrasekhar–Fermi (CF) method; Davis 1951; Chandrasekhar & Fermi 1953; see also Houde et al. 2016 and Pattle et al. 2017). The CF method infers the magnetic field strength by statistically comparing the dispersion in the polarization orientation with the dispersion in velocity. Therefore, the magnetic field strength projected on the plane of the sky can be calculated by

$$\begin{eqnarray}&&{B}_{p}={ \mathcal Q }\sqrt{4\pi \rho }\ \displaystyle \frac{\delta {v}_{\mathrm{los}}}{\delta \theta },\end{eqnarray} \tag{ 7 }$$

assuming that velocity perturbations are isotropic (Ostriker et al. 2001). In Equation (7), ${ \mathcal Q }$ is a factor to account for various averaging effects (see Crutcher et al. 2004 and Houde 2004 for details), ρ is the mean density of the cloud, δv_los is the rms line-of-sight velocity, and δθ is the dispersion in the polarization angle. To estimate the magnetic field strength in the ρ Oph-A core region, a correction factor of ${ \mathcal Q }=0.5$ is adopted here because the magnetic field appears to be ordered (Ostriker et al. 2001; Houde 2004; also see Falceta-Gonçalves et al. 2008; Novak et al. 2009; Pattle et al. 2017). Since we apply this formula only to the subregions where the angle dispersion is relatively small (≤25°), and the velocity dispersion of the molecular lines tracing high-density regions is available in the next paragraph, ${ \mathcal Q }=0.5$ is appropriate, as simulated by Ostriker et al. (2001). Note that if the turbulence correlation length is not resolved and therefore their simulation assumption is not valid, the case of the ${ \mathcal Q }$ factor can be much lower (Heitsch et al. 2001; see also Houde et al. 2009). Then Equation (7) can be expressed as follows (Lai et al. 2002):

$$\begin{eqnarray}&&{B}_{p}=8.5\displaystyle \frac{\sqrt{{n}_{{\rm{H}}2}/({10}^{6}\,{\mathrm{cm}}^{-3})}\,{\rm{\Delta }}{\rm{v}}/(\mathrm{km}\,{{\rm{s}}}^{-1})}{\delta \theta (^\circ )}\,\mathrm{mG}.\end{eqnarray} \tag{ 8 }$$

Here ${n}_{{{\rm{H}}}_{2}}$ is the number density of hydrogen molecules and Δv is the line width.

As mentioned in the previous section, there are several magnetic field components in the ρ Oph-A core region. Since they are different from each other either in direction or degree of polarization, we estimate the magnetic field strength of each component separately. To investigate their magnetic field strengths individually, we estimated the median polarization position angles, which indicate the local average magnetic field directions of each component. Table 2 shows the median degrees of polarizations and position angles calculated using Stokes Q and U in each region. Figure 8 shows the vectors in each region averaged over each region.

The local average density in each of components a–j is calculated from our Stokes I data, assuming that the core depth is equal to the geometric mean size of each subcore where the polarization data exist and ranges from 2 × 10⁶ to 7 × 10⁴ cm⁻³. Since the ρ Oph-A core has a complex magnetic field structure showing various directions in each subcore, we do not attempt to apply to the CF method to the entire core, but only to the subcores showing a relatively well-defined magnetic field direction. In components a, d, and e, André et al. (2007) estimated a velocity dispersion of 0.26, 0.15, and 0.17 km s⁻¹, respectively. These are nonthermal line dispersions from N₂H⁺ (1–0) observations. Using these values with the standard deviation in field direction of 15, 58, and 27 found in each region, the magnetic field strength projected on the plane of the sky is calculated as Bp ∼ 5, 0.2, and 0.8 mG (cf. Table 2). The estimated magnetic field strength in the ρ Oph-A core region is larger than that in other molecular clouds derived using the CF method (namely, 20–200 μG; Andersson & Potter 2005; Poidevin & Bastien 2006; Alves et al. 2008; Kwon et al. 2010, 2011; Sugitani et al. 2011; Kusune et al. 2015) but comparable to that in the Orion A region (see, e.g., Pattle et al. 2017). Our high magnetic field strengths may be attributed to using the higher H₂ densities associated with the subcores rather than the lower H₂ densities associated with the larger ρ Oph-A core. Thus, we conservatively take these field strengths as an order-of-magnitude estimate. These values are still representative of the field strength toward the subcores in ρ Oph-A and can be taken as an upper limit for the surrounding gas.

Finally, it should be noted that there are certain limitations in the CF technique, such as the effect of the limited telescope resolution (Heitsch et al. 2001). Also note that our estimates are only for some subregions where the field dispersions are relatively small. Therefore, both of these effects tend to bias toward a high magnetic field strength. In addition, more sophisticated applications of the CF technique, such as those described in Hildebrand et al. (2009) or Pattle et al. (2017), would be desirable in future works.

Our polarimetric data will be useful to discuss the correlation between the magnetic field and the velocity field in each core. However, this is beyond the scope of this first-look paper. Therefore, in this section, we show an example of a possible correlation between magnetic field and velocity gradient.

