Chains of dense cores in the Taurus L1495/B213 complex
AAstronomy & Astrophysics manuscript no. 24576ms c (cid:13)
ESO 2018September 13, 2018
Chains of dense cores in the Taurus L1495/B213 complex (cid:63) (cid:63)(cid:63)
M. Tafalla and A. Hacar Observatorio Astronómico Nacional (IGN), Alfonso XII 3, E-28014 Madrid, Spain e-mail: [email protected] Institute for Astrophysics, University of Vienna, Türkenschanzstrasse 17, A-1180 Vienna, Austriae-mail: [email protected]
ABSTRACT
Context.
Cloud fragmentation into dense cores is a critical step in the process of star formation. A number of recent observationsshow that it is connected to the filamentary structure of the gas, but the processes responsible for core formation remain mysterious.
Aims.
We studied the kinematics and spatial distribution of the dense gas in the L1495 / B213 filamentary region of the Taurus molec-ular cloud with the goal of understanding the mechanism of core formation.
Methods.
We mapped the densest regions of L1495 / B213 in N H + (1–0) and C O(2–1) with the IRAM 30m telescope, and comple-mented these data with archival dust-continuum observations from the Herschel Space Observatory.
Results.
The dense cores in L1495 / B213 are significantly clustered in linear chain-like groups about 0.5 pc long. The internal motionsin these chains are mostly subsonic and the velocity is continuous, indicating that turbulence dissipation in the cloud has occurred atthe scale of the chains and not at the smaller scale of the individual cores. The chains also present an approximately constant abundanceof N H + and radial intensity profiles that can be modeled with a density law that follows a softened power law. A simple analysisof the spacing between the cores using an isothermal cylinder model indicates that the cores have likely formed by gravitationalfragmentation of velocity-coherent filaments. Conclusions.
Combining our analysis of the cores with our previous study of the large-scale C O emission from the cloud, wepropose a two-step scenario of core formation in L1495 / B213. In this scenario, named “fray and fragment” , L1495 / B213 originatedfrom the supersonic collision of two flows. The collision produced a network of intertwined subsonic filaments or fibers ( fray step ).Some of these fibers accumulated enough mass to become gravitationally unstable and fragment into chains of closely-spaced cores.
Key words.
Stars: formation – ISM: abundances – ISM: kinematics and dynamics – ISM: molecules – Radio lines: ISM
1. Introduction
Star formation requires a high degree of cloud fragmentation.A typical dark cloud is tens of parsecs in size, but the coresthat undergo gravitational collapse and form stars are less than0.1 pc in diameter. Understanding how a large-scale cloud of gasfragments into a small number of dense cores remains a criticalchallenge in the field of star formation (di Francesco et al. 2007;Ward-Thompson et al. 2007; Bergin & Tafalla 2007).A clue to understanding fragmentation comes from cloudmorphology. Molecular clouds are known to present complexfilamentary distributions over multiple size scales, and a connec-tion between this filamentary structure and the process of cloudfragmentation has long been proposed (Schneider & Elmegreen1979; Larson 1985; Hartmann 2002; Myers 2009). Interest onthis connection has been boosted by the large-scale cloud imagesfrom the Herschel Space Observatory, which display a strikingprevalence of filamentary structures in the distribution of cloudmaterial (André et al. 2010; Molinari et al. 2010). These newHerschel images show that dense cores often lie along large-scale filaments like beads in a string, and leave little doubt thatsome type of filamentary fragmentation must be responsible fortheir condensation (see André et al. 2013 for a recent review). (cid:63)
Based on observations carried out with the IRAM 30m Telescope.IRAM is supported by INSU / CNRS (France), MPG (Germany), andIGN (Spain). (cid:63)(cid:63)
Herschel is an ESA space observatory with science instruments pro-vided by European-led Principal Investigator consortia and with impor-tant participation from NASA.
While filamentary fragmentation appears to produce cores,the exact manner in which this process operates is far from clear.Filamentary structures are often as large as the clouds them-selves and involve most of the cloud mass, but core and starformation have an e ffi ciency of only a few percent (Evans etal. 2009). Filaments therefore cannot completely fragment intocores, and some process must prevent most mass in a filamentto end up forming cores and stars. What limits fragmentation isstill a mystery, especially considering that many of the observedfilaments have estimated mass-per-unit-lengths that greatly ex-ceed the limit of gravitational stability (Arzoumanian et al. 2011;Hennemann et al. 2012; Palmeirim et al. 2013)In a previous study of the filamentary region L1495 / B213in Taurus (Hacar et al. 2013), we used velocity information de-rived from the C O emission to decompose what looks like asingle filament in optical and continuum images into a complexnetwork of 35 smaller filamentary structures. These structures,referred to as “fibers” to distinguish them from the large-scalefilament, present properties that di ff er significantly from thoseof the 10 pc-long L1495 / B213 region. The fibers, for exam-ple, have typical sizes around 0.5 pc, coherent velocity fields,and mass-per-unit-lengths that lie within uncertainties in the ex-pected range of gravitational equilibrium values. These fibersseem to represent a size scale intermediate between the largefilamentary cloud and the smaller dense cores, and have likelyformed by some type of fragmentation process associated withthe dissipation of turbulence (Hacar et al. 2013).When the C O data of L1495 / B213 were complementedwith N H + observations, which highlight the dense cores, the Article number, page 1 of 17 a r X i v : . [ a s t r o - ph . GA ] D ec & A proofs: manuscript no. 24576ms fibers were found to divide into two groups. Most fibers did notcontain embedded cores, and were referred to as “sterile,” buta small group of fibers contained the totality of the cores andwere classified as “fertile.” This di ff erence between sterile-fertilefibers was significant. Sterile fibers did not contain cores, but fer-tile fibers contained around three cores on average. As a result,most cores in the L1495 / B213 were found to be located in char-acteristic closely-packed linear groups.The low angular resolution observations of Hacar et al.(2013), made with the 14m FCRAO telescope, limited the studyof the closely-packed cores to only the most basic global prop-erties. To remedy this, we carried out higher-resolution obser-vations using the IRAM 30m telescope. These new observationsallow us to resolve the internal structure of the core linear groupsand to study the connection between the di ff erent cores formedfrom a single fiber. In the following sections we present the anal-ysis of the core emission with emphasis on the kinematics ofthe gas. In the last section, we present a simple scenario of coreformation that combines the large-scale analysis of Hacar et al.(2013) with the results from the new IRAM 30m data.
2. Observations
We observed selected regions of the L1495 / B213 cloud withthe IRAM 30m telescope during one session in February-March2013 and another one the following June. In both sessionswe used the EMIR heterodyne receiver (Carter et al. 2012) infrequency-switching mode together with the VESPA autocorre-lator.The observations consisted of simultaneous on-the-fly mapsin the lines of N H + (1–0) (93.17 GHz) and C O(2–1) (219.56GHz) in dual polarization mode. The maps covered the regionsidentified by Hacar et al. (2013) as bright in N H + (1–0) andtherefore indicative of dense core formation. To resolve the linesin velocity, the VESPA autocorrelator was set to a frequency res-olution of 20 kHz, which corresponds to 0.063 km s − at the fre-quency of N H + (1–0) and to 0.027 km s − at the frequency ofC O(2–1)The data were calibrated by observing a combination of am-bient and cold loads plus the blank sky every 10 minutes approx-imately. The resulting T ∗ A scale was converted into main beambrightness temperature T mb using the facility-recommendedmain beam e ffi ciencies of 0.8 and 0.6 for N H + and C O, re-spectively. All intensities in this paper are reported in T mb scaleand have an estimated uncertainty of 10-15%.Additional o ff -line data processing was carried out using theGILDAS program CLASS , and included convolving the datawith a Gaussian kernel to resample the observations into a reg-ular spatial grid (Mangum et al. 2007), folding the spectra tocorrect for frequency switching, and subtracting a polynomialto flatten the baseline. In some steps of the analysis, the data,which have an original angular resolution of 26 (cid:48)(cid:48) and 12 (cid:48)(cid:48) forN H + and C O, were further convolved with a Gaussian of fullwith half maximum (FWHM) of 20 (cid:48)(cid:48) to enhance the sensitivity.These convolved data have an angular resolution of 33 (cid:48)(cid:48) (N H + )and 23 (cid:48)(cid:48) (C O).To complement the IRAM 30m observations, we usedarchival data from the Herschel Space Observatory (Pilbratt etal. 2010). These data consisted of dust continuum maps of theL1495 / B213 region observed with the SPIRE instrument at 250,350, and 500 µ m (Gri ffi n et al. 2010). They were obtained as partof the Herschel Gould Belt Survey (HGBS, André et al. 2010), Fig. 1.
Large-scale view of the L1495 / B213 complex. The grey scaleshows the Herschel-SPIRE 250 µ m emission mapped by Palmeirimet al. (2013), and the red contours represent the N H + (1–0) emissionmapped by Hacar et al. (2013) with the FCRAO telescope. The blackdashed lines show the limits of the FCRAO observations, and the bluesolid lines enclose the regions newly mapped with the IRAM 30m tele-scope. and have been previously presented by Palmeirim et al. (2013)and Kirk et al. (2013). The data used here (OBSID 1342202254)were obtained through the Herschel Science Archive and corre-spond to level 2.5 as reduced with version 9.1.0 of the StandardProduct Generation (SPG) software.
3. Results
Figure 1 presents a large-scale view of the L1495 / B213 complex.The grey background represents the Herschel-SPIRE 250 µ mdust continuum emission mapped by Palmeirim et al. (2013),which traces the distribution of material in the region. As can beseen, this emission delineates a ≈ / B213 region isnot the only large-scale filamentary structure in the Taurus darkcloud. Images of Taurus using both molecules and dust revealthat most of the material in the cloud is distributed in a net-work of crisscrossing filaments of di ff erent sizes and orientations(Barnard 1907; Dobashi et al. 2005; Lombardi et al. 2010; Gold-smith et al. 2008; Kirk et al. 2013). In this sense, the L1495 / B213region is a prominent but still representative part of the Taurusmolecular cloud.Fig. 1 also shows the N H + (1–0) emission mapped by Hacaret al. (2013) using the FCRAO telescope with an angular reso-lution of about 60 (cid:48)(cid:48) (red contours). This N H + emission tracesthe denser, chemically-evolved gas that has condensed out of themore di ff use material in the cloud. This gas occupies only a smallfraction of the total cloud area and forms linear structures withtypical length of ≈ . Article number, page 2 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex
Fig. 2.
