EVLA Observations of the Barnard 5 Star-Forming Core: Embedded Filaments Revealed
Jaime E. Pineda, Alyssa A. Goodman, Héctor G. Arce, Paola Caselli, Steven Longmore, Stuartt Corder
DD RAFT VERSION
2, N
OVEMBER
1, 2018, JEP
Preprint typeset using L A TEX style emulateapj v. 11/10/09
EVLA OBSERVATIONS OF THE BARNARD 5 STAR-FORMING CORE: EMBEDDED FILAMENTS REVEALED J AIME
E. P
INEDA , A
LYSSA
A. G
OODMAN , H ÉCTOR
G. A
RCE , P AOLA C ASELLI , S TEVEN L ONGMORE , S TUARTT C ORDER
Draft version 2, November 1, 2018, JEP
ABSTRACTWe present a ∼ (cid:48) × (cid:48) Expanded Very Large Array (EVLA) mosaic observations of the NH (1,1) emissionin the Barnard 5 region in Perseus, with an angular resolution of 6 (cid:48)(cid:48) . This map covers the coherent region,where the dense gas presents subsonic non-thermal motions (as seen from single dish observations with theGreen Bank Telescope, GBT). The combined EVLA and GBT observations reveal, for the first time, a strikingfilamentary structure (20 (cid:48)(cid:48) wide or 5,000 AU at the distance of Perseus) in this low-mass star forming region.The integrated intensity profile of this structure is consistent with models of an isothermal filament in hydro-static equilibrium. The observed separation between the B5–IRS1 young stellar object (YSO), in the centralregion of the core, and the northern starless condensation matches the Jeans length of the dense gas. This sug-gests that the dense gas in the coherent region is fragmenting. The region observed displays a narrow velocitydispersion, where most of the gas shows evidence for subsonic turbulence, and where little spatial variationsare present. It is only close to the YSO where an increase in the velocity dispersion is found, but still displayingsubsonic non-thermal motions. Subject headings:
ISM: clouds — stars: formation — ISM: molecules — ISM: individual (Perseus MolecularComplex, B5) INTRODUCTIONMolecular clouds (MCs) observed using low-density trac-ers display velocity dispersions much larger than the thermalvalue (e.g., Zuckerman & Evans 1974; Larson 1981; Myers1983). These supersonic motions are usually attributed to“turbulence," and a variety of recent numerical models of tur-bulence can reproduce qualitatively realistic clouds. Measure-ments of MCs’ energy budgets show that turbulence must bedissipated in order for dense cores (where individual or smallgroups of stars form) to collapse and form stars.Dense cores have been studied using NH (1,1) maps,which traces material with densities of a few 10 cm − . Ithas been found that dense cores show an almost constantlevel of non–thermal motions, within a certain “coherence”radius (Goodman et al. 1998). The term “coherent core” isused to describe the dense gas where non–thermal motionsare roughly constant, and typically smaller than the thermalmotions, independent of scale (see also Caselli et al. 2002).Recent observations carried out with the 100-m Robert F.Byrd Green Bank Telescope (GBT) have allowed us to cre-ate large scale maps of NH (1,1) towards four star-formingregions in the Perseus Molecular Cloud: B5, IC348–SW,L1448, and L1451 (Pineda et al. 2010, 2011). One of the mainresults of Pineda et al. (2010) is the clear observation (for the [email protected] ESO, Karl Schwarzschild Str. 2, 85748 Garching bei Munchen, Ger-many UK ALMA Regional Centre Node, Jodrell Bank Centre for Astro-physics, School of Physics and Astronomy, University of Manchester,Manchester, M13 9PL, UK Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cam-bridge, MA 02138, USA Department of Astronomy, Yale University, P.O. Box 208101, NewHaven, CT 06520-8101, USA School of Physics and Astronomy, University of Leeds, Leeds LS29JT, UK North American ALMA Science Center, 520 Edgemont Road, Char-lottesville, VA 22903, USA National Radio Astronomy Observatory, 520 Edgemont Road, Char-lottesville, VA 22903, USA first time) of the sharp transition (at 30 (cid:48)(cid:48) resolution) betweenthe coherent section of the B5 dense core and the more tur-bulent dense gas outside it (see also Pineda et al. 2011, forresults on the other regions surveyed). However, the angularresolution of these observations did not allow us to study ingreat detail the spatial variations of the velocity dispersion orcolumn density.In this letter, we present new NH (1,1) observations ofB5 obtained with the Expanded Very Large Array (EVLA)through the Open Shared Risk Observing (OSRO) program,see Perley et al. (2011) for a description of the EVLA project.These data are combined with previous