Strong Alfvénic turbulence develops eddy-like motions perpendicular to the local magnetic field direction (Goldreich & Sridhar 1995). Very recently, González-Casanova & Lazarian (2017) proposed that this fact can be used to study the direction of magnetic fields by using the velocity gradient calculated from the centroid velocity. The centroid velocity is an intensity-weighted average velocity along the line of sight (e.g., Miesch et al. 1999). Here we try to compare the magnetic field direction in the ρ Oph-A core region with the centroid velocity components.

André et al. (2007) measured subsonic or transonic levels of internal turbulence within the condensations, and their result supports the view that most of the L1688 starless condensations are gravitationally bound and prestellar in nature. Figure 10 shows a comparison between the magnetic field direction (this work) and the centroid velocity components of N₂H⁺(1–0) spectra (André et al. 2007). The apparent main velocity core gradient (indicated by arrows in Figure 10) appears to be roughly perpendicular to the magnetic field orientation traced by POL-2. Certainly, these observations should be compared with theoretical modeling using the physical parameters of the ρ Oph-A core in future.

Zoom In Zoom Out Reset image size

Polarimetry in the ρ Oph-A core region was reported previously by several authors at other wavelengths. Sato et al. (1988) carried out near-infrared polarimetry (in the K band only) of 20 sources that are embedded within the densest region of the ρ Oph dark cloud with a single-channel detector, and they suggested that there are three dominant components of the polarization position angles 0°, 50°, and 150°. Recently, Kwon et al. (2015) presented wide and deep near-infrared polarimetry (in the JHK_s bands) of the ρ Oph regions, which corresponds to the densest part of L1688. Since they cover a wider region than our observations (but much more sparsely due to the limited number of stars available for the aperture polarimetry), we compare their polarimetry data covering a 40' × 40' region with our submillimeter data. In this active cluster-forming region, they found that the magnetic fields appear to be connected from core to core, rather than as a simple overlap of the different cloud core components. Putting it differently, the magnetic field morphology seems to be connected between different cores in the ρ Oph molecular cloud complex. In addition, comparing their near-infrared polarimetric results with the large-scale magnetic field structures obtained from a previous optical polarimetric study (Vrba et al. 1976), they suggested that the magnetic field structures in the ρ Oph core were distorted by the cluster formation in this region, which may have been induced by shock compression due to wind/radiation from the Scorpius–Centaurus association. Also note that there is 350 μm submillimeter polarization from the CSO in Dotson et al. (2010) for ρ Oph-A. Their data are broadly consistent with our 850 μm map.

Our new submillimeter polarimetry demonstrates that one of the main polarization position angles in Oph-A is approximately 50° (see Figures 3–10) and so is well aligned with the 50° magnetic field found in the near-infrared (Kwon et al. 2015; see also Figure 5 in this paper for the comparison within the same field of view). Kwon et al. (2015) found that the "50° component" is the dominant magnetic field component in the observed region; it can be seen as a distinct clump in the diagram plotting degree of polarization versus polarization angle (Figure 9 of Kwon et al. 2015) and in the histogram of polarization position angles (Figure 10 of Kwon et al. 2015). This component is seen in the northeast regions of ρ Oph-A (and in ρ Oph-B and ρ Oph-E on a large scale, regions not covered in this work). The "0° component" can be seen from ρ Oph-A toward ρ Oph-AC (located at the southeastern region of ρ Oph-A, which is not shown in our submillimeter map; cf. Kwon et al. 2015). In contrast, in Oph-A, both the 0° and the 50° components exist.

Figure 11 shows the histogram of polarization position angles for the 90° rotated submillimeter polarization vectors, as well as for the H-band polarization position angles from Kwon et al. (2015). The distribution is relatively widespread, but if we refer to both the H-band polarization vector map (Figure 8 of Kwon et al. 2015) and this histogram, we see several components, of which the components at 0° and 50° are most clearly seen. As shown in Figure 11, the distribution of the polarization position angles obtained from submillimeter polarimetry is in relatively good agreement with that obtained from near-infrared polarimetry for the 0° and 50° components but not for the 150° component. Note that since our submillimeter map covers a small part of the area covered by the near-infrared polarimetry survey and we see much higher column density regions of ρ Oph-A, there is also some inconsistency between the distributions of submillimeter and near-infrared polarization angles. Therefore, our results indicate that both submillimeter emission polarization and near-infrared dichroic polarization may trace the magnetic field structures associated with the ρ Oph-A core region at different spatial scales and regions along the line of sight.

Zoom In Zoom Out Reset image size

Previous observations have shown agreement between the magnetic field structures seen at various wavelengths, such as near- and far-infrared or submillimeter wavelengths (e.g., Tamura et al. 1996, 2007; Kandori et al. 2007; Kwon et al. 2011). Our new results are consistent with this behavior, although the greatest-density regions can be traced only by submillimeter polarimetry. The 50° component seen in the lower-density regions of the submillimeter map around the edge of the core (the northeast filament; cf. Figure 1) is consistent with the 50° component seen in the lower-density tracer of near-infrared polarization (Kwon et al. 2015), giving us still further confidence in our observations. Our data are also consistent with the recently released HAWC+ data taken by SOFIA (Santos et al. 2018). A combination of polarimetric observations over wavelengths and scales observed by instruments such as ALMA and by 8 m class optical/infrared telescopes will become more important in the future to test the range of scales over which this behavior holds.