Integrated intensity maps of the N H + (1–0) emission showing how the dense cores in L1495 / B213 form linear chain-like structures (IRAM30m data). All maps have the same linear and intensity scales, and the first contour and contour interval are 0.5 K km s − . The star symbols indicatethe position of the YSOs in the compilation of Luhman et al. (2010), with solid symbols representing Class I objects. The central positions, in( α (J2000), δ (J2000)) coordinates are (04:19:42.5, + + + + + scale direction of the cloud. The higher resolution IRAM 30mobservations presented here cover the regions indicated by bluelines in the figure. H + maps Figure 2 shows the new IRAM 30m maps of N H + (1–0) in-tegrated intensity toward the regions with significant emission,each labeled with a Barnard cloud name following the conven-tion of Hacar et al. (2013). The figure also shows the locationof the young stellar objects (YSOs) identified by the Spitzer sur-vey of Luhman et al. (2010) (see Rebull et al. 2010 for a similarcompilation).The new IRAM 30m maps agree with the lower resolutionFCRAO maps of Hacar et al. (2013), and with maps of some ofthe individual regions previously presented by Lee et al. (2001)and Tatematsu et al. (2004). These new maps provide a sharper and more sensitive view of the dense cores, and highlight thetendency of the cores to lie in elongated structures, which wewill refer to as “chains” due to their linear appearance and thepresence of multiple peaks.Many N H + peaks in Fig. 2 correspond to dense cores in thesense used by previous studies of NH or N H + emission, likethose of Benson & Myers (1989) and Caselli et al. (2002). Thisis the case of the peaks in the B218 chain and those in the north-ern part of B213. These N H + peaks have a strong degree ofcentral concentration and typical sizes of 0.05-0.1 pc, implyingthat the emitting gas is self-gravitating and likely evolving to-ward star formation. Indeed, several of these peaks contain em-bedded Class 0 or Class I YSOs indicating that star formationhas already taken place.Not all the N H + peaks in Fig. 2 however have the dis-tinct central concentration that we normally associate with densecores. Most peaks in the B10 chain, for example, are true emis- Article number, page 3 of 17 & A proofs: manuscript no. 24576ms sion maxima, but they barely stand out above the extended fil-amentary emission that surrounds them. This lack of contrastand central concentration makes the nature of these peaks un-clear. They seem to represent pockets of dense gas that arestill connected to their surroundings, and that therefore havenot yet evolved into gravitationally decoupled objects similar tothe standard dense cores. If so, these N H + peaks must corre-spond to an evolutionary stage earlier than the dense core phase,but that is more evolved than the extended material traced withC O by Hacar et al. (2013) since the gas is chemically evolved.The variety of peak contrasts in Fig. 2 therefore implies that theN H + data cover an almost complete sequence of core evolu-tionary stages that goes from the barely-discernible peak near ∆ δ = (cid:48)(cid:48) in B10 to the highly compact peak near the origin ofthe B213 map, which is associated with the well-known outflowsource IRAS 04166 + H + peaks islikely a consequence of the continuous transition between theambient and core regimes, and implies that distinguishing be-tween “true” dense cores and precursors of dense cores is animpossible (or arbitrary) task. For this reason, here we will notattempt to distinguish between N H + peaks that we consider tobe “cores” and those that we consider as not having reached thecore phase yet, and we will treat them equally in our analysis.The lack of a clear distinction between cores and pre-cores, how-ever, seems not enough reason to stop using the term “core.” It isjust a reminder that when studying core formation, some inter-mediate structures will unavoidably end up having an ambiguousnature. Hacar et al. (2013) used the mean surface density of com-panions (MSDC) to quantify the clustering of dense coresin L1495 / B213. The MSDC measures the average number ofneighbors per unit area that one object has as a function of angu-lar separation. It has often been used, together with its equiv-alent the two-point correlation function, to determine the de-gree of clustering of stars in Taurus (Gomez et al. 1993; Larson1995; Simon 1997; Hartmann 2002). Studying the distribution oflow-density condensations identified from from NIR extinctionmeasurements, Schmalzl et al. (2010) found an excess of powerin the MSDC at small angular separations. Hacar et al. (2013)found also an excess of nearby companions, but in a di ff erenttype of structures: the dense cores derived from N H + observa-tions.The new IRAM 30m data provide an improved descriptionof the N H + emission in L1495 / B213, especially at small an-gular scales. Thus, as a first step in our analysis, we have re-evaluated the MSDC of dense cores in the cloud. To do this, wehave determined the location of all the N H + emission peaks inthe maps of Fig. 2. We have counted 22 distinct peaks, whichare three more than the 19 found by Hacar et al. 2013 becausethe new data identify additional peaks in the B10 region. Usingthe position of these N H + peaks, we have calculated the MSDCfollowing the procedure described in Hacar et al. (2013). To ex-tend the MSDC to the largest angular separations, we have usedthe large-scale information from the FCRAO data, since the newthe new 30m observations do not recover new peaks in regionswith no FCRAO detection, but only separate better the regionsalready known to have bright N H + emission.The new MSDC determined from the IRAM 30m data isshown in Figure 3 with red squares. This MSDC has a finer Fig. 3.
Mean surface density of companions as a function of angularseparation for the 22 cores identified in L1495 / B213. The red squaresrepresent observed data. The black line and the blue-shaded region rep-resent the mean and the ± rms interval from 100 Monte Carlo simu-lations of a random distribution of 22 cores. The departure of the redsquares from the shaded region at small angles indicates a significantlevel of clustering at scales less than 700 (cid:48)(cid:48) . sampling than that determined with the FCRAO data due to thehigher sensitivity and resolution of the new observations, butapart from that, the two MSDCs are consistent with each other.Also shown in Fig. 3 is the expected MSDC for a uniformly ran-dom distribution of cores. This distribution was determined us-ing a set of 100 Monte Carlo simulations in which 22 dense coreswere assigned random coordinates inside a rectangular region ofdimensions approximately equal to those of L1495 / B213. Themean value of this model MSDC is indicated by the black line,and its rms interval is contained in the blue-shaded region.As Fig. 3 shows, the MSDC for the cores in L1495 / B213has a significant excess over the random distribution for angularseparations smaller than about 700 (cid:48)(cid:48) (0.5 pc for a Taurus distanceof 140 pc, Elias 1978). This excess means that for a given core,the probability of having a neighbor closer than about 0.5 pc ishigher than it would be if the cores were distributed randomlyover the cloud. The excess confirms the visual impression fromFig. 1 that the N H + peaks in L1495 / B213 tend to cluster insmall chains, and that the chains have a typical length of 0.5 pc.The chains, therefore, are true physical structures, and not merechance groupings of dense cores in the cloud.The clustering of dense cores into chains implies that coreformation in L1495 / B213 is a highly correlated process. A num-ber of authors have previously emphasized that core formationresults from the fragmentation of filamentary clouds (Schnei-der & Elmegreen 1979; Larson 1985; Hartmann 2002; Myers2009; André et al. 2010; Molinari et al. 2010). Our observationsof L1495 / B213 go further than that by showing that fragmen-tation does not occur equally distributed along the length of afilament, but that it favors special locations where multiple coresare formed in chains. The reason for this selectivity is likely as-sociated with the multiplicity of filaments found by Hacar et al.(2013), who argued that the L1495 / B213 large-scale filament isin fact a collection of 35 intertwined velocity-coherent filaments.If most of these filaments are “sterile” and do not form cores,while a small minority are “fertile” and form multiple cores, aclustering of the cores into a few elongated chains is naturallyexpected. In Sect. 4 we will discuss in more detail the implica-tions of this result to our understanding of core and star forma-tion.
Article number, page 4 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex
Fig. 4.
Correlation between observed N H + (1–0) integrated intensity and H column density derived from SPIRE dust-continuum data. Pointswithin 45 (cid:48)(cid:48) of an embedded YSO have been excluded to avoid e ff ects of stellar heating. The open squares in the B7 / L1495 panel indicate pointslikely a ff ected by V892 Tau (see text). The dashed lines represent the analytic expression discussed in the text. H + abundance While N H + is a tracer of choice for dense core gas due to itsresistance to freeze out, its formation is enhanced by the disap-pearance of CO from the gas phase, so it can su ff er systematicabundance variations during the evolution of a core (Caselli et al.1999; Bergin et al. 2002; Tafalla et al. 2002; Aikawa et al. 2005).To quantify possible N H + abundance variations in the chains ofL1495 / B213, we need to compare the N H + data with data froma di ff erent tracer that is insensitive to freeze out. As mentionedbefore, observations of the dust emission and absorption providean independent estimate of the gas column density, and there-fore represent the ideal counterpart to the molecular line datapresented here.The Herschel Gould Belt Survey (HGBS) used the HerschelSpace Observatory to produce very high quality images of thedust continuum emission from the Taurus molecular cloud (An-dré et al. 2010; Palmeirim et al. 2013; Kirk et al. 2013). Thesepublicly-available images are an excellent counterpart to theN H + line data, since they have high angular resolution, covermultiple wavelengths, and trace optically thin emission. As afist step, we compared the dust and the N H + data by super-posing the N H + (1–0) maps of Fig. 2 with the 250, 350, and500 µ m maps made by the HGBS team with the SPIRE instru-ment. These SPIRE maps cover the longest wavelengths observ-able with Herschel, and provide the highest sensitivity to colddust emission.The SPIRE-N H + comparison showed that in places withsignificant N H + (1–0) emission ( ≥ . − ), the dust con-tinuum flux and the N H + (1–0) integrated intensity have simi-lar spatial distributions. This implies that in N H + -bright places,most of the dust continuum emission arises from the componentemitting N H + (in N H + -weak places, the dust emission tracesthe extended cloud). As a result, we can use the dust continuumemission from the chains to estimate an associated H columndensity, and with it, an N H + abundance. To carry out this es-timate, we first convolved the 250 µ m map to match the 35 (cid:48)(cid:48) . µ m map, which is also similar to the 33 (cid:48)(cid:48) resolution of the convolved N H + data (Sect. 2). We thenfollowed standard practice and assumed that the dust emitted asan optically thin grey body with an emissivity that varies withfrequency as ν (Hildebrand 1983), and used the fluxes at thetwo wavelengths to derive a dust temperature and an H columndensity for each position.Our choice of the 500 µ m dust opacity was κ µ m = .