observations from theGBT to study the dense gas traced by NH at high angular andspectral resolution. Here, we present results from the com-bined EVLA and GBT observations that enable us to analyzethe kinematic properties of the dense gas traced by NH (1,1)and also to study the radial profile of new filamentary struc-ture found within this core. DATAObservations were carried out with the EVLA of the Na-tional Radio Astronomy Observatory on May 16th, 2010(project 10A-181). We used the high-frequency K-bandreceiver and configured the WIDAR correlator to observea 1 MHz window around the NH (1,1) rest frequency(23.6944950787 GHz). The correlator generated 256 chan-nels across the window, giving a 3.90625 kHz channel sepa-ration, equivalent to 0 .
049 km s − at the observed frequency.This configuration covers the main hyperfine component andalso one of the inner pair of satellite lines for NH (1,1). Atthis frequency the primary beam of the array is about 1 . (cid:48) .The array was in the compact (D) configuration, which coversbaselines from 35 meters up to 1 km, and included 26 anten-nas. The observations covered the entire region of interestusing 27 pointings. The observations were carried out underfair weather ( τ = 0 . X -band a r X i v : . [ a s t r o - ph . GA ] J un J. E. Pineda et al.reference pointing checks were performed every 60 minutesusing the quasar J0336+3218.The raw-data were reduced using CASA image processingsoftware. The signal from each baseline was inspected, andbaselines showing spurious data were removed prior to imag-ing. Each channel was cleaned separately according to thespatial distribution of the emission. The clean map was im-aged using a 6 (cid:48)(cid:48) circular beam and corrected by the primarybeam. The images were created using multi-scale clean (withscales of 4 and 12 arcsec and smallscalebias=0.2) with a ro-bust parameter of 0 . (1,1) emissionwhich arises from angular scales not sampled by the interfer-ometer. We included the NH (1,1) single dish data obtainedwith the GBT (Pineda et al. 2010) as a prior (or model) duringthe imaging process to recover the large scale emission. Thefinal noise achieved is 14 mJy beam − per channel.The integrated intensity map (see Figure 1) shows a few re-gions with negative emission, which would suggest the pres-ence of imaging artifacts. However, careful inspection of thedata confirms that the spectra towards those positions do notshow any clear negative bowl associated with imaging arti-facts, in fact, these regions are about the − σ level still con-sistent with the map noise. RESULTSLeft panel of Figure 1 shows the integrated intensity mapobtained using the GBT at 30 (cid:48)(cid:48) resolution. Gray contours inFigure 1 show the extent of NH (1,1) emission. The orangecontours show the regions in the GBT data within which thenon-thermal velocity dispersion is sub-sonic. The blue con-tour shows the region observed with the EVLA and presentedin the right panel, and it covers the entire central region wherethe sub-sonic non-thermal velocity dispersions are observed.The resulting NH (1,1) integrated intensity map for B5 isshown in right panel of Figure 1, and it covers a region of size6.5 (cid:48) × (cid:48) .The new high resolution integrated intensity map (rightpanel of Figure 1) reveals that within the region of subsonicnon-thermal motions found in the single dish data (orangecontour) filamentary structures appear. These filaments arenarrow, with widths of ≈ (cid:48)(cid:48) or 5,000 AU at the distance ofPerseus (250 pc; Hirota et al. 2008).An important feature of molecular line observations is theability to probe the kinematics of the gas. Here we fit simulta-neously all hyperfine components of the NH (1,1) line usinga forward model previously presented by Rosolowsky et al.(2008), see also Pineda et al. (2010). This method describesthe emission at every position with a centroid velocity ( v LSR ),velocity dispersion ( σ v ), kinetic temperature ( T k ), excitationtemperature ( T ex ) and opacity ( τ ), while also including theresponse of the frequency channel using a sinc profile. Sincethe kinetic temperature is only used to predict the NH (2,2)line (not observed due to the constraints in the OSRO pro-gram), it does not have any effect on the remaining parame-ters. We use a fixed value of the kinetic temperature of 10 Kfor the entire region, which is consistent with the results ob-tained by Pineda et al. (2011) using the single dish data. Ifthe resulting fit does not provide an accurate velocity disper-sion, σ v < .
