WARM EXTENDED DENSE GAS AT THE HEART OF A COLD COLLAPSING DENSE CORE

Low-mass stars form through gravitational contraction within cold dense condensations, called dark cloud dense cores. The dense cores remain roughly isothermal as long as radiative cooling is comparable to the heating from compression of the medium. This compression is due to the gravitational contraction over ∼10⁵ years (e.g., Masunaga et al. 1998). When the central density reaches 10¹⁰–10¹¹ cm⁻³, compression heating becomes dominant and a condensed, more or less star-sized object, protostellar core, forms at the heart of a dense core (Boss & Yorke 1995; Masunaga et al. 1998).

A starless dense core is a cold (∼10 K) dense (≳10⁴ cm⁻³) large (∼20,000 AU) condensation of gas and dust (e.g., Bergin & Tafalla 2007). It does not harbor any embedded infrared point source (hereafter IPS). Some evolved starless cores have dynamically collapsing envelopes. On the other hand, class 0 objects (e.g., André et al. 2000), which are embedded in protostellar dense cores with a temperature of ∼30 K, a density of ≳10⁵ cm⁻³, and a size of ≲10,000 AU, are characterized by the existence of IPSs and large-scale molecular bipolar outflows driven by protostars. Note that there are dense cores that harbor very low luminosity embedded sources within them but do not have large-scale bipolar outflows. These objects were used to be categorized as starless cores, however, embedded sources were revealed by deep water maser emission survey at the Nobeyama 45 m telescope (e.g., Furuya et al. 2001) and deep infrared observations by the Spitzer Space Telescope (e.g., Dunham et al. 2008). They are in a stage more evolved than the starless core stage and younger than the class 0 stage. The formation of the protostellar cores must occur in such a transition stage.

In order to investigate when and how the birth of a protostellar core occurs, we observed four dense cores in the Taurus molecular cloud using two CO lines in submillimeter bands, i.e., J = 6 − 5 (wavelength λ 433.544 μm) and 7 − 6 (λ 371.650 μm) transitions. The energy levels at J = 5 and J = 6 correspond to 83 K and 116 K, respectively. The critical densities at the temperatures of 100 K traced by the molecule are several times 10⁵ cm⁻³. With the molecular lines that trace both warm and dense gas components, one can study the physical properties of the regions closely related to the final processes of protostellar core formation when material at the center starts warming up due to the formation of the protostellar core.

Our surveyed dense cores include L1521E, L1498, L1521F, and L1544. These cores are selected because (1) they are well studied and known as objects in early stages of star formation, i.e., the starless core stage or the stage of a dense core that harbors a very low luminosity IPS and does not show a large-scale bipolar outflow, and (2) they are reasonably isolated hence good targets to study internal star formation processes without any external disturbance. Among the aforementioned dense cores, L1521E is in the youngest evolutionary stage because of its richness in carbon-chain molecules (e.g., Hirota et al. 2002). All the cores except for L1521E show a gas infall signature (Hirota et al. 2002; Tafalla et al. 1998, 2004; Onishi et al. 1999). L1498 has a relatively low density of 10⁴ cm⁻³ and is in a younger evolutionary stage as compared to L1544 and L1521F. L1521F (Mizuno et al. 1994; Codella et al. 1997; Onishi et al. 1999; Shinnaga et al. 2004; Crapsi et al. 2004) has been one of the primary targets among dense cores because of the short dynamical timescale due to the high density of 10⁶ cm⁻³ at the center (Onishi et al. 1999), chemically evolved features (Shinnaga et al. 2004; Crapsi et al. 2004), and the discovery of the IPS, L1521F-IRS, which may be a protostar in the making, within the dense core using the Spitzer Space Telescope (Terebey et al. 2005, 2009; Bourke et al. 2006). L1544 shows a similar nature to L1521F, however, no IPS has been detected to date. Based on the observed properties, L1521F is the most evolved sample in the set.

For L1521F, the Spitzer observation imaged a bipolar nebula of scattered light extending in the east–west direction associated with L1521F-IRS (Bourke et al. 2006; Terebey et al. 2009), which may trace a cavity. However, no large-scale bipolar outflow is observed using spectroscopy. A deep centimeter continuum observation confirms that there are no shock-ionized inner regions of a bipolar outflow from a ∼0.1 L_☉ source (Harvey et al. 2002). L1521F-IRS is not categorized as a class 0 source owing to the lack of a large-scale bipolar outflow. It is in an evolutionary category similar to GF 9-2 (e.g., Furuya et al. 2006, 2009) that shows very weak water maser emission from an embedded source and does not have a large-scale bipolar outflow.