03 cm g − , based on the matching between our SPIRE-derivedH column densities with the extinction-derived column densi-ties of Schmalzl et al. (2010), which were kindly provided byMarkus Schmalzl. This choice is only 20% lower than the valueassumed by the HGBS team (André et al. 2010; Könyves et al.2010; Arzoumanian et al. 2011; Palmeirim et al. 2013; Kirk etal. 2013), and lies within the range of values used or derived byother authors from Herschel data (Henning et al. 2010; Juvelaet al. 2011; Launhardt et al. 2013; Suutarinen et al. 2013). Still,it should be noted that the dust opacity has an uncertainty ofat least 50%, and that it could su ff er variations with density, asshown by the detailed analysis of Juvela et al. (2011), Suutari-nen et al. (2013), and Ysard et al. (2013). The uncertainty in thedust opacity represents the largest source of uncertainty in ourH column density estimate.Fig. 4 compares the SPIRE-derived H column density withthe N H + (1–0) integrated intensity in all the regions with N H + emission. The B213 data have been split into two panels to dis-tinguish the region two main components (labeled SE and NW),and the southern part of B7 / L1495 has been treated separatelyfrom the north one (and labeled with open squares), due to itsanomalous N H + abundance further discussed below. As the fig-ure shows, the N H + (1–0) integrated intensity and the H col-umn density appear to be correlated in all regions. An estimateof the Pearson’s r coe ffi cient confirms this impression and re-turns values that range from 0.81 in B213-NW (lowest) to 0.94in B218 (highest), all indicative of a significant degree of corre-lation.While significant, the correlation in the panels of Fig. 4presents a non-negligible level of scatter. Part of it appears to Article number, page 5 of 17 & A proofs: manuscript no. 24576ms arise from the contribution of gas with either low density or lowN H + abundance (or likely both). This gas does not contributeto the N H + intensity, but increases the H column density andshifts some of the points along the x-axis, broadening the distri-bution in the plots. An extreme example of this e ff ect can be seenin the B213-NW panel, where in addition to the main diagonalband of points there is a secondary band that has weaker N H + emission and is shifted horizontally by about 10 cm − . Thisband is associated with a small condensation near ∆ α = − (cid:48)(cid:48) seen in Fig.2. A less extreme example occurs in B218, where thebroad diagonal band of points is in fact the superposition of twoslightly shifted and narrower bands, each one due to one of thebright cores in the chain. In contrast, the B211 region presentsonly a single dense core in the N H + maps, and its distributionof points presents the narrowest correlation of the sample.If the scatter in the panels of Fig. 4 results from additionalgas components along the line of sight, the slope is an indica-tor of the N H + abundance in the dense gas. This is expectedbecause the N H + integrated intensity represented in the y-axisis proportional to the N H + column density (assuming that theemission is optically thin, see below), and as a result, the slopeof the correlation equals the ratio between the N H + and H col-umn densities, which is an estimate of the N H + abundance. Adetailed radiative transfer model presented in the next section tofit the radial profiles of N H + emission confirms this interpreta-tion, and shows that there is an almost linear relation betweenthe N H + integrated intensity and the H column density. This isillustrated in Fig. 4 with a series of black lines that represent thepredictions from the radiative transfer model assuming that allchains have the same N H + abundance of 5 × − with the ex-ception of B7 / L1495-south, where the abundance is 1 . × − .To fit the data in Fig. 4, we added small horizontal o ff sets of7 × cm − , 6 × cm − , and 3 × cm − to the fits of B211,B7 / L1495-north, and B10, respectively. The o ff set in B211 isin fact expected, since this region contains two additional C Ocomponents (numbers 9 and 12 in the decomposition of Hacar etal. 2013) that do not emit in N H + and clearly contribute to theH column density (see Fig. 7 below). The o ff sets in B7 / L1495and B10 are also likely related to the presence of lower-densitygas toward these two regions.While uncertain, the N H + abundance in B7 / L1495-south ismuch lower than in other chains (by a factor of 3), and is theonly one that deviates from a pattern of almost constant abun-dance. To investigate its origin, we have inspected the SPIRE im-ages of B7 / L1495 at di ff erent wavelengths. These images showthat B7 / L1495-south lies inside a region of bright and extendedFIR emission in the vicinity of V892 Tau, a Herbig Ae / Be starfirst identified by Elias (1978) and with a total luminosity of ≈
400 L (cid:12) (Sandell et al. 2011; Mooley et al. 2013) that lies about300 (cid:48)(cid:48) (0.2 pc) in projection from B7 / L1495-south. A grey bodyanalysis of the SPIRE emission, indicates that the dust temper-ature in the region is elevated, and that it gradually increasestoward V892 Tau, where it reaches about 15 K, or 50% higherthan in B7 / L1495-north. In addition, the C O maps of Fig. 7(discussed below) show that B7 / L1495-south coincides with aregion of bright C O emission. This is in contrast with the otherN H + -bright regions, which coincide with weak C O emissiondue to CO freeze out. Thus, it appears that molecular deple-tion, and its resulting N H + enhancement, are anomalously lowin the dense gas of B7 / L1495-south. The low N H + abundancein B7 / L1495-south seems therefore a result from the action ofV892 Tau. Whether this is a consequence of simple dust heat-ing or of a di ff erent energetic process requires a more detailedinvestigation. Fig. 6.
Histogram of the N H + (1–0) mean optical depth for all chainpositions used in the radiative transfer analysis. In the maps, the core chains appear irregular in shape and di ff er-ent from each other. A closer inspection of the emission, how-ever, shows that they have a similar internal structure. This canbe seen in the radial profiles of N H + emission presented inFig. 5. These profiles were created by following the emissionof each chain in the map with a cursor and defining the line ofrelative maxima as the axis of the chain. Using this axis, the ra-dial distance of each observed position was calculated, and theintensity of the emission was plotted as a function of it.As figure Fig. 5 shows, the emission from each chain followsa radial profile that consists of a flat inner region and a power-lawtail, similar to that often found in filamentary clouds (Arzouma-nian et al. 2011; Hacar & Tafalla 2011; Palmeirim et al. 2013).The B213 and B218 chains present a larger level of dispersionnear the axis because they contain bright cores separated by re-gions of weak emission, so points with the same axial radius canhave a large range of intensities. The B10 and B7 / L1495-northchains, on the other hand, present a less clumpy and more pris-tine appearance, and their radial profiles have a lower dispersionnear the axis (as in Fig. 4, the emission from B7 / L1495 has beenseparated into north and south components).The combination of clumpiness due to the embedded coresand comparable radial profiles implies that the chains startedtheir evolution with a similar density structure, and that laterevents added di ff erent fragmentation patterns to each one. Ourgoal in this section is to determine this common underlyingstructure, since it represents the initial conditions of core for-mation. For this, we have modeled the N H + radial profiles as-suming that the chains are cylindrically symmetric, and that theyhave a density profile of the form n ( r ) = n ◦ + ( r / r ) α , (1)where n ◦ is the central density, r the half-density radius, and α the asymptotic power index. This type of radial profile haspreviously been used to fit the density structure of both starlesscores (Tafalla et al. 2002) and filamentary clouds (Arzoumanianet al. 2011; Hacar & Tafalla 2011; Palmeirim et al. 2013; Ysardet al. 2013).To compare the model with the observations, we have solvedthe equations of radiative transfer and predicted the radial profileof N H + intensity. Lacking a 2D model for the radiation trans-fer in a cylinder, we have instead used a spherically symmetric Article number, page 6 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex
Fig. 5.
Radial profiles of N H + (1–0) integrated intensity in log-log scale. The red squares represent the data, and the black lines are our models.B7 / L1495 north data and model have been shifted by a factor of two to ease visibility. model that has the same density radial profile. This model prop-erly accounts for the radial drop of collisional excitation causedby the density law, although it likely underestimates the excita-tion due to photon trapping because photons escape more easilyfrom a sphere that from a cylinder. The e ff ect of this di ff erence inthe trapping, however, is likely to be very small, since the opti-cal depth of the N H + (1–0) emission is low. This is illustrated inFig. 6 with a histogram of the mean N H + (1–0) optical depth es-timated from the data of all bright positions in our survey (4,947spectra in total). The mean optical depth was defined by divid-ing the total optical depth of the N H + (1–0) transition (deter-mined with the HFS hyperfine analysis in the CLASS program)between seven, which is the number of components. It thereforerepresents the average optical depth of an individual N H + (1–0)component.As can be seen, the histogram of mean optical depths is dom-inated by low values. In all regions but B7 / L1495, ∼
90% of thepoints have a mean optical depth lower than 1, and less than1% of the points have a mean optical depth larger than 2. Thecluster-forming B7 / L1495 region is significantly more opaqueand is responsible for the small group of points between opti-cal depths 1 and 2 in the histogram. Still, 56% of its positionshave a mean optical depth lower than 1, and no point exceedsa value of 2. Under these conditions, using spherical geometryto simulate the radiative excitation in a cylinder appears to bean acceptable approximation, especially considering that the as-sumption of cylindrical symmetry is itself a large simplificationof the true geometry of the chains.To solve the radiative transfer equations we used the MonteCarlo code of Bernes (1979) previously applied to analyze theemission from starless cores (Tafalla et al. 2002, 2004a). Thiscode was implemented with the molecular parameters of N H + from the LAMDA web site (Schöier et al. 2005), which includethe collision rates of Daniel et al. (2005). These rates include theindividual hyperfine components of each rotational transition (upto J = H + inthe dense gas and whose emission in Taurus cores indicates amedian temperature of 9.5 K (which little dispersion, Jijina et al.1999). Recent large-scale ammonia mapping of the L1495 / B213complex by Seo et al. (in preparation) confirms this assumption,and indicates that while there are small local variations of 1-2 Kin some cores, there are no global temperature gradients in thechains. Even the lower-density gas that surrounds the core chainsappears to have a similar temperature, since the CO-based esti-mate of Goldsmith et al. (2008) indicates that the majority ofpoints in this region (their Mask 2) have temperatures that liein the 6-12 K range. This constant temperature of the gas in thedensity range of interest (10 -10 cm − , see below) is expectedfrom detailed modeling of the gas heating and cooling, and con-trasts with the well-measured temperature gradient of the dustcomponent found by Palmeirim et al. (2013), which is expectedfrom heating by the interstellar radiation field (Evans et al. 2001;Galli et al. 2002).Additional assumptions of the model were a non-thermalFWHM linewidth of 0.25 km s − , as suggested by the analy-sis of Sect. 3.7.2, and a maximum radius of 400 arcsec (0.27 pc),although the comparison with the data is restricted to the cen-tral 100 arcsec due to limited signal to noise (the exact size hasonly a small e ff ect on the result). Following our experience withthe analysis of dense cores, we divided the cloud model into 200shells, used 2,000 photons, and iterated the calculation 40 times.To simulate the IRAM 30m observations, plus the additional 20 (cid:48)(cid:48) Gaussian smoothing applied to the data to enhance its signal tonoise, the emerging intensity distribution was convolved with aGaussian of 33 (cid:48)(cid:48)
FWHM. Finally, the data were scaled up bya factor of 1.4 to simulate a 45 degree inclination angle of themodel with respect to the line of sight. The L1495 / B213 cloudappears as a relatively long filament in the sky ( ≈
10 pc), so itis unlikely to be highly inclined; using a moderate angle of 45seems like a reasonable assumption that is unlikely to introducea large error.To find the best fit to the data, we explored di ff erent valuesof the N H + abundance and the density law. As discussed in theprevious section, the choice of N H + abundance has a direct ef- Article number, page 7 of 17 & A proofs: manuscript no. 24576ms
Table 1.
