A&A 512, A3 (2010)

L1506: a prestellar core in the making

L. Pagani¹ - I. Ristorcelli² - N. Boudet² - M. Giard² - A. Abergel³ - J.-P. Bernard²

1 - LERMA & UMR 8112 du CNRS, Observatoire de Paris, 61 Av. de l'Observatoire, 75014 Paris, France
2 - CESR & UMR 5187 du CNRS/Université de Toulouse, 9 Av. du Colonel Roche, BP 4346, 31028 Toulouse Cedex 4, France
3 - IAS, Bât. 121, Université Paris-Sud, 91435 Orsay, France

Received 9 July 2009 / Accepted 2 December 2009

Abstract
Context. Exploring the structure and dynamics of cold starless clouds is necessary to understand the different steps leading to the formation of protostars. Because clouds evolve slowly, many of them must be studied in detail to identify different moments in a cloud's lifetime.
Aims. We study a fragment of the long filament L1506 in the Taurus region, which we name L1506, a core with interesting dust properties observed by the PRONAOS balloon-borne telescope.
Methods. To trace the mass content of L1506C and its kinematics, we mapped the dust emission, and the line emission of two key species, C¹⁸O and N₂H⁺. We also observed ¹³CO and C¹⁷O. We model the species emission using 1D Monte Carlo models.
Results. This cloud is reminiscent of L1498 but also exhibits peculiar features: i) a large envelope emitting solely in ¹³CO that contains a much smaller core with strong C¹⁸O depletion at its centre despite a low maximum opacity (  20 mag); ii) extremely narrow C¹⁸O lines indicating a low, non-measurable turbulence; iii) contraction traced by C¹⁸O itself (plus rotation); iv) unexpectedly, the kinematical signature of the external envelope is opposite to that of the core, the ¹³CO and C¹⁸O velocity gradients having opposite directions and the C¹⁸O line profile being blue peaked in contrast to the ¹³CO one which is red peaked. The core is large (r = 3  10⁴ AU) and not very dense (n(H₂)  5  10⁴ cm^-3, possibly less). This core is therefore not yet prestellar.
Conclusions. All these properties imply that the core is kinematically detached from its envelope and in the process of forming a prestellar core. This is the first time that the dynamical formation of a prestellar core has been witnessed. The extremely low turbulence could explain the strong depletion of this core despite its relatively low density and opacity in contrast to undepleted cores such as L1521E, which exhibits a turbulence at least 4 times as high.

Key words: stars: formation - ISM: clouds - ISM: abundances - ISM: molecules - ISM: structure - ISM: individual objects: L1506

1 Introduction

On their way to forming stars, clouds go through a contraction phase leading to the formation of prestellar cores (n  1  10⁵ cm^-3, Keto & Caselli 2008), which are usually identified by their lack of internal heating sources, their large CO depletion, and their strong NH₃ and N₂H⁺ emission lines for which turbulence is often subthermal. These cores subsequently collapse to form protostars. Many details pertaining to these two condensation steps remain obscure and we do not understand the conditions necessary to produce stars of different masses, high or low, the conditions necessary to initiate the contraction of the cloud or the collapse of the prestellar core, nor the rôle of the magnetic field in supporting the clouds against collapse. Obviously, since the evolution time is long, one has to observe many clouds as a substitute to tracking any single cloud in the process of forming a star. For each cloud, physical and chemical properties should be studied in detail to assess all important parameters such as temperature, density, kinematics, and their respective gradients, if any, and also chemical abundances by volume. By accumulating this type of studies, we can hope to evaluate the impact of cloud age or density upon the depletion phenomenon, distinguish contraction from oscillation, and describe how prestellar cores and protostars form.

Many prestellar cores have been identified and some have already been studied in detail, including L1506, L1506, L1506, and L1506. Among these objects, a few cases have been identified as early prestellar cores if their density is already higher than 1  10⁵ cm^-3, or simple cores probably about to form prestellar cores if not. The first type includes L1506, L1506, and L1506 (Hirota et al. 2004; Tafalla & Santiago 2004; Hirota et al. 2002), which, despite having a high density, show no sign of depletion (strong CO lines, weak or absent NH₃ and N₂H⁺ lines) and are considered to be both physically and chemically young, while the second type is represented by two cases presently, L1506 (Shirley et al. 2005; Tafalla et al. 2004) and the very dense globule L1506 in the Coalsack complex (Lada et al. 2004). In the study by Shirley et al. (2005), L1498 exhibits a low density (1-3  10⁴ cm^-3) and molecular depletion (as already reported by several other authors, especially Willacy et al. 1998, who, by studying L1498, reported the first C¹⁸O depletion in a prestellar core by comparison with a dust map). The G2 gas content has not yet been studied in detail and only its peculiar, ring-like dust shape indicates its unstable nature despite its low density. This core is expected to turn into a prestellar core (Lada et al. 2004). In this second category, which is considered to represent cores that are physically young but chemically evolved, we present a newcomer, a fraction of the L1506 filament.