The CO observations in submillimeter bands were carried out at the Caltech Submillimeter Observatory (CSO). The CO J = 6 − 5 (frequency ν 691.473076 GHz; Goldsmith et al. 1981) and 7 − 6 (ν 806.651801 GHz; Schultz et al. 1985) data were taken on 2006 February 6–8 and on 2009 January 22 UT, respectively. The high-altitude dry site, sensitive receivers, and high efficiency of the telescope combined to permit the observations. The Dish Surface Optimization System was used during the observations to compensate for the gravitational deformation of the 10.4 m diameter reflector (Leong 2005). The telescope's pointing was checked every 1–2 hr using planets. The pointing accuracy is estimated to be ∼3'' for both observations, and the accuracy of the map registration of the two CO transitions is within 3''.

Cryogenically cooled SIS receivers operation at 4 K at the CSO that we used for the observations produced typical single sideband system temperature of ∼3900 K at 433.5 μm and ∼2300–3000 K at 371.7 μm (Kooi et al. 2000) measured with a 50 MHz bandwidth spectrometer. The beam chopping method was used. The beam sizes at the two wavelengths are about 10'' and 9'' at 433.5 and 371.7 μm, respectively. From observations of Mars and Saturn, the main beam efficiencies are measured to be ∼30% and ∼35% at 433.5 μm and 371.7 μm at elevations of 30° and 48°, respectively. The rms noise levels for both observations are about 0.6–1.4 K in T_mb. The velocity resolution of both observations is ∼0.02 km s⁻¹.

Among the observed dense cores, only one, L1521F, shows significant CO line emission at both transitions (Figure 1). The rest of the dense cores do not show the CO 6 − 5 line emission brighter than 0.3 K in the observed main beam temperatures (T_mb). Observed properties of CO lines of L1521F are summarized in Table 1. T_mb values of the CO transitions are weak (see Table 1), suggesting that the emission may be optically thin. The peak T_mb is observed at the position of R.A. = 4^h25^m3429, decl. = 26°45'87 (B1950.0) for both transitions. Note that L1521F-IRS is located at 9'' south of the peak position. Considering the thermal velocity width of the CO gas at 100 K is only about 0.5 km s⁻¹, a significant fraction of the broad velocity widths originates from non-thermal motions.

Zoom In Zoom Out Reset image size

We estimate the temperature of the observed warm dense gas components. Taking advantage of the similar beam sizes of the observations of the two transitions, one can derive the excitation temperature (T_ex) using the following equation (e.g., Shinnaga et al. 2008):

$\begin{equation} T_{\rm ex} = \frac{38.714\; {\rm K}}{\ln \big[\left(\frac{7}{6}\right)^2\frac{T_{\rm {mb CO} 6-5}}{T_{{\rm mb CO} 7-6}}\big]}. \end{equation} \tag{ 1 }$

T_ex values estimated at the positions of (R.A. offset, decl. offset) = (0'', 0''), (−10'', +10''), (−10'', 0''), and (−10'', −10'') are 57 ± 1.7, 40 ± 7, 74 ± 19, and 68 ± 5.5 K, respectively. Note that the CO 7 − 6 emission is not detected toward the position of L1521F-IRS, indicating that the temperature near the IPS is still low (≲30 K). Over the central 30'' × 30'' region (corresponds to 4200 AU), the averaged T_ex is calculated to be 34.4 ± 0.94 K. These temperatures are much higher than the T_ex measured with an N₂H⁺ transition, 5.0 K (Shinnaga et al. 2004).

Figure 2 presents the total integrated intensity map of the CO J = 6 − 5 emission that traces the warm extended dense gas (hereafter WEDG), overlaid on the dust continuum map at 850 μm that traces the cold (∼10 K) extended (16,000 AU) dense (≳10⁵ cm⁻³) condensation (Shinnaga et al. 2004). The observed effective radius of WEDG within the 3σ contour is 2400 AU (0.012 pc). WEDG sits in the central region of the cold condensation and is distributed somewhat asymmetrically, extending to the north and northwest directions from L1521F-IRS. The peak T_mb of the CO 6 − 5 