Best-fit chain parameters.
Chain n r X(N H + ) M / L (cm − ) ( (cid:48)(cid:48) ) ( M ◦ pc − )B213-SE 6 ×
50 5 × − ×
50 5 × − ×
35 5 × − ×
50 5 × − / L1495-N 7 ×
50 5 × − / L1495-S 7 ×
50 1 . × − ×
40 5 × − ×
50 5 × − H + intensity andthe H column density, and the plots of Fig. 4 were used to derivea constant value of 5 × − for all chains but B7 / L1495-south(were the best fit is 1 . × − ). To fit the density law, we usedthe radial profiles of N H + (1–0) intensity and explored the e ff ectof each of the three free parameters. The power law index is onlyweakly constrained, since the radial profile only approaches thisasymptotic behavior at large radius, where the data have a lowsignal to noise ratio. Reasonable fits were achieved with valuesclose to 3 (as found for the filaments in L1517, Hacar & Tafalla2011), so this parameter was fixed to 3 in all the chains. Theremaining two parameters, central density and half-maximumradius, are somewhat correlated, since both contribute linearlyto the central column density, and they need to be distinguishedby fitting the profiles at large radii (with the already mentionedproblem of low signal to noise and certain dependence on thepower-law index). After exploring a number of combinations, wedetermined as best fit values those given in Table 1, which pro-duce the radial profiles shown with black lines in Fig. 5. Sincewe were interested in the density structure of the chains as pos-sible indicator of the initial conditions of core formation, the fitswere purposely chosen to fit the points with lowest intensity nearthe axis and to avoid the brighter points that arise from the densecores.As can be seen in Table 1, both the central density andthe half-maximum radius vary little over the sample of chains(6 − × cm − and 35 (cid:48)(cid:48) − (cid:48)(cid:48) , respectively). This small vari-ation agrees with our expectation of a common internal den-sity structure based on the similarity of the radial profiles, andstrengthens the idea that the di ff erent chains may have formedin a similar manner. The fit values, however, have a significantlevel of uncertainty due to the uncertainty in the the dust opacitydiscussed before. Also, the large scatter in the radial profiles isa remainder that cylindrical symmetry is an over-simplificationof the 3D geometry of the chains. For this reason, the best-fitparameters in Table 1 should be considered only as a first-orderapproximation to the true parameters of the chain gas, whichlikely have an uncertainty level of a factor of 2.Even if approximate, the parameters of Table 1 can be usedto explore the physical state and gravitational stability of the corechains. To do this, we compare our best fit models with the clas-sical solution of an isothermal cylinder in equilibrium, first stud-ied by Stodólkiewicz (1963) and Ostriker (1964). This solutionhas an asymptotic power-law index of -4, while our best fit mod-els are slightly flatter and have a power-law index of -3. Moreimportantly, the isothermal cylinder has an equilibrium mass perunit length of 16.6 M (cid:12) pc − , assuming a gas kinetic temperatureof 10 K. Table 1 shows that the mass per unit length values ofour best-fit models are systematically larger, although only by at most a factor of 2. Palmeirim et al. (2013) also found a larger-than-equilibrium mass per unit length in the B213 / B211 filamentas a whole using dust continuum measurements, although theseauthors treated the region as a single object and ignored the pres-ence of multiple velocity components. Whether this larger massper unit length means that the chains are significantly out of equi-librium is unclear, especially considering the uncertainty in thedust opacity and that additional support mechanisms, such asmagnetic fields or temperature gradients can increase the equi-librium mass per unit length (Stodólkiewicz 1963; Nakamura etal. 1993; Recchi et al. 2013). Further understanding of the physi-cal state of the chains requires the analysis of their internal kine-matics, which is the topic of the next two sections. O data: multiple components
The C O molecule freezes out rapidly onto the dust grains atdensities typical of the cores and the chains, so it is a poor tracerof the dense gas kinematics. It is however a faithful tracer of themotions in the lower-density gas that surrounds the chains, sincein this regime C O is chemically stable, easily thermalized, anddoes not su ff er appreciably from saturation due to its low abun-dance. Hacar et al. (2013) showed that in the L1495 / B213 re-gion, the velocity fields of C O and N H + are similar, indicat-ing that the dense cores and their surrounding environment areclosely coupled kinematically. Before studying the kinematics ofthe dense gas with N H + in the next section, it is therefore con-venient to use the C O emission to determine the properties ofthe velocity field in the vicinity of the chains. These propertiesprovide important context and help solve some of the ambigui-ties that a ff ect the more selective N H + emission.As analyzed in detail by Hacar et al. (2013), the C O ve-locity field in L1495 / B213 is complex. It consists of about 35intertwined filamentary components (or fibers) that appear in thespectra as multiple velocity peaks. To disentangle these compo-nents, Hacar et al. (2013) used a combination of Gaussian fitsto the spectra and the Friends In VElocity (FIVE) algorithm,which connects spatially the velocity components from nearbypositions. The IRAM 30m maps discussed here are less extendedthan the FCRAO maps of Hacar et al. (2013), and the focus ofour analysis is limited to the C O emission related to the densegas in the chains. For this reason, we have carried out a simpli-fied analysis of the velocity structure of the C O emission basedon the inspection of the spectra and the use of velocity-integratedmaps.An inspection of the C O data reveals that each mapped re-gion contains at least several positions where the spectrum hastwo peaks separated by more than one full linewidth. Thesedouble-peaked spectra do not originate from self-absorption,since, when detected, the optically thin isolated component ofN H + (1–0) matches the velocity of one of the two C O com-ponents, instead of appearing at the intermediate velocity thatwould be expected in the case of self-absorption. The doublepeaks therefore arise from the multiple velocity componentsstudied by Hacar et al. (2013) when they overlap along somelines of sight. Examples of these double-peaked spectra can beseen in the top panel of Fig. 7 for each of the five regions asso-ciated with dense gas.To determine the spatial distribution of the C O velocitycomponents in the vicinity of the core chains, we have dividedthe emission into two velocity intervals centered approximatelyon each of the C O peaks. The resulting maps, presented in thebottom panels of Figure 7, show that in each chain, the two C Ocomponents di ff er markedly in spatial distribution. In B213, the Article number, page 8 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex
Fig. 7.
Velocity structure of the C O(2–1) emission in L1495 / B213.
Top: C O(2–1) spectra from selected positions illustrating the presence ofmultiple velocity components along the line of sight. The selected positions are indicated with red circles in the maps below.
Bottom:
Maps ofC O(2–1) emission integrated in two velocity intervals that approximately coincide with the components in the top spectra. All maps have thesame physical scale, color code (shown in the B7 / L1495 panel), and contour scale (first contour and interval are 0.3 K km − ). Coordinate centersand star symbols as in Fig. 2. Article number, page 9 of 17 & A proofs: manuscript no. 24576ms blue component extends to the NW of the mapped region and isunrelated to the chain of N H + cores, that has a di ff erent velocityand spatial distribution. This blue C O component correspondsto component number 18 in the cloud decomposition of Hacar etal. (2013). The red component, on the other hand, is associatedwith the chain of dense cores, and its velocity and large-scale ori-entation are similar to those of the chain. In contrast with N H + ,the C O emission presents strong evidence of large-scale freezeout. It misses some of the brightest N H + and continuum peaks,such as the core around IRAS 04166 + O freeze out in this region seems thereforenot limited to the dense cores, but occurs at the scale of the fullchain, and is only reversed locally in the vicinity of some YSOs.In the single-core region B211, also the red component is as-sociated with the N H + dense core, since it has the same velocityand a similar spatial distribution. The unrelated blue componentarises from a long diagonal filament that is in fact the superposi-tion of the parallel components 9 and 12 in the velocity decom-position of Hacar et al. (2013).The more complex B10 region appears in the N H + mapsof Fig. 2 as consisting of two roughly parallel chains plus anisolated core in the south. The C O spectrum and maps in Fig. 7show now that the western chain of B10 coincides with a regionwhere two velocity components that are separated by 1 km s − spatially overlap. These two components seem to be responsiblefor the apparent velocity jump of about 1 km s − seen in N H + (Sect. 3.7.1), indicating that the western chain of B10 n is infact the overlap of two separate structures. This interpretationis in agreement with the decomposition of the large-scale C Oemission by Hacar et al. (2013), who divided this region into twocomponents labeled 6 and 8.The C O maps of Fig. 7 also show that chain-wide COfreeze out has also taken place in B10, since the eastern chain isvery prominent in N H + but only marginally visible in C O. Asmentioned before, this eastern chain shows little fragmentation.This indicates that CO depletion precedes the fragmentation ofthe chain into cores.In the B7 / L1495 region, the blue C O velocity component isassociated with the isolated N H + core to the SW, while the redcomponent is associated with the N H + chain of cores. As canbe seen in Fig. 7, the blue C O brightens significantly towardthe south end of the map, which coincides with the region wherethe dust temperature increases due to heating by V892 Tau. Thisbrightening of the C O emission implies that CO depletion maybe lower closer to V892 Tau, and this may explain the anoma-lously low N H + abundance inferred from the comparison withthe SPIRE data.Finally, in B218, the two C O components present anticor-related spatial distributions. The blue component peaks towardthe NE and SW of IRAS 04248 + H + starless cores at each side of the YSO becausethey match both in position and velocity. The red C O com-ponent, on the other hand, peaks toward the IRAS source andextends slightly toward the NW. The nature of this componentis unclear due to its limited extent. A likely possibility is that itis caused by the action of the YSO on its surrounding gas, sinceIRAS 04248 + O emission (Gomez et al. 1997).In this interpretation, the B218 region would consists of a sin-gle C O velocity component that corresponds to the blue C O regime. Further observations of this region are needed to clarifyits kinematics.To summarize, the C O data show that the presence of mul-tiple velocity components in the vicinity of the chains is com-mon. These components are separated by supersonic speeds anddo not seem to be interacting, since they are forming densecores at their own systemic velocity, and not at the intermedi-ate velocity that would be expected if core formation occurredthrough collisions. The components in each region, however, arenot completely unrelated. In B10 and B7 / L1495, for example,the two components have produced cores in close proximity,which given the strong clustering of cores in the cloud seemsan unlikely random event. In other regions, like B213 and B211,di ff erent filamentary components are almost parallel, suggestingthat they have some type of relation or common origin. Fromtheir large-scale study of the C O emission, Hacar et al. (2013)found that indeed, most velocity components in L1495 / B213 be-long to groups or bundles with a common origin, and proposedthat some type of turbulent fragmentation process was responsi-ble for their origin. A number of recent hydrodynamical simu-lations have shown that bundles of filamentary components likethose in L1495 / B213 arise naturally from the combination of tur-bulent motions and self gravity (Kritsuk et al. 2013; Smith et al.2013; Moeckel & Burkert 2014; Myers et al. 2014). This im-plies that the formation of dense cores and chains is precededby a step of fragmentation whose product are the C O compo-nents shown in Fig. 7. Thus, the multiplicity of components inthe C O spectra near the core chains is not a mere superposi-tion coincidence, but a natural consequence of the hierarchicalfragmentation required to form dense cores. Further discussionon this topic is presented below after the analysis of the N H + kinematics. H + data The velocity structure of the N H + emission is simpler than thatof C O due to the more selective nature of this tracer. In gen-eral, the N H + (1–0) spectra present a single velocity component,although split into seven features due to hyperfine structure. Afew spectra show hints of two velocity components, like nearB213 (900 (cid:48)(cid:48) , − (cid:48)(cid:48) ), but the components are so weak that isnot possible to analyze them using multiple fits. For this reason,we have fitted the N H + (1–0) spectra assuming a single velocitycomponent, using for this the CLASS program and the numeri-cal parameters of the hyperfine structure derived by Caselli et al.(1995). This single-component fit analysis determines both theline center velocity and the full width at half maximum (FWHM)corrected for optical depth broadening. Subtracting the thermalcontribution of a gas at 10 K, the FWHM can be converted intoan estimate of the non-thermal velocity dispersion in the gas.Figure 8 shows in color the distribution of N H + line cen-ter velocity and non-thermal FWHM as derived from the hyper-fine analysis. To ensure the quality of the data, the figure onlypresents results from fits that appear reliable under visual inspec-tion, which approximately corresponds to an intensity thresh-old of 0.5 K km s − . As can be seen, the line center velocitychanges smoothly over each chain, with a typical size scale forthe changes of the order of a core diameter. The accompany-ing linewidth maps also show a smooth behavior, although thereare several regions of high dispersion that we discuss in moredetail below. Since the sound speed linewidth corresponds to0.45 km s − , the maps in Fig. 8 indicate that the gas in the chainsis mostly subsonic, and that only a few locations have supersoniclinewidths. Article number, page 10 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex
Fig. 8.
Velocity structure of the N H + emission as determined from hyperfine fits to the spectra. For each region, the first panel shows (in km s − )the spatial distribution of the velocity centroid, and the second panel shows (also in km s − ) the distribution of non-thermal linewidth (FWHM).The black contours show the distribution of integrated emission to help identify the main dense gas features. All plots use the same spatial scaleand color code, which is indicated by wedges in the maps of B213. Coordinate centers, contour levels, and star symbols as in Fig. 2. While the maps in Fig. 8 provide a good representation of thegas velocity field in two dimensions, they provide limited quan-titative information on the gas kinematics. This information isbetter appreciated in the velocity profiles of Fig. 9. These profilesrepresent, as a function of distance along the axis of each chain,the line center velocity (blue), the non-thermal FWHM (red),and for reference, the N H + (1–0) integrated intensity (black).To ensure physical proximity between the points, the figure onlyshows data from positions within 30 arcsec from each chain axis.All panels use the same linear scale, and as a result, the figure is dominated by the data from the B213 region, which has a lengthalmost as large as the rest of the chains combined. We first study the line-center velocity, which is represented withblue squares in the top panels of Fig. 9. As can be seen, thisparameter presents little dispersion and an almost oscillatory be-havior in most panels. In the longest B213 chain, the line cen-ter velocity oscillates repeatedly without deviating by more than
Article number, page 11 of 17 & A proofs: manuscript no. 24576ms
Fig. 9.
Velocity structure of the N H + emission along the axis of the core chains. For each chain, the plot shows the velocity centroid in the toppanel (blue symbols), the non-thermal FWHM in the middle panel (red symbols), and the integrated intensity in the bottom panel (black symbols).Only points within 30 (cid:48)(cid:48) of the chain axis are shown to ensure proximity. The horizontal dashed line indicates the FWHM-equivalent of the soundvelocity. Note the smooth oscillations in the velocity centroid and the predominance of subsonic values in the non-thermal linewidth. about 0.3 km s − from the mean value over its full length ofalmost 1.5 pc. The smaller B10-E (eastern branch of B10) andB7 / L1495 chains also show smooth velocity oscillations, againwith close-to-constant mean values and amplitudes of the orderof 0.2 km s − .In contrast with the other chains, the western branch of B10presents a jump in velocity of about 1 km s − near D = (cid:48)(cid:48) .This jump most likely results from the presence in B10-W oftwo separate chains. As discussed in Sect. 3.6 and evidenced bythe double-peaked C O spectrum of Fig. 7, two cloud compo-nents with velocities around 6 and 7 km s − coexist and over-lap in B10-W. The blue C O component lies mostly toward theNW, and the red component lies mostly toward the SW. This isalso the distribution of the N H + line center velocities, whichalso match the velocities of the two C O components. Since theN H + center velocity remains almost constant toward each sideof the jump, and the jump coincides with a sharp drop of N H + emission (as shown in the 2D map of Fig. 8), the most naturalinterpretation of the velocity jump in B10-W is that it representsthe transition between the two C O components. This impliesthat B10-W is not a single chain of cores, but the chance align-ment of two di ff erent velocity components. The recent numericalsimulations of turbulent molecular clouds by Moeckel & Burk-ert (2014) show that this type of chance alignment is expected inregions like the L1495 / B213 cloud.Another discontinuity in the velocity profiles of Fig. 9 oc-curs in B218 near D = (cid:48)(cid:48) . As shown in 2, the B218 chainconsists of three N H + cores, two of them starless and located ateach end of the chain and a weaker one located near the centerand associated with the Class I object IRAS 04248 + − , but that the region between them, wherethe N H + emission weakens and the IRAS source lies, presentsa rapid shift in velocity towards the red. This reddening of theN H + emission coincides with the reddening of the C O emis-sion discussed in Sect. 3.6, and is highly localized toward thevicinity of the YSO (see Fig. 8). As discussed in Sect. 3.6 thisreddening of the emission likely results from the interaction ofthe YSO with the surrounding cloud, and does not represent aremnant of the pre-stellar motions in the chain.The velocity oscillations in Fig. 9 are similar to those foundin L1517, also in the Taurus complex, by Hacar & Tafalla (2011).In L1517, several filaments presented a sinusoidal velocity pat-tern that was approximately shifted by λ/ λ/ ff set betweendensity and velocity in the case of an unstable (core-forming)perturbation (Gehman et al. 1996; Hacar & Tafalla 2011). Thecore chains in L1495 / B213 provide an ideal place to search forsimilar core-forming motions, since the multiplicity of cores pro-vides a strong constraint in the displacement between the veloc-ity and column density profiles. The B213-NW chain, for exam-ple, contains 4 almost equally-spaced cores, while the filamentsin L1517 contained only two cores. For this reason, we have fit-ted the N H + intensity profiles in Fig. 9 with simple sinusoidalfunctions (after subtracting a mean value), and we have com-pared the velocity profiles with shifted versions of the intensitysinusoids. While occasional matches for individual cores can befound, no chain presents a systematic displacement between its Article number, page 12 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex velocity and intensity profiles that applies to all cores and thatcould be interpreted unambiguously as evidence of core-formingmotions along the chain axis. This lack of a systematic shift be-tween the velocity and column density patterns implies that thevelocity oscillations in the chains are not entirely due to the core-forming motions predicted by the analytic theory.There are several possible origins for the velocity oscilla-tions seen in N H + . One possibility is that they still arise fromcore-forming motions, but that the motions are more complexthan what the simple analytic model assumes. Most chains con-tain a mixture of starless cores and cores with embedded YSOs,and this indicates that the contraction history of the chain musthave been more irregular than what is assumed by the simplemodel, in which all cores are formed simultaneously throughthe exponential growth of a single sinusoidal perturbation. An-other possibility is that the oscillations arise from motions thatpre exist the formation of the chains and the cores. Hacar etal. (2013) found that most C O filamentary components in thecloud present oscillations in their velocity field irrespectively oftheir core-forming status. In this case, the chains must start theirevolution with an already perturbed velocity field, and identify-ing any core-forming motions in such conditions may require amore complex analysis. Numerical simulations of the formationand fragmentation of realistic filamentary structures are neededto clarify this issue.
The non-thermal N H + component of the velocity field is repre-sented with red squares in the middle panels of Fig. 9. Like theline center velocity, the non-thermal component presents both asmooth oscillatory behavior along each of the chains and a verylow level of dispersion, except for a few regions of moderatescatter. The great majority of points lie below the sonic thresholdof 0.45 km s − , indicating that subsonic gas motions dominatethe chains. For the data presented in the figure, which consists ofpoints with separations of less than 30 (cid:48)(cid:48) from the chain axis, thefraction of supersonic points is only 2.1%. Increasing the separa-tion threshold to 90 (cid:48)(cid:48) adds more weak points, but only increasesthe fraction of supersonic points to 3.7%. Subsonic points aretherefore the norm, and supersonic positions are rare.In addition to being rare, the supersonic points tend to lie insmall groups, implying that they result from localized causes.Feedback from outflows is one of them, as illustrated by thepoints in the vicinity of IRAS 04169 + D = (cid:48)(cid:48) . This Class I source powers a well-known bipolar outflow previously studied in CO by Moriarty-Schieven et al. (1992), Bontemps et al. (1996), and Narayananet al. (2012). As Fig. 8 shows, the points of supersonic N H + linewidth lie north and south of IRAS 04169 + H + linewidths of possibleoutflow origin are the vicinity of the three low-mass sources near D = (cid:48)(cid:48) in B213 SE, which appears to have already beenevacuated of dense gas, and the already-mentioned vicinity ofIRAS 04248 + + D = (cid:48)(cid:48) . Thisis another outflow source of low luminosity that has evacuateda cavity in the surrounding dense core (Bontemps et al. 1996;Tafalla et al. 2004b; Santiago-García et al. 2009).Not all regions with supersonic linewidth result from out-flow feedback. A group of supersonic points in B10-NW co- Fig. 10.
Structure function of the dense-gas velocity field. The color-coded solid squares show the structure function as a function of lagin B213 (black), B10 (red), B218 (green), and B7 / L1495 (blue). Theblue dashed line represents the classical relation from Larson (1981),and the red dashed line represents the recently-determined core-velocitydi ff erence of Qian et al. (2012). incide with the region of double-peaked C O spectra shownin Fig. 7, and their enhanced linewidth likely results from thismulti-component kinematics. Indeed, some N H + (1-0) spectrain this region show two velocity components, and confusion be-tween these components and the hyperfine structure seems tocause the larger linewidths seen in this region. A similar mixingbetween components is likely to occur toward the western part ofB213-SE (D ≈ (cid:48)(cid:48) in Fig. 9), where two velocity componentsat 6.7 and 7.0 km s − seem to coexist.While we cannot exclude that some of the remaining super-sonic N H + linewidths correspond to positions with intrinsicallysupersonic motions, the overwhelming dominance of subsonicpositions ( > The combination of subsonic motions and large-scale continu-ity in the velocity field of the B213 gas filaments led Hacar et al.(2013) to refer to these structures as “velocity coherent” (see alsoHacar & Tafalla 2011). This term stresses the quiescent state ofthe filaments compared to the cloud as a whole, which is charac-terized by a Kolmogoro ff -type relation between size and velocityindicative of turbulent motions (Larson 1981). To further com-pare the kinematics of the gas in the chains with that of the large-scale cloud, we have carried out a structure-function analysis ofthe N H + line-center velocities. The structure function measuresthe mean di ff erence in velocity between positions separated by adistance l , and is defined as δ v ( l ) = (cid:104) ( v ( x ) − v ( x + l )) (cid:105) , where v ( x ) is the gas velocity at an arbitrary position x and thebrackets represent a spatial average over all positions separatedby l . This function is commonly used as a descriptor of the ve-locity field in a cloud, where it is systematically found to dependon l as a power law (Elmegreen & Scalo 2004; Heyer & Brunt2004).To estimate the structure function of the gas in theL1495 / B213 chains, we have used the N H + line-center veloc-ities derived in Sect. 3.7 from hyperfine fits. These velocities Article number, page 13 of 17 & A proofs: manuscript no. 24576ms
Fig. 11.
Fray and fragment scenario of core formation. The three panels represent three di ff erent stages in the evolution of a large-scale filamentlike L1495 / B213. In the first stage (left panel), two gas flows collide and produce a filamentary density enhancement. With time (middle panel),a combination of residual turbulent motions and gravity splits the gas into a series of intertwined velocity-coherent fibers. Finally (right panel),some fibers accumulate su ffi cient mass to reach the limit for gravitational fragmentation and form chains of dense cores. measure the radial component of the velocity field, so our es-timate of the structure function refers only to this radial com-ponent. Fig. 10 shows in colored squares the structure functionfor all the chains but B10-W, which was found to be the super-position of two di ff erent gas components. As can be seen, thestructure functions of B213, B10E, and B7 / L1495 are approxi-mately flat or present only smooth oscillations as a function of l .The structure function of B218 (green squares) presents a spikenear l = .
15 pc, but this feature is caused by the strong redden-ing of the emission near the YSO, which we saw is likely due tooutflow feedback. At larger distances, corresponding to the sep-aration between the starless cores at each side of the YSO, thestructure function of B218 converges to a low value comparableto that of the other chains.The flat structure functions of the chains contrast with thepower-law functions of the extended gas indicative of turbulentmotions (Larson 1981; Elmegreen & Scalo 2004; Heyer & Brunt2004). This is illustrated in Fig. 10 with two dashed lines. Theblue line represents the classical Larson (1981) relation, and thered line is the structure function estimated by Qian et al. (2012)for Taurus using a large-scale map of the CO emission (Gold-smith et al. 2008). As seen in the figure, both dashed lines followa similar power-law increase with l and deviate systematicallyfrom the flat structure functions of the chains. While the devi-ation between chains and cloud is most prominent in the B213chain due to its larger length, it is also significant in the otherthree chains due to the consistency of their behavior.The flatness of the structure functions in Fig. 10 adds evi-dence to the suggestion by Hacar et al. (2013), from C O data,that the gas in the velocity-coherent filaments has decoupledfrom the turbulent velocity field that dominates the cloud as awhole. The observations, therefore, imply that turbulence doesnot dissipate at the ≈ . O velocity-coherent fibers. As a result ofthis dissipation of the turbulence at larger scales, the condensa-tion of the cores out of the chain gas must involve little kinematicchange, which is in agreement with the smooth oscillations seenin the radial profiles of velocity. Thus, while turbulence dom-inates the motions of the cloud gas at large scales, it appears to have been dissipated before the gas condenses into close-to-spherical dense cores.
4. Implications for core and star formation
We now combine our analysis of the internal structure of theN H + chains with the study of the large-scale C O emission byHacar et al. (2013) and a number of arguments from analytic the-ory and numerical simulations to present a scenario of core andstar formation in the L1495 / B213 region. A schematic view ofthis scenario is shown in Fig. 11 with a three-step time sequenceof the gas evolution in the cloud. For reasons that should be clearbelow, we refer to this scenario as “fray and fragment.”
The first step in the scenario consists of the formation of the10 pc-long L1495 / B213 region. This event, like the formationof the rest of the Taurus molecular cloud, most likely resultedfrom the collision between two large-scale flows of gas, as pro-posed by a number authors and implied by numerical simulations(Ballesteros-Paredes et al. 1999; Hartmann et al. 2001; Padoan &Nordlund 2002; Mac Low & Klessen 2004; Vázquez-Semadeniet al. 2007; Heitsch et al. 2008). A lower limit to the velocity ofthis collision can be estimated using the velocity spread of thegas in the L1495 / B213 filament. The C O data of Hacar et al.(2013) show some spectra containing multiple velocity compo-nents that range in LSR centroid velocity between 4.8 and 7.0km s − (their Fig. 8). This spread implies that the relative veloc-ity of the converging flows was at least 2.2 km s − .Since the velocity spread of 2.2 km s − refers to C O-emitting gas, which has a typical temperature of 10 K, it cor-responds to a Mach number of about 11. This implies that theinternal motions in the large-scale filament of L1495 / B213 defi-nitely belong to the supersonic regime.The second step in the scenario of Fig. 11 is the generationof substructure inside the 10 pc-long L1495 / B213 filamentarycloud. This substructure was studied in detail by Hacar et al.(2013), who analyzed the C O emission from the cloud with theFriends In VElocity (FIVE) algorithm and identified 35 distinctvelocity-coherent fibers. These fibers run approximately parallelto the 10 pc-long L1495 / B213 cloud and criss-cross each otherlike the threads of a frayed rope.
Article number, page 14 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex
Fig. 12.
Distance to the nearest neighbor among the N H + cores inL1495 / B213 (no projection correction applied). Note how most valueslie below 0.2 pc.
While the relative velocity between fibers spans the full2.2 km s − range, the velocity dispersion of the gas inside eachfibers is much lower and lies in the subsonic or slightly tran-sonic regime. Such combination of low internal velocity dis-persion and large-scale continuity of the velocity implies thatthese fibers have already decoupled from the large-scale turbu-lent velocity field of the cloud, and that therefore represent acritical scale of velocity dissipation. The recent numerical simu-lations of cloud turbulence by Moeckel & Burkert (2014) seemto confirm this. The simulations show that large-scale filamen-tary structures quickly evolve into bundles of intertwined fiberslike those of L1495 / B213 due to the combined e ff ect of vorticityand self gravity (see also Kritsuk et al. 2013, Smith et al. 2013,and Myers et al. 2014).The final step of the scenario shown in Fig. 11 is the forma-tion of dense cores out of the velocity-coherent fibers. Hacar etal. (2013) showed that only a few of these fibers are responsiblefor the formation of all dense cores cores in L1495 / B213, and re-ferred to them as “fertile.” These fertile fibers preferentially formmultiple cores, so they seem to have an intrinsic predispositionto core formation along most of their length. A natural origin forthis predisposition is gravitational instability, as implied by thehigher value of the mass-per-unit-length in these fibers.To test this interpretation, we have studied the typical sepa-ration between dense cores along the chains. This has been doneusing the core positions determined in Sect. 2 and calculating thedistribution of distance to the nearest neighbor. Fig. 12 shows ahistogram with the results. As can be seen, the distribution isdominated by values smaller than 0.2 pc, with 3 / / B213 complex. Thus, we have assumed a moderateinclination angle of 45 degrees, which implies a foreshorteningfactor of only 1.4. With this correction, the true median distancebetween cores is estimated to be approximately 0.2 pc.To compare the above inter-core distance with the expecta-tion from gravitational fragmentation, we use as a guide the self-gravitating isothermal cylinder model. This model can only rep-resent a first-order approximation to the chains, since we have seen that their mass per unit length likely exceed the model pre-diction. In addition, the turbulent origin of the fibers and chainsmakes it unlikely that the gas in them has had su ffi cient time tofully relax to an isothermal equilibrium configuration, or evento acquire the axial symmetry required for a cylindrical geome-try. Still, the deviations from equilibrium are unlikely to be verylarge, since the prevalence of subsonic linewidths in the spec-tra indicates an absence of shocks. Under these conditions, theisothermal cylinder model must provide a reasonable approxi-mation, especially considering that the characteristics of gravita-tional fragmentation are robust with respect to changes in geom-etry and contributions from rotation and magnetic fields (Larson1985).As shown by Stodólkiewicz (1963), an isothermal cylinderwith an equilibrium mass per unit length is unstable to sinu-soidal perturbations that have a wavelength larger than a criti-cal value of 3 . c / √ G ρ , where ρ c is the central density of thefilament. These perturbations fragment the cylinder into a a se-ries of equally-spaced condensations whose separation is givenby the wavelength of the perturbation applied to the system, orif no single-wavelength perturbation is imposed, by the wave-length of the fastest-growing unstable mode, which is twice thecritical wavelength (Nagasawa 1987, also Larson 1985; Inutsuka& Miyama 1997; Fischera & Martin 2012). In Sect. 3.5 we esti-mated that the core chains have a typical pre-fragmentation cen-tral density of ≈ . × cm − . For the isothermal cylindermodel, this density implies a critical wavelength of 0.14 pc, anda fastest-growing mode wavelength of 0.28 pc, assuming the gasis at 10 K. These values are in reasonable agreement with ourestimate of a typical inter-core distance of 0.2 pc. Strictly speak-ing, an inter-core value closer to 0.3 pc would have been ex-pected, but it is also possible that the initial central density of thefibers has been underestimated, and that this has led to an over-estimate of the expected critical wavelength and the separationbetween the cores. Given all the uncertainties in our estimates ofthe chain physical parameters, it appears that gravitational frag-mentation is the likely mechanism responsible for the formationof the dense cores inside the chains.If correct, the picture of core formation by gravitational frag-mentation has several important consequences. In first place,fragmentation can only occur if an isothermal cylinder has su ffi -cient mass per unit length. Since core formation in L1495 / B213is restricted to a small number of fertile fibers, it appears thatonly fertile fibers reach the required mass-per-unit-length, andthat the rest of the fibers remain sterile because they never accu-mulate enough mass (this is supported by the analysis of Hacaret al. 2013). This interpretation suggests that core formation inL1495 / B213 is regulated by how much mass the fibers can ac-cumulate, and that inability to reach the critical mass is a mainbottleneck in the core and star formation process. Given the tur-bulent state of the large-scale cloud, early dissipation of ster-ile fibers by shocks is the most likely mechanism to limit theirgrowth. Star formation is thus limited, not by failed cores, but byfailed (or sterile) fibers.Another consequence of the gravitational fragmentation sce-nario is that dense cores may not be equilibrium structures, sincethey originate from an instability that cannot be reversed withoutexternal energy injection. This conclusion may seem to contra-dict the idea that cores are in gravitational equilibrium, as im-plied by the good fit of their radial density profiles with mod-els of isothermal (Bonnor-Ebert) equilibrium (Alves et al. 2001).Equilibrium density profiles, however, are not a guarantee of truehydrostatic equilibrium, since they are expected to occur alsoduring the first stages of gravitational collapse (Kandori et al.
Article number, page 15 of 17 & A proofs: manuscript no. 24576ms H + maps, the chainshave a typical full length of 0.5 pc, which is only about twicethe wavelength of the fastest-growing mode. The chains there-fore hardly qualify as “infinitely long,” and their fragmentationis likely to have been a ff ected by edge e ff ects. Chain edges,however, are unlikely to be sharp, since this would favor therapid production of condensations near the boundaries (Burkert& Hartmann 2004), which is not seen in the maps. More likely,the edges of the chains are characterized by a smooth densitydecrease, which is expected to favor fragmentation in the chaininterior (Nelson & Papaloizou 1993). Other deviations from theidealized infinite cylinder model include the natural bending ofsome of the chains, like B10, and the likely presence of signifi-cant initial perturbations in the density profile as a result of theirturbulent environment. These additional elements are likely re-sponsible for the already-discussed irregular pattern of fragmen-tation in the chains, by which cores that have already formedstars are intermixed with starless cores at an earlier stage of evo-lution. Numerical simulations of bundles of filaments, like thosepresented by Moeckel & Burkert (2014), are needed to explorehow gravitational fragmentation occurs under more realistic con-ditions.We finish by noting that the fray and fragment scenario pro-posed here has been inspired by the analysis of L1495 / B213,but that it could apply to other regions of core and star forma-tion. The analysis of the N H + emission from the NGC 1333cluster-forming region by Hacar et al. (2014, in preparation), forexample, shows that the dense gas in this region forms a net-work of velocity-coherent fibers very similar to those found inL1495 / B213, although with a higher density of fibers per unitvolume, as it would be expected for a cluster-forming region.In addition, a number of large-scale dust-continuum images ofclouds show that what initially appears as a single large-scale fil-ament, on close inspection is resolved into a network of closely-aligned, small-scale fibers containing dense cores and embeddedyoung stars. This is noticeable in some of the SPIRE images ofthe Herschel Gould Belt Survey, and is especially prominent inthe APEX image of Orion presented by the ESO press release eso1321 (see also Stanke et al. 2013). Velocity-resolved obser-vations of these filamentary structures are needed to test whethera fray and fragment scenario can explain core formation in otherregions and in more massive environments.
5. Conclusions
We have presented new N H + (1–0) and C O(2–1) observationsof the dense regions of L1498 / B213 in Taurus made with theIRAM 30m telescope and complemented with archival dust-continuum observations from the Herschel Space Observatory.From the analysis of the combined dataset, we have reached thefollowing main conclusions.1. The dense cores in L1498 / B213 tend to cluster in lineargroups of three cores on average, which we call chains. Thisclustering is evident in the maps of integrated emission, and pro-duces a significant peak in the mean surface density of compan-ions for separations smaller than about 500 (cid:48)(cid:48) ( ≈ . (cid:48)(cid:48) (0.2 pc).
2. The N H + integrated intensities are correlated with the H column densities estimated from the SPIRE-Herschel data. Thiscorrelation is approximately linear, and in some chains it has asmall threshold of H column density that seems to arise fromadditional gas components. The linear correlation indicates thattowards the chains the N H + emission traces most of the materialseen in the Herschel images, and therefore provides velocity in-formation to the continuum data. It also indicates that the N H + abundance is approximately uniform with a value of 5 × − .3. A simplified Monte Carlo model of the N H + radial pro-files indicates that the density of the chains decreases withradius as a softened power law. Typical central densities are6 − × cm − and half-maximum diameters are 0.05 pc.4. While the C O emission reveals the presence of multiplevelocity components separated by supersonic speeds (analyzedin detail by Hacar et al. 2013). the N H + emission shows thatthe dense gas is overwhelmingly subsonic and continuous in ve-locity. This gas has appears to have decoupled from the turbulentvelocity field of the cloud and does not follow the standard Lar-son relation despite extending for up to 1 pc in length.5. When combined with the analysis of the C O emissionfrom Hacar et al. (2013), our observations suggest a scenarioof core formation which we refer to as fray and fragment . Inthis scenario, a collision between two supersonic gas flows hascreated the large-scale L1498 / B213 filament. Due to a combina-tion of turbulence and self-gravity, the large-scale filament hassplit into a network of smaller and intertwined filamentary struc-tures or fibers ( fray step). Some of these fibers have accumu-lated enough material to exceed the mass-per-unit-length limit ofgravitational instability and to fragment forming chains of densecores. Although this scenario is motivated by L1498 / B213, ad-ditional observations suggest that it may apply to other regions.
Acknowledgements.
We thank the IRAM sta ff for support during the observa-tions, and Markus Schmalzl for generously providing us with his extinction dataand for information on the relation between NIR extinction and SPIRE 500 µ memission. We also thank Joaquín Santiago-García for his rendering of the frayand fragment scenario shown in Fig. 11. This research was performed in partthanks to financial support from projects FIS2012-32096 and AYA2012-32032of Spanish MINECO and from the MICINN program CONSOLIDER INGE-NIO 2010, grant “Molecular Astrophysics: The Herschel and ALMA era - AS-TROMOL” (ref.: CSD2009-00038). AH acknowledges support from the Aus-trian Science Fund (FWF). This research has made use of NASA’s AstrophysicsData System Bibliographic Services together with the SIMBAD database and theVizieR catalogue access tool operated at CDS, Strasbourg, France. References
Aikawa, Y., Herbst, E., Roberts, H., & Caselli, P. 2005, ApJ, 620, 330Alves, J. F., Lada, C. J., & Lada, E. A. 2001, Nature, 409, 159André, P., Men’shchikov, A., Bontemps, S., et al. 2010, A&A, 518, L102André, P., Di Francesco, J., Ward-Thompson, D., et al. 2013, arXiv:1312.6232Arzoumanian, D., André, P., Didelon, P., et al. 2011, A&A, 529, L6Ballesteros-Paredes, J., Vázquez-Semadeni, E., & Scalo, J. 1999, ApJ, 515, 286Barnard, E. E. 1907, ApJ, 25, 218Benson, P. J., & Myers, P. C. 1989, ApJS, 71, 89Bergin, E. A., Alves, J., Huard, T., & Lada, C. J. 2002, ApJ, 570, L101Bergin, E. A., & Tafalla, M. 2007, ARA&A, 45, 339Bernes, C. 1979, A&A, 73, 67Bontemps, S., Andre, P., Terebey, S., & Cabrit, S. 1996, A&A, 311, 858Burkert, A., & Hartmann, L. 2004, ApJ, 616, 288Carter, M., Lazare ff , B., Maier, D., et al. 2012, A&A, 538, A89Caselli, P., Benson, P. J., Myers, P. C., & Tafalla, M. 2002, ApJ, 572, 238Caselli, P., Myers, P. C., & Thaddeus, P. 1995, ApJ, 455, L77Caselli, P., Walmsley, C. M., Tafalla, M., Dore, L., & Myers, P. C. 1999, ApJ,523, L165Cernicharo, J., Bachiller, R., & Duvert, G. 1985, A&A, 149, 273Daniel, F., Dubernet, M.