The cloud that we present in this paper, L1506, attracted attention during a large-scale ¹³CO survey (Abergel et al. 1994) because of its low (though not unique) DIRBE / and IRAS /  colour ratios to which the ¹³CO maps were compared. Its dust properties were studied later on using the PRONAOS balloon-borne experiment continuum emission data in the submillimetre range (200-600 m) (Stepnik et al. 2003), albeit with a limited angular resolution (2-3.5). A significant change in the dust properties was detected between the diffuse and the dense part of the cloud: no emission from transiently heated small particles and a strong enhancement (by a factor of 3.4) in the submillimetre emissivity of the latter. This was interpreted as the signature of dust coagulation leading to the formation of fluffy aggregates (consisting of a mixture of very small and large grains). To understand this evolution and investigate the physical conditions associated with efficient aggregate formation (ice mantles, level of turbulence, densities, ...), we concentrate on the gas properties. It is presently clear that depletion plays a key rôle in the gas phase of dark clouds and basically two species are needed to map the gas inside the clouds, namely C¹⁸O and N₂H⁺. We may also use NH₃ as a tracer but N₂H⁺ has some advantages over NH₃ among which the possibility to use the same telescope as for C¹⁸O, giving a similar resolution and beam correction and also superior diagnostic capabilities as discussed by Pagani et al. (2007). By observing these species, both the undepleted (outer) and the depleted (inner) cores can be traced and their kinematics and physical properties can be modeled. With ¹³CO, we are also able to trace a fraction of the extended envelope. We present and model these observations in this paper and discuss the properties of this cloud.

2 Observations

All observations were performed using the IRAM-30 m telescope. The dark nebula L1506 is an elongated object (Onishi et al. 1996; Nercessian et al. 1988) and the SIMBAD reference position (  = 41831.1  = +251925 ) is outside the part of the cloud that we present here, which is centred on the PRONAOS emission peak (  = 41850  = +251915 , Stepnik et al. 2003) that we propose to name L1506 (A and B are already defined, Lee & Myers 1999). Its distance is estimated to be 140 pc (Elias 1978; Kenyon et al. 1994).

In September 2004, we mapped the filament dust emission at 1.2 mm with the MAMBO II bolometre. Skydips, pointings, and calibration source observations were regularly performed, and data reduction was executed using IRAM proprietary software. The cold filament studied by Stepnik et al. (2003) is clearly visible at the centre (Fig. 1). Spectroscopic observations were performed over several runs: 29 and 30 November 2003, 22 and 23 May 2004, 28 July 2004, and 18 July 2008. In all cases, the VESPA autocorrelator was used with frequency sampling varying from 6 to 20 kHz for 3 mm lines and 40 kHz for 1.3 mm lines (Table 1). All observations were performed in frequency switching mode. Receivers A&B were used except in July 2008, when the 1 mm 9 pixel cameras HERA 1&2 were used in the On-the-Fly mode still using frequency switching. The C¹⁸O and ¹³CO (J:1-0) lines were observed simultaneously by tuning the receiver halfway between the two frequencies and offsetting the VESPA autocorrelator subwindows by 210 MHz in front of the two lines. Pointing was regularly checked and found to be stable to within 3 .

Table 1:   Telescope settings.

3 Results and analysis

Figure 2 shows a large-scale ¹³CO (J:1-0) map obtained with POM-1, a 2.5 m dish telescope with 4.5 resolution and 100 kHz filters - 0.27 km s^-1 - that was in service at Bordeaux Observatory. The original data were taken from Nercessian et al. (1988) and the region of interest is delineated by a square. This map traces the large-scale envelope of the cloud. On the right-hand side, the C¹⁸O (J:2-1) integrated intensity map obtained on-the-fly with the 30-m over L1506C is shown. We degraded the resolution to 24 to increase the signal-to-noise ratio. A large hole in emission is visible in the centre of the map which is caused by CO depletion, as now well established. The PRONAOS strip centred on the 100 m emission peak traverses the most depleted part of the filament. This depletion is also remarkable in terms of spectrum peak temperature (Fig. 3) or integrated intensity (Fig. 4) along the PRONAOS strip. In these figures, the decrease in C¹⁸O intensity is correlated with the appearance of N₂H⁺, another sure sign of CO depletion as CO readily destroys N₂H⁺ (Caselli 2002; Pagani et al. 2005; Bergin & Langer 1997). To verify that the C¹⁸O lines do not decrease for opacity reasons in the centre of the cloud, we observed 3 positions in C¹⁷O (J:1-0). We chose two positions close to the C¹⁸O peaks on each side of the strip (75 ) and one in the middle, close to the local minimum of the C¹⁸O emission (Fig. 5). These 3 positions are indicated in Fig. 2 by 3 small white boxes superimposed on 3 of the crosses along the PRONAOS strip. Several interesting features are visible:

1.