05 km s − (velocity dispersion narrower than theexpected thermal value for gas at 6 K) or σ σ v > . σ v (signal- to-noise for the velocity dispersion less than 5), the fit is re-peated but with a fixed value of 5 K for excitation temperature.The centroid velocity and velocity dispersion maps are shownin Figure 2.The centroid velocity map shows little variation ( < . − ) across the entire region, see Figure 2a. The regionclose to the YSO displays a velocity gradient at a positionangle that is close to the outflow (shown by the arrows) butin opposite direction. The velocity dispersion map, see Fig-ure 2b, shows vast regions where small and uniform velocitydispersion are found. It is only towards positions close to theYSO where slightly broader lines are found.Figure 3 presents the distribution of the derived velocitydispersion towards B5. Two histograms are shown Figure 3depending on the proximity to the YSO in B5: (a) positionsclose to the YSO (separated by 2 beams or less, < (cid:48)(cid:48) ) in red,and (b) all other pixels in black. For points far from the YSOthe velocity dispersions are small and the velocity dispersiondistribution (black histogram in Figure 3) is narrow, almostevery pixel at a distance larger than 2 beams (12 (cid:48)(cid:48) ) from thecentral YSO displays sub-sonic non-thermal motions. Thevelocity dispersion distribution of pixels far from the YSO,black histogram in Figure 3, peaks at a value lower than whatis expected if the non-thermal component, σ NT , is equal tohalf the thermal velocity dispersion, 0 . s , ave . The velocitydispersion of points close to the YSO, red histogram in Fig-ure 3, presents lines broader than the rest of the core, but theystill display velocity dispersions with a subsonic non-thermalcomponent. This increase in the velocity dispersion might bethe effect of the radiation from the embedded YSO or due tothe interaction between the outflow or stellar wind and thedense gas.Herschel observations of the IC 5146 star forming region(Arzoumanian et al. 2011) revealed filamentary structure seenin the column density maps (see also André et al. 2010). Ar-zoumanian et al. (2011) fitted the column density profile offilaments with a cylindrical filament model, Σ ( r ) = A p ρ c R f lat (cid:0) + ( r / R f lat ) (cid:1) ( p − / , (1)where ρ c is the filament’s central density, r is the cylindri-cal radius, p is the power-law density exponent at large radii, R f lat is the radius of the density profile inner flat section, and A p is a finite constant factor dependent on p and the filamentinclination angle. They find filaments which are well fit with adensity exponent in the range p = 1 . − . p = 4, is expected (Ostriker 1964).Here we focus our attention towards the filament shown inFigure 1 by the yellow box. Since this filament is almost per-fectly aligned in the North-South direction we perform a se-ries of horizontal cuts, and define the radius as the distancefrom the peak at a given cut. The average velocity disper-sion and integrated intensity emission profiles along the fil-ament are shown in panels (a) and (b) of Figure 4, respec-tively, where the spread in the distribution is shown by theyellow area. Figure 4a shows that the velocity dispersiondoes not change across the filament, and it is consistent withsubsonic non-thermal velocity dispersions (delimited by thedotted line). Figure 4b shows the integrated intensity profile,which is much wider than the beam of the combined EVLAand GBT data (shown by the blue dash line). Since we doarnard 5: Filaments revealed 3 h m s m s s s s s RA (J2000)32 ° ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ D e c ( J ) h m s m s s s s s RA (J2000)32 ° ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ D e c ( J ) GBT beam0.1 pc 03 h m s s s RA (J2000)32 ° ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ ʹ ʹʹ D e c ( J ) EVLA 5,000 AU (K km s -1 ) -1 2 5 (mJy beam -1 km s -1 ) -2 49 100 StarlesscondensationYSO B5-IRS1
Figure 1.