-L., Meuwly, M., Cernicharo, J., & Pagani, L. 2005,MNRAS, 363, 1083 Article number, page 16 of 17. Tafalla and A. Hacar : Chains of dense cores in the Taurus L1495 / B213 complex di Francesco, J., Evans, N. J., II, Caselli, P., et al. 2007, Protostars and Planets V,17Dobashi, K., Uehara, H., Kandori, R., et al. 2005, PASJ, 57, 1Elias, J. H. 1978, ApJ, 224, 857Elmegreen, B. G., & Scalo, J. 2004, ARA&A, 42, 211Evans, N. J., II, Dunham, M. M., Jørgensen, J. K., et al. 2009, ApJS, 181, 321Evans, N. J., II, Rawlings, J. M. C., Shirley, Y. L., & Mundy, L. G. 2001, ApJ,557, 193Fischera, J., & Martin, P. G. 2012, A&A, 542, A77Gaida, M., Ungerechts, H., & Winnewisser, G. 1984, A&A, 137, 17Galli, D., Walmsley, M., & Gonçalves, J. 2002, A&A, 394, 275Gehman, C. S., Adams, F. C., & Watkins, R. 1996, ApJ, 472, 673Goldsmith, P. F., Heyer, M., Narayanan, G., et al. 2008, ApJ, 680, 428Gomez, M., Hartmann, L., Kenyon, S. J., & Hewett, R. 1993, AJ, 105, 1927Gomez, M., Whitney, B. A., & Kenyon, S. J. 1997, AJ, 114, 1138Goodman, A. A., Barranco, J. A., Wilner, D. J., & Heyer, M. H. 1998, ApJ, 504,223Gri ffi n, M. J., Abergel, A., Abreu, A., et al. 2010, A&A, 518, L3Hacar, A., & Tafalla, M. 2011, A&A, 533, A34Hacar, A., Tafalla, M., Kau ff mann, J., & Kovács, A. 2013, A&A, 554, A55Hartmann, L., Ballesteros-Paredes, J., & Bergin, E. A. 2001, ApJ, 562, 852Hartmann, L. 2002, ApJ, 578, 914Heitsch, F., Hartmann, L. W., Slyz, A. D., Devriendt, J. E. G., & Burkert, A.2008, ApJ, 674, 316Hennemann, M., Motte, F., Schneider, N., et al. 2012, A&A, 543, L3Henning, T., Linz, H., Krause, O., et al. 2010, A&A, 518, L95Heyer, M. H., & Brunt, C. M. 2004, ApJ, 615, L45Hildebrand, R. H. 1983, QJRAS, 24, 267Inutsuka, S.-i., & Miyama, S. M. 1997, ApJ, 480, 681Jijina, J., Myers, P. C., & Adams, F. C. 1999, ApJS, 125, 161Juvela, M., Ristorcelli, I., Pelkonen, V.-M., et al. 2011, A&A, 527, A111Kandori, R., Nakajima, Y., Tamura, M., et al. 2005, AJ, 130, 2166Kirk, J. M., Ward-Thompson, D., Palmeirim, P., et al. 2013, MNRAS, 432, 1424Könyves, V., André, P., Men’shchikov, A., et al. 2010, A&A, 518, L106Kritsuk, A. G., Lee, C. T., & Norman, M. L. 2013, MNRAS, 436, 3247Larson, R. B. 1981, MNRAS, 194, 809Larson, R. B. 1985, MNRAS, 214, 379Larson, R. B. 1995, MNRAS, 272, 213Launhardt, R., Stutz, A. M., Schmiedeke, A., et al. 2013, A&A, 551, A98Lee, C. W., Myers, P. C., & Tafalla, M. 1999, ApJ, 526, 788Lee, C. W., Myers, P. C., & Tafalla, M. 2001, ApJS, 136, 703Lombardi, M., Lada, C. J., & Alves, J. 2010, A&A, 512, A67Luhman, K. L., Allen, P. R., Espaillat, C., Hartmann, L., & Calvet, N. 2010,ApJS, 186, 111Mac Low, M.-M., & Klessen, R. S. 2004, Reviews of Modern Physics, 76, 125Mangum, J. G., Emerson, D. T., & Greisen, E. W. 2007, A&A, 474, 679Moeckel, N., & Burkert, A. 2014, arXiv:1402.261Molinari, S., Swinyard, B., Bally, J., et al. 2010,Mooley, K., Hillenbrand, L., Rebull, L., Padgett, D., & Knapp, G. 2013, ApJ,771, 110 A&A, 518, L100Moriarty-Schieven, G. H., Wannier, P. G., Tamura, M., & Keene, J. 1992, ApJ,400, 260Myers, P. C. 2009, ApJ, 700, 1609Myers, A. T., Klein, R. I., Krumholz, M. R., & McKee, C. F. 2014, MNRAS,439, 3420Nagasawa, M. 1987, Progress of Theoretical Physics, 77, 635Nakamura, F., Hanawa, T., & Nakano, T. 1993, PASJ, 45, 551Narayanan, G., Snell, R., & Bemis, A. 2012, MNRAS, 425, 2641Nelson, R. P., & Papaloizou, J. C. B. 1993, MNRAS, 265, 905Ostriker, J. 1964, ApJ, 140, 1056Padoan, P., & Nordlund, Å. 2002, ApJ, 576, 870Palmeirim, P., André, P., Kirk, J., et al. 2013, A&A, 550, A38Pilbratt, G. L., et al. 2010, A&A, 518, L1Qian, L., Li, D., & Goldsmith, P. F. 2012, ApJ, 760, 147Rebull, L. M., Padgett, D. L., McCabe, C.-E., et al. 2010, ApJS, 186, 259Recchi, S., Hacar, A., & Palestini, A. 2013, A&A, 558, A27Sandell, G., Weintraub, D. A., & Hamidouche, M. 2011, ApJ, 727, 26Santiago-García, J., Tafalla, M., Johnstone, D., & Bachiller, R. 2009, A&A, 495,169Schmalzl, M., Kainulainen, J., Quanz, S. P., et al. 2010, ApJ, 725, 1327Schneider, S., & Elmegreen, B. G. 1979, ApJS, 41, 87Schöier, F. L., van der Tak, F. F. S., van Dishoeck, E. F., & Black, J. H. 2005,A&A, 432, 369Simon, M. 1997, ApJ, 482, L81Smith, R. J., Shetty, R., Beuther, H., Klessen, R. S., & Bonnell, I. A. 2013, ApJ,771, 24Stanke, T., Stutz, A., Megeath, T., & HOPS Team 2013, Protostars and PlanetsVI Posters, 4Stodólkiewicz, J. S. 1963, Acta Astronomica, 13, 30Suutarinen, A., Haikala, L. K., Harju, J., et al. 2013, A&A, 555, A140Tafalla, M., Myers, P. C., Caselli, P., Walmsley, C. M., & Comito, C. 2002, ApJ,569, 815Tafalla, M., Myers, P. C., Caselli, P., & Walmsley, C. M. 2004, A&A, 416, 191Tafalla, M., Santiago, J., Johnstone, D., & Bachiller, R. 2004, A&A, 423, L21Tatematsu, K., Umemoto, T., Kandori, R., & Sekimoto, Y. 2004, ApJ, 606, 333Vázquez-Semadeni, E., Gómez, G. C., Jappsen, A. K., et al. 2007, ApJ, 657, 870Ward-Thompson, D., André, P., Crutcher, R., et al. 2007, Protostars and PlanetsV, 33Ysard, N., Abergel, A., Ristorcelli, I., et al. 2013, A&A, 559, A133mann, J., & Kovács, A. 2013, A&A, 554, A55Hartmann, L., Ballesteros-Paredes, J., & Bergin, E. A. 2001, ApJ, 562, 852Hartmann, L. 2002, ApJ, 578, 914Heitsch, F., Hartmann, L. W., Slyz, A. D., Devriendt, J. E. G., & Burkert, A.2008, ApJ, 674, 316Hennemann, M., Motte, F., Schneider, N., et al. 2012, A&A, 543, L3Henning, T., Linz, H., Krause, O., et al. 2010, A&A, 518, L95Heyer, M. H., & Brunt, C. M. 2004, ApJ, 615, L45Hildebrand, R. H. 1983, QJRAS, 24, 267Inutsuka, S.-i., & Miyama, S. M. 1997, ApJ, 480, 681Jijina, J., Myers, P. C., & Adams, F. C. 1999, ApJS, 125, 161Juvela, M., Ristorcelli, I., Pelkonen, V.-M., et al. 2011, A&A, 527, A111Kandori, R., Nakajima, Y., Tamura, M., et al. 2005, AJ, 130, 2166Kirk, J. M., Ward-Thompson, D., Palmeirim, P., et al. 2013, MNRAS, 432, 1424Könyves, V., André, P., Men’shchikov, A., et al. 2010, A&A, 518, L106Kritsuk, A. G., Lee, C. T., & Norman, M. L. 2013, MNRAS, 436, 3247Larson, R. B. 1981, MNRAS, 194, 809Larson, R. B. 1985, MNRAS, 214, 379Larson, R. B. 1995, MNRAS, 272, 213Launhardt, R., Stutz, A. M., Schmiedeke, A., et al. 2013, A&A, 551, A98Lee, C. W., Myers, P. C., & Tafalla, M. 1999, ApJ, 526, 788Lee, C. W., Myers, P. C., & Tafalla, M. 2001, ApJS, 136, 703Lombardi, M., Lada, C. J., & Alves, J. 2010, A&A, 512, A67Luhman, K. L., Allen, P. R., Espaillat, C., Hartmann, L., & Calvet, N. 2010,ApJS, 186, 111Mac Low, M.-M., & Klessen, R. S. 2004, Reviews of Modern Physics, 76, 125Mangum, J. G., Emerson, D. T., & Greisen, E. W. 2007, A&A, 474, 679Moeckel, N., & Burkert, A. 2014, arXiv:1402.261Molinari, S., Swinyard, B., Bally, J., et al. 2010,Mooley, K., Hillenbrand, L., Rebull, L., Padgett, D., & Knapp, G. 2013, ApJ,771, 110 A&A, 518, L100Moriarty-Schieven, G. H., Wannier, P. G., Tamura, M., & Keene, J. 1992, ApJ,400, 260Myers, P. C. 2009, ApJ, 700, 1609Myers, A. T., Klein, R. I., Krumholz, M. R., & McKee, C. F. 2014, MNRAS,439, 3420Nagasawa, M. 1987, Progress of Theoretical Physics, 77, 635Nakamura, F., Hanawa, T., & Nakano, T. 1993, PASJ, 45, 551Narayanan, G., Snell, R., & Bemis, A. 2012, MNRAS, 425, 2641Nelson, R. P., & Papaloizou, J. C. B. 1993, MNRAS, 265, 905Ostriker, J. 1964, ApJ, 140, 1056Padoan, P., & Nordlund, Å. 2002, ApJ, 576, 870Palmeirim, P., André, P., Kirk, J., et al. 2013, A&A, 550, A38Pilbratt, G. L., et al. 2010, A&A, 518, L1Qian, L., Li, D., & Goldsmith, P. F. 2012, ApJ, 760, 147Rebull, L. M., Padgett, D. L., McCabe, C.-E., et al. 2010, ApJS, 186, 259Recchi, S., Hacar, A., & Palestini, A. 2013, A&A, 558, A27Sandell, G., Weintraub, D. A., & Hamidouche, M. 2011, ApJ, 727, 26Santiago-García, J., Tafalla, M., Johnstone, D., & Bachiller, R. 2009, A&A, 495,169Schmalzl, M., Kainulainen, J., Quanz, S. P., et al. 2010, ApJ, 725, 1327Schneider, S., & Elmegreen, B. G. 1979, ApJS, 41, 87Schöier, F. L., van der Tak, F. F. S., van Dishoeck, E. F., & Black, J. H. 2005,A&A, 432, 369Simon, M. 1997, ApJ, 482, L81Smith, R. J., Shetty, R., Beuther, H., Klessen, R. S., & Bonnell, I. A. 2013, ApJ,771, 24Stanke, T., Stutz, A., Megeath, T., & HOPS Team 2013, Protostars and PlanetsVI Posters, 4Stodólkiewicz, J. S. 1963, Acta Astronomica, 13, 30Suutarinen, A., Haikala, L. K., Harju, J., et al. 2013, A&A, 555, A140Tafalla, M., Myers, P. C., Caselli, P., Walmsley, C. M., & Comito, C. 2002, ApJ,569, 815Tafalla, M., Myers, P. C., Caselli, P., & Walmsley, C. M. 2004, A&A, 416, 191Tafalla, M., Santiago, J., Johnstone, D., & Bachiller, R. 2004, A&A, 423, L21Tatematsu, K., Umemoto, T., Kandori, R., & Sekimoto, Y. 2004, ApJ, 606, 333Vázquez-Semadeni, E., Gómez, G. C., Jappsen, A. K., et al. 2007, ApJ, 657, 870Ward-Thompson, D., André, P., Crutcher, R., et al. 2007, Protostars and PlanetsV, 33Ysard, N., Abergel, A., Ristorcelli, I., et al. 2013, A&A, 559, A133