The (0, 0) and (+75 , 0) offset positions (along the cut) display extremely narrow C¹⁷O and C¹⁸O lines (FWHM = 0.16 to 0.26 km s^-1, where FWHM represents full width half maximum).

2.

The existence of a velocity gradient across the strip, which possibly traces rotation in the C¹⁸O data.

3.

The existence of another velocity gradient for ¹³CO of opposite direction.

4.

The C¹⁸O (J:1-0) splitting towards the reference position, which is not very visible in the (J:2-1) line but is clearly so in the C¹⁷O (J:1-0) line, as the dotted lines indicate in Fig. 5.

Since most of the properties of this cloud change gradually with radius, it is difficult to differentiate between the core and the envelope. One parameter that changes abruptly however is the velocity field. We therefore define the envelope as the region where ¹³CO exhibits a different velocity field from the other species, while C¹⁸O and N₂H⁺ both trace the core, the outer, undepleted part the former and the inner depleted one the latter.

As far as we know, these CO narrow lines are the narrowest C¹⁸O and C¹⁷O lines reported so far, being comparable to the very narrow features seen in NH₃ and N₂H⁺ in sources such as L183 (Pagani et al. 2007). The main difference is that these C¹⁸O and C¹⁷O lines are emitted outside the depleted core, in a region where turbulence is usually still high. The lines are narrow enough for the C¹⁷O :1_7/2-0_5/2 and :1_3/2-0_5/2 hyperfine components, which are usually blended, to be completely separated here.

Figure 6 confirms the extraordinary narrowness of the lines towards the centre of the cloud. These C¹⁸O spectra were taken at (0, -25 ) in the frame of the PRONAOS cut, which is (-21.6 , -12.6 ) in the Equatorial J2000 frame. The (J:1-0) spectrum clearly contains two components, which we fitted with 2 independent Gaussians. We also fitted the (J:2-1) spectrum with two unconstrained Gaussians. The velocity of each of the two components agree for both lines within the uncertainty of the fit (a difference of 10 m s^-1 for an uncertainty of 30 and 40 m s^-1 for the left and right components, respectively). As for C¹⁷O in Fig. 5, the splitting of the two components is 200 m s^-1 and the width of the individual components is 150 (right) and 180 (left) m s^-1. For a standard gas kinetic temperature of 10 K, for which the optically thin thermal linewidth of CO is 0.12 km s^-1, this constrains the turbulent width contribution to the linewidth to be 90 m s^-1.

If the C¹⁸O and C¹⁷O splitting was caused by self-absorption, the exact reverse would be true, the C¹⁸O (J:2-1) line would have contained the deepest self-absorption feature and normally no such feature would have been seen in C¹⁷O since, due to its low abundance and to its hyperfine structure, it is very difficult to make it optically thick. These symmetric features cannot be caused by rotational Döppler shifting because for a narrow beam directed at the rotational axis, most of the material picked up by the beam moves at right angles to the line of sight with therefore no contribution to the velocity component along it. Only material very close to the rotational axis would contribute to the Döppler shifted emission. Since the inner core is depleted in CO, this contribution near the axis is not possible and therefore rotation cannot explain this splitting. To explain this double peak feature, the only possibility left is that the CO envelope is radially moving. The direction of movement (expansion or contraction) can be inferred from the asymmetry of the C¹⁸O (J:1-0) and (J:2-1) lines. Here, these C¹⁸O lines are stronger in the blue component than in the red component. As we confirm using the Monte Carlo model (Sect. 4), this can only be explained by contraction and not expansion despite the absence of heavy self-absorption, which is usually invoked to justify blue peaked profiles (Myers et al. 1996). Since the lines are almost independent because of their narrowness, the blue and red line contributions are each emitted from only one side of the cloud - front or back depending on the movement, expansion or contraction - and have no or very little interaction. The line formed in the rear part of the core originates in the (backside) low density envelope and obtains contribution from a material that becomes denser as the photons travel towards the observer, until they reach the depleted zone and then cross the remaining cloud without interaction. For a constant kinetic temperature, this means that the excitation temperature of the contributing material increases (or remains constant if thermalized) towards the observer and the successive layers add their contribution to the emergent signal. In contrast, the line formed on the near side of the cloud originates in the highest density and excitation temperature layer, just outside the depleted zone and obtains contributions from increasingly less dense material with therefore decreasing excitation temperature. When the excitation temperature decreases, a net absorption of photons can occur if opacity is not null. The C¹⁸O lines are not really optically thin (opacities are in the range of 0.1 to 0.5) and this is enough to slightly differentiate between the line that originates in the rear part of the cloud, which experiences no attenuation after leaving the highest excitation temperature layer and the line which comes from the front part and experienced a small absorption due to the outer, less excited layers. The effect is clearly visible towards the (0, 0) position: the C¹⁸O (J:1-0) line is 20% stronger on the blue side (1.81 and 1.48 K peak temperature for the blue and red components, a difference greater than 10, with  = 27 mK), and the C¹⁸O (J:2-1) line has lost its red peak because of higher opacity and higher sensitivity to density conditions (Fig. 5). The red component is therefore that emitted from the front part of the cloud, the blue component from the rear part and this indicates that the cloud is contracting rather than expanding. The amount of contraction of the C¹⁸O core is indicated by the two dotted lines in Fig. 5, namely 100 m s^-1. Therefore, the C¹⁸O outer core both rotates (from the (75 , 0) symmetrical velocity offsets seen in Fig. 5 as mentioned above) and contracts. To our knowledge, this is the first time that core contraction is observed based on C¹⁸O and C¹⁷O lines.