Left panel:
Integrated intensity map of B5 in NH (1,1) obtained with GBT. Gray contours show the 0.15 and 0.3 Kkms − level in NH (1,1) integratedintensity. The orange contours show the region in the GBT data where the non-thermal velocity dispersion is sub-sonic. The young star, B5–IRS1, is shown bythe star in both panels. The outflow direction is shown by the arrows. The blue contour shows the area observed with the EVLA, and the red box shows thearea shown in the right panel. Right panel:
Integrated intensity map of B5 in NH (1,1) obtained combining the EVLA and GBT data. Black contour shows the50 mJybeam − kms − level in NH (1,1) integrated intensity. The yellow box shows the region used in Figure 4. The northern starless condensation is shown bythe dashed circle. h m s s s s s RA (2000)32 ° ′ ′′ ′ ′′ ′ ′′ ′ ′′ ′ ′′ ′ ′′ ′ ′′ D e c ( ) λ J = 44,300 AUEVLA beam5,000 AU(a) (b) V LSR (km s -1 ) σ v (km s -1 ) Figure 2. (a)
Centroid velocity map of B5 obtained by fitting all observed NH hyperfine components simultaneously. The position of the YSO is shown by thestar, and the orientation of the outflow is shown by the arrows. The estimated Jeans length, λ J , of the dense gas is shown by the vertical scale bar, see Section 4.The EVLA beam size is shown at the bottom left. The orange contour shows the region with subsonic non-thermal motions identified from the GBT data. Graycontour shows the 50 mJybeam − kms − level in NH (1,1) integrated intensity, same as right panel of Figure 1. The Black box shows the region used in Figure 4. (b) Velocity dispersion map derived from fitting all NH hyperfine components simultaneously. J. E. Pineda et al. σ v (km s -1 )0.000.020.040.060.08 F r equen cy Far from YSOClose to YSO (<12 arcsec)B5 0.5 c s,ave = σ NT c s,ave = σ NT Figure 3.
Velocity dispersion distribution for positions at a distance to B5–IRS1 larger (smaller) than 12 (cid:48)(cid:48) is shown in black (red). Blue vertical linesshow the expected velocity dispersion for two values of the velocity disper-sion non-thermal component ( σ NT ): 0 . s , ave , and c s , ave , where c s , ave is thesound speed of the average particle ( µ = 2 .