The N₂H⁺ hyperfine structure lines do not have such a high signal-to-noise ratio and do not show structured profiles. However, their displacement in velocity along the cut is similar to that of C¹⁸O as we can see in Fig. 7.

The ¹³CO (J:1-0) line, which was observed simultaneously with the C¹⁸O (J:1-0) line, has a very different behaviour (Figs. 3, 5 and 7). Its peak intensity is redshifted and its velocity gradient along the cut is opposite to that of the C¹⁸O line. From the large-scale ¹³CO map obtained with POM-1, we know that the L1506 filament has a width of 40 in the region around L1506C (Fig. 2). This is 6 times larger than the C¹⁸O extent. As we do not know how the ¹³CO velocity behaves beyond these central 6, we cannot claim that this velocity drift is the signature of a rotation of the large-scale filament but it definitely indicates that it is opposite to the C¹⁸O rotation (Figs. 5 and 7). Similarly, the red peak in the ¹³CO velocity profile is probably a sign of gas expansion (Figs. 3 and 5), while C¹⁸O traces a contraction. Therefore, the extended envelope has a dynamical behaviour that is the opposite of that of the core.

Figure 4 shows the asymmetry of the cloud in both dust and gas. It is remarkable that the C¹⁸O lines on each side of the cut have the same maximum integrated intensity despite a variation of 25% in peak intensity. It indicates that the quantity of gas is comparable and that the main reason for the intensity change is the line width difference. We also note that the N₂H⁺ integrated intensity profile is strongly asymmetric and varies abruptly at the top, two features that we cannot reproduce with our standard one-dimensional (1D) model (see Sect. 4). Its peak position is close to but not coincident with the dust peak position.

We observed the position with the second strongest N₂H⁺ integrated intensity (offset +25 along the PRONAOS strip) in N₂D⁺ (J:1-0) to check the level of deuteration in the inner core (the strongest peak - offset +37.5  - was identified afterwards). We found only a very weak signal (Fig. 8), the main hyperfine component being only 70 mK. A hyperfine fit was performed using the CLASS fitting routine MINIMIZE with the HFS option.

4 Models

4.1 Density profile

The density radial distribution was derived from the MAMBO map emission using a Wiener linear inversion method as described by Dupac & Giard (2002). Considering the elongated distribution (along the RA axis), we modelled the filament with a cylindrical geometry approximation, its axis being in the plane of the sky and perpendicular to the PRONAOS strip, as in Stepnik et al. (2003). We also accounted for a temperature inward decrease from 15 K to 8 K. This range was inferred from the  profile of the filament (Stepnik et al. 2003) combined with the predictions of either the Zucconi et al. (2001) or Bernard et al. (1992) dust temperature models in cold cores. We discretized the cylinder into iso-density and iso-temperature rings, and defined the array A to be

or

where (i,j) are, respectively, the indices of the rings and the lines of sight,  the column density, (i) the gas density in the ring i,  the observed intensity along the line of sight j, b is the noise level,  (T_i) is the black-body function evaluated for the temperature T of the ring i,  is the proton mass,  is the mean molecular weight of interstellar material in molecular clouds and   is the mass absorption coefficient which is assumed to be constant in the core and derived from the dust aggregates emissivity model by Ossenkopf & Henning (1994). The local density  in each ring (i) is assumed to be   , where we adopt the optimized inversion matrix

(cf. Dupac & Giard 2002), where

The most successful optimization was obtained using a ring thickness value of 50 , the result leading to an inner density of 2.4  10⁴ cm^-3 (averaged over a 50  radius core).

4.2 Line modeling

The density profile derived from the dust data was used as a first estimate in our 1D line radiative transfer models but we allowed the profile to change because the raw dust map is noisy and its final resolution in the inversion method is only 50 compared to the C¹⁸O (J:1-0) 24 and (J:2-1) 12 and because the dust properties are not strongly constrained and can vary, therefore causing a typical uncertainty of a factor of 2 in the derived density.

Table 2:   The Monte Carlo cloud model used to fit the spectra (Figs. 7 and 9-12).

For all three species, ¹³CO, C¹⁸O, and N₂H⁺, we fit a non-local thermodynamic equilibrium radiative transfer Monte Carlo model and attempt to reproduce the spectra (Figs. 9-12) and the velocity-position diagram (Fig. 7). The 1D Monte Carlo model was originally developed by Bernes (1979), and incorporates both microscopic turbulence and a radial velocity field. It was subsequently modified to include rotation (Pagani & Breart de Boisanger 1996) and a variant was developed to treat the hyperfine structure of N₂H⁺ (Pagani et al. 2007). The 1D cloud parameters are shown in Table 2 and the species abundances are traced as a function of radius in Fig. 13. Here the model is spherical with a radius equal to the dust cylinder radius and a rotation axis similar to the cylinder axis of the dust model, i.e., assumed to be aligned with the embedding filament elongation and therefore perpendicular to the PRONAOS strip. The model data to be compared with the observations are taken at the equator that is considered to be a good approximation of a cylinder slice, as already discussed in Pagani et al. (2007) for a similar case. For the ¹³CO emission, since our high spatial and velocity resolution strip only extends across 6 and because the ¹³CO line is optically thick and far more extended, it is difficult to develop a simple 1D model to reproduce and constrain the 30-m ¹³CO observations (the POM-1 data are both too coarse and noisy for this purpose). More generally, the cloud is not symmetrical and therefore our 1D Monte Carlo model cannot reproduce the various differences between positive and negative offsets. However, we attempted to fit globally the observations and overall to use a similar cloud description for the N₂H⁺ and C¹⁸O line models. Though the ¹³CO data are too difficult to reproduce with the 1D Monte Carlo code, we attempted to fit them as well, at least to obtain some estimate of the physical conditions in the outer parts to help explain the observations. From the C¹⁸O and N₂H⁺ data, we find that the cloud rotational axis could also be slightly offset along the PRONAOS cut (about +12 ) with respect to our reference position (which was defined somewhat arbitrarily). The peak N₂H⁺ integrated intensity is even farther away from the reference position (+37.5 ) but assuming the rotational axis to be this far does not fit with the overall velocity profile. Despite the cloud extension on the positive offset side being smaller than on the negative offset side, the fit is relatively good, especially towards the centre of the cloud. As mentioned above, we attempted to adjust the density profile finding a solution that is close to the formula of Tafalla et al. (2002) for the N₂H⁺ - C¹⁸O region

where n₀(H₂) = 5  10⁴ cm^-3, r₀ = 1  10⁴ AU, and  = 2.5. This profile provides a peak dust emission that is comparable to that inferred from the MAMBO observation, and the average density of 4  10⁴ cm^-3 in a 50  radius is within a factor of 2 of the peak density derived from the MAMBO observations. Apart from the dust properties that are not strongly constrained, an equally probable explanation of this factor of 2 difference is that the He + N₂H⁺ collisional coefficients that we presently use (Daniel et al. 2005) overestimate the true H₂ density needed to collisionally excite the N₂H⁺ to its J = 1 level as discussed in Pagani et al. (2007). From dust and N₂H⁺ measurements, we find that the density in this core is rather low, a factor of 2 to 4 below the threshold proposed by Keto & Caselli (2008) for prestellar cores. Therefore, it is not such a core.

The CO+N₂H⁺ model that we present here is by no means unique but represents a family of possible solutions. This family has remarkable features that are common to all solutions:

1.

A peak column density of N(H₂)  2  10²² cm^-2, which is equivalent to   20 mag (following Bohlin et al. 1978). This is low compared to most other cases with similar strong C¹⁸O depletion but slightly higher than in L1498, which has a peak column density of N(H₂)  1.3  10²² cm^-2 (Shirley et al. 2005).

2.

A strong central depletion (30) of C¹⁸O.

3.

C¹⁸O depletion that begins at a density between 1.5 and 2  10⁴ cm^-3.

Figure 13: H2 density and molecular abundance (relative to H2) as a function of radius.

4.

For both C¹⁸O lines (towards the positive offsets) and the N₂H⁺ line (for all offsets), the linewidth is so narrow that the turbulence contribution to the width is negligible and defined to be   68 m s^-1. To our knowledge, this has been the lowest turbulence reported for C¹⁸O lines. The line width is mostly dominated by the thermal width and the enlargement caused by macroscopic kinematics (rotation and infall).

5.

The radial velocity is strongly constrained in the C¹⁸O envelope but not so much in the N₂H⁺ internal part. However, a constant infall around 0.1 km s^-1 is compatible with the data. No infall is not.

6.

The rotational velocity decreases with increasing radius. Solid rotation is therefore excluded and velocity dependences of the type V  with provide good solutions.

7.

The temperature pattern is indicative of a slightly higher temperature in the N₂H⁺ inner core (10 K) than in the C¹⁸O outer core (8 K). If correct this would resemble to the B68 case (Bergin et al. 2006).

The C¹⁸O temperature is determined by the necessity to reproduce both the (J:1-0) and (J:2-1) lines with a relatively tightly constrained dust profile, whereas the N₂H⁺temperature is the minimum we could define while keeping the needed density as low as possible. Despite the observation of a single transition for N₂H⁺, we are constrained by the hyperfine structure (see Pagani et al. 2007). However, this temperature difference could possibly prove to be inaccurate if the actual H₂-N₂H⁺ collisional coefficients were available. As they are expected to be higher than the present He-N₂H⁺ ones, they would probably allow us to fit the observations with both a lower temperature and a lower density. In any case, the gas appears cooler than the dust ( K) as estimated by Stepnik et al. (2003). Their resolution may be too coarse and a large fraction of the (warmer) envelope emission intermingled with the core emission, or their single temperature fit too simple by lack of constraints (we have shown that in the case of the L1506 cold core, the 200 m emission does not trace the same dust as the submm emission and that a temperature gradient is compulsory, Pagani et al. 2004). Alternatively the gas and the dust being decoupled at these low densities (n(H₂)  5  10⁴ cm^-3 or less), the dust could possibly be warmer than the gas though this is not usually expected.

The modeling of the N₂D⁺ (J:1-0) line implies a column density of 4 to 8  10¹¹ cm^-2 depending on the chosen N₂D⁺ abundance profile, which we are unable to constrain with only one observation. The peak N₂H⁺ column density is 2  10¹³ cm^-2 and therefore the deuterium enrichment is 2-4%. Compared to L183 (Pagani et al. 2009b,2007), we identify a similar deuteration level at similar densities indicating that the evolutionary speed of the deuterium enrichment is probably similar and therefore can be used to determine the age of the core.

Beyond the core characterized by C¹⁸O narrow lines, the envelope, as traced by ¹³CO, is most accurately described by an almost constant density (in the range 200-1000 cm^-3) and a large amount of turbulence (  = 0.4 km s^-1). To follow its macroscopic velocity drift, we had to introduce a counter-rotating pattern in the external layers of a constant velocity of -0.3 km s^-1. We do not claim that the ¹³CO is indeed rotating but this is the only way to approach the velocity profile of this line with our model. Similarly, to reproduce the redshifted peak, we had to introduce an expansion velocity of 90 m s^-1, which is comparable in amplitude to the infall velocity of the core. The fit is not perfect inasmuch as we were unable to centre the peak emission in our model on the same position as in the observed position-velocity diagram (Fig. 7) but the fit bears some ressemblance to reality as shown in Fig. 11. The derived parameters are therefore reasonably close to the true situation and this envelope is clearly decoupled from the core in terms of density profile as much as in terms of turbulence and large-scale motions. This region also has a very high abundance of ¹³CO (6  10^-6) and an absence of C¹⁸O (<5  10^-9, an abundance of 1  10^-8 is marginally consistant with the observations). The density is not well constrained in the external envelope but since the ¹³CO abundance is already high (its standard value is 1-2  10^-6, i.e., 3 to 6 times less), the density cannot be lower than we have assumed here since a lower density would produce an even greater increase in the ¹³CO abundance. In contrast, we could increase the density to lower the abundance of ¹³CO, but it would become higher than the density in the outermost C¹⁸O layers. In the absence of any additional ¹³CO transitions, it is difficult to correctly constrain the density and abundance in these last, turbulent layers.

The collapsing core has a size of 3  10⁴ AU in radius and contains a mass of 4 . This size is larger than that of L1498, which was shown to be the largest by 50% of the nearby prestellar cores mentioned in the study by Shirley et al. (2005). Our mass estimate is somewhat approximate as the density of the N₂H⁺ region is possibly overestimated but on the other hand, the core could be slightly more extended than a sphere, and closer to a cylinder. The estimate derived from the dust observations is similar but more poorly constrained because of our lack of knowledge about the dust emissivity and temperature in the core on scales smaller than the PRONAOS resolution. The ¹³CO low density filament contains at least 20  accounting only for the part along its main axis that covers the core. Its diameter is 250 000 AU or 1.25 pc.

5 Discussion

Several properties of this area of the L1506 filament are remarkable and presently difficult to explain. The expanding and counter-drifting external envelope is one of them. One partially similar case has however been reported: Young et al. (1981) showed that the L1506 core as traced by C¹⁸O had a position/velocity profile gradient that had an opposite sense to the ¹²CO one, which they interpreted as a counter-rotating core (but they did not discuss the radial component). Contemporary theoretical work (Mouschovias & Paleologou 1979; Gillis et al. 1979) was invoked to explain this situation: a radial, frozen-in magnetic field perpendicular to the rotation axis could slow down the rotation of the cloud, but because the field remains anchored to the external regions, the magnetic torques do not vanish immediately and counter-rotation is induced in the core, which would be able to oscillate several times before coming to a halt. Whether this is truly the case here is unclear because 1) we are uncertain that the ¹³CO itself traces rotation; 2) the expansion of the ¹³CO envelope is not predicted by the model; and 3) we know nothing about the magnetic field inside this core (external parts have however been mapped, Moneti et al. 1984; Heiles 2000).

The strong and abrupt depletion of C¹⁸O at relatively low density is another remarkable property. This is similar to the L1498 case but in stark contrast to other very young cores, such as L1495B, L1521B, and L1521E, which exhibit no depletion despite a density that is typically 5 times higher (and possibly 10 times if He-N₂H⁺ collisional coefficients indeed over-estimate the true density in L1506C or if we compare to the L1498 peak density derived by Shirley et al. 2005). Based mainly on this absence of depletion, these cores have been judged to be young but it is unusual that after recently acquiring their present density state, they show no kinematical sign of contraction. It is thus conceivable that they are older than assumed and that the absence of depletion could be due to another mechanism, such as strong desorption, an ill-understood mechanism (Roberts et al. 2007). It is understood that energetic cosmic ray impacts can desorb the most volatile ices such as CO (Hasegawa & Herbst 1993; Bringa & Johnson 2004; Leger et al. 1985; Roberts et al. 2007). The desorption efficiency is inversely correlated with the size of the grains because smaller grains are easier to warm up. Therefore, the question arises of whether or not cores such as L1495B have smaller grains than L1506C or not and what the difference in size of these grains should be to allow or prevent an efficient desorption of CO. As explained by Roberts et al. (2007), all desorption mechanisms remain poorly known and it is presently difficult to go further. However, if this explanation proves to be valid, then one has yet to understand why grains coagulate in some sources and not in others. Turbulence could be one of the main key parameters involved in this differentiation between cores. Based on the analyses of Chokshi et al. (1993) and Poppe & Blum (1997) works, Flower et al. (2005) demonstrate that dust coagulation efficiency reaches a maximum for turbulent velocities in the range 10-80 m s^-1 (see their Fig. A.1), a condition that is hardly achieved even in the most quiescent prestellar cores but fits the present case. In contrast, L1521E shows no depletion, while the turbulence is suprathermal (even for N₂H⁺,  = 0.27 km s^-1, Tafalla & Santiago 2004, a factor of 4 at least larger than in L1506C). If the size of the grains in L1521E is not known, large grains have been advocated to explain submillimetre emission in L1506C. Stepnik et al. (2003) compiled a model where the very small grains, emitting at 60 m, disappear for offsets lower than r₀=4 from the centre, while the submm emissivity increases by a factor of 3.4 because of the formation of fluffy aggregates. Their observations were taken with an angular resolution of 2-3.5', so the value of r₀ could be lower and fit the extent (200 ) of the low turbulence region in L1506C. Grain growth, depletion, and desorption are certainly not completely understood at the moment.

The absence of C¹⁸O across 5/6th of the external envelope width as traced by the POM-1 ¹³CO observations is another puzzle. The ¹³CO abundance itself seems to be higher than usual so that the ¹³CO/C¹⁸O ratio reaches 600-1200 over a large region. Although ¹³CO fractionation does exist, it is weak and does not seem to be able to explain this difference. More observations of the envelope (including ¹³CO (J:1-0) and (J:2-1) and very low noise C¹⁸O transitions) are needed to address this issue.

Clear traces of collapse among prestellar cores was found in a few cases. Lee et al. (1999,2004) observed a large number of asymmetrically blue profiles towards a sample of starless cores but they traced the CS kinematics with respect to N₂H⁺ or DCO⁺ and, since CS is depleted inside the N₂H⁺ region, they traced the inward motion of the envelope only. A few prestellar cores are indeed modeled with a collapsing core, the most well studied being L1506 (e.g., van der Tak et al. 2005; Williams et al. 1999; Tafalla et al. 2002), which was reported to clearly undergo a collapse from line profile fitting. Tafalla et al. (2004) also invoke the possibility of infall from the observations of two other cores, L1506 and L1506 despite the absence of a clear velocity gradient and Williams et al. (2006) found that L1506 has a profile reminiscent of L1544, which is also indicative of probable infall. This was confirmed by Lee et al. (2007) who also reported the infall of L1506. Here the amplitude of the inward motion of N₂H⁺ is not strongly constrained but is clearly necessary to reproduce the velocity-profile plot: no infall at all is clearly the wrong result but the details of the radial velocity profile are uncertain. The inward motion is also clearly indicated by the C¹⁸O and C¹⁷O lines. It is the first time to our knowledge that these lines trace the collapse of a core. We understand that this is possible here because several favorable factors are met:

1.

No CO isotopologues are present in the central part of the core, which clearly separates the front and rear moving layers as indicated by C¹⁷O.

2.

No C¹⁸O in the extended envelope, which would have either hidden or blurred the faint variation in intensity between the blue and red emission that enable us to differentiate between expansion and contraction.

3.

Low C¹⁸O velocity dispersion, which increases the line opacity near unity (  = 0.56,  = 0.91). Combined with the outwardly decreasing excitation temperature, it allows for weak but measurable differential absorption between the blue and red components.

The most surprising aspect is that this core with a density of lower than 1  10⁵ cm^-3 is supposed to be dynamically stable given its total mass (less than 10 , Keto & Caselli 2008), which is clearly not the case here (the core mass being estimated to be 4 ). However, as noted by Keto & Caselli (2008), most clouds without non-thermal energy support would be unstable and, in the present case, turbulence is abnormally low, non-measurable in several parts of the core. This would support the idea of a possible collapse though we have no information about the magnetic field intensity and direction. If a magnetic field is indeed present and induces the core counter-rotation, it is not incompatible with infall because it is included in the magnetic braking model of Mouschovias & Paleologou (1979). Oscillations or sound waves were invoked in the case of e.g., B68 (Maret et al. 2007); one could be tempted to invoke a similar argument here but the rotation of C¹⁸O and ``counter-rotation'' of ¹³CO seem difficult to explain in terms of sound waves.

6 Conclusions

The large size of L1506C and its low density clearly indicate that it is not yet a prestellar core but its inward motion and its kinematical decoupling from the outer envelope are signs that it is in the process of turning into one. It is the first time that a low density core is clearly observed to collapse to form a prestellar core, both the kinematical and density status of L1498 being somewhat unclear. Two different evolutionary paths from normal clouds to prestellar cores seem to exist: one that begins with density enhancement first much before any depletion occurs and one that begins with depletion first. Two questions arise: what is the reason for this differentiation (turbulence via its action on grain growth, itself limiting the desorption capability of cosmic rays) and do both paths really produce standard prestellar cores? Coming observations with the Herschel Space Observatory will help us to address these questions and the nature of the dust inside this object.

Acknowledgements

We want to thank an anonymous referee for her/his fruitful comments which helped to improve the manuscript and P.F. Goldsmith for fruitful discussions. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France.

References

Footnotes

... making

Based on observations made with the IRAM-30 m. IRAM is supported by INSU/CNRS (France), MPG (Germany), and IGN (Spain).

... DIRBE

Diffuse Infra Red Background Experiment, http://lambda.gsfc.nasa.gov

... IRAS

Infra Red Astronomical Satellite, http://lambda.gsfc.nasa.gov

... PRONAOS

PROgramme NAtional d'Observations Submillimétriques.

... telescope

http://www.iram-institute.org/

... SIMBAD

http://simbad.u-strasbg.fr/sim-fid.pl

... MAMBO

MAx-Planck-Millimeter-BOlometer, see IRAM website.

... VESPA

VErsatile Spectroscopic and Polarimetric Analyzer, see IRAM website.

... HERA

HEterodyne Receiver Array, see IRAM website.

... POM-1

Petite Opération Millimétrique.

... expanding

There exists the alternative possibility that the blue peak is stronger than the red peak because of a difference in column densities between front and rear parts of the cloud. If this were the case in such a manner that it would hide the cloud expansion, the difference in column density between the two components would be 40% in the optically thin limit (and even more with growing opacity). Indeed, since the intensity difference is about 20% towards the (0, 0) position in C¹⁸O (J:1-0), it would have to be compensated twice to go from a stronger red component (due to expansion combined with differential excitation) to a weaker red component (as observed) and the effect would be more pronounced for the (J:2-1) component. This possibility seems therefore quite unlikely and even if the two sides of the cloud have different column densities, contraction remains the most probable explanation.

...option

http://www.iram.fr/IRAMFR/GILDAS/

...(Shirley et al. 2005)

In terms of the total dust amount and the peak density of L1498, note that Shirley et al. (2005) disagree with Tafalla et al. (2004). If we were to assume that the results of the latter authors be true, L1498 would be closer to a normal prestellar core than to a simple core. The discrepancy has been only partly explained by Shirley et al. (2005).

All Tables

Table 1:   Telescope settings.

Table 2:   The Monte Carlo cloud model used to fit the spectra (Figs. 7 and 9-12).

All Figures

---

Copyright ESO 2010