33) assuming T kin =10 K. The ve-locity dispersions, σ v , obtained far enough from the YSO (black histogram)are consistently below the values expected for supersonic non-thermal mo-tions. Positions close to the YSO (red histogram) display larger velocity dis-persions than the rest of the region, but they are still consistent with subsonicnon-thermal motions. not observe the NH (2,2) with the EVLA we cannot providea column density profile, but we use NH (1,1) integrated in-tensity as a proxy for the total column density (Rosolowskyet al. 2008; Friesen et al. 2009; Foster et al. 2009; Pineda et al.2011). The integrated intensity profile is fitted with a modelwhich follows equation 1, and the best fit results for p = 2 and p = 4 are shown in Figure 4. The figure shows that this fila-ment in B5, in opposition to what is found by Arzoumanianet al. (2011) on larger filaments, is better fit by an isothermalhydrostatic filament model (see also Hacar et al., submitted,or Bourke et al., submitted, for other examples of objects withsimilar column density profiles). DISCUSSION AND CONCLUSIONThe observations presented here show that subsonic non-thermal velocity dispersions display little variations acrossthe region of coherence, and even across a filament. Also,the coherent region is far from presenting an uniform columndensity (as traced by the NH (1,1) integrated intensity), asfilamentary substructures are revealed. But the presence ofsubstructures within the region of coherence should not besurprising, the lack of turbulent support might allow exter-nal forces or Jeans-like instabilities to generate over-densitieswithout much difficulties.Filaments have received special attention recently. Theyappear as a natural outcome from turbulent simulations ofmolecular clouds (e.g., Klessen et al. 2004), and also fromthe fragmentation of a modulated layer (Myers 2009). RecentHerschel observations have shown that filaments are com-monly found in star-forming regions (André et al. 2010; Ar-zoumanian et al. 2011). However, these filaments presentdifferent properties than those presented here : (a) they arebigger structures, with a characteristic width of ∼ (b)B5 Filament ProfileBeamModel (p=4)Model (p=2)-30 -20 -10 0 10 20 30Radius (arcsec)020406080 F l u x ( m Jy bea m - k m s - ) σ v ( k m s - ) (a) 0.5 c s,ave = σ NT c s,ave = σ NT Figure 4. (a)
Average velocity dispersion profile perpendicular to the fil-ament shown by the yellow box in Figure 1. The dispersion of the radialprofile along the filament is shown by the yellow area. Similar to Figure 3,blue horizontal lines show the expected velocity dispersion for two values σ NT : 0 . s , ave , and c s , ave . (b) Average integrated emission profile perpendic-ular to the filament, with the dispersion shown in yellow. The beam responseis shown by the blue line, while two filament models are shown by the redcurves. ments previously studied. The filaments in B5 are embeddedin a dense region with subsonic turbulence, and the filamentshown in Figure 4 is better fitted with the hydrostatic isother-mal filament model. The differences between the filamentsidentified using Herschel and EVLA are due to physical dif-ferences between the structures. The Herschel identified fil-aments are found at low column densities and have a char-acteristic size of ∼ . ∼ λ J = 44 , (cid:18) T
10 K (cid:19) / (cid:18) (cid:104) n (cid:105) cm − (cid:19) − / (cid:16) µ . (cid:17) − AU , (2)where T is the temperature of the gas, (cid:104) n (cid:105) is the average den-sity, and µ is the mean molecular weight. For the averagevalues of density and temperature in B5 from the dense gasarnard 5: Filaments revealed 5( T = 10 K and (cid:104) n (cid:105) = 10 cm − , Pineda et al. 2011) the Jeanslength correspond to the separation between the B5–IRS1YSO and the starless condensation in the northern part of theregion observed. Fragmentation has been argued to explainthe stars separation in young star-forming regions (e.g., Hart-mann 2002; Teixeira et al. 2006, 2007) and the mass functionof cores or condensations (e.g., Testi & Sargent 1998; Motteet al. 1998; André et al. 2007). Here we show evidence forfragmentation occurring within the coherent region, which isnot necessarily linked to the filament formation process.These results support the idea that a coherent region needsto be created to form a low-mass star. It is in this coherent re-gion where the dense gas that has lost its turbulent support ac-cumulates, and then it can easily fragment, create a filament,and/or undergo gravitational collapse to form a star. We ex-pect that future EVLA and ALMA observations of star form-ing regions will allow us to produce temperature and velocitydispersion maps which will allow us to perform a much betterjob on characterising the physical properties of filaments andof these coherent regions.We would like to thank Alvaro Hacar for discussions re-garding the properties of filaments. The Expanded Very LargeArray is operated by the National Radio Astronomy Observa-tory. The National Radio Astronomy Observatory is a facilityof the National Science Foundation, operated under coopera-tive agreement by Associated Universities, Inc. This materialis based upon work supported by the National Science Foun-dation under Grant AST-0908159 to AAG and AST-0845619 to HGA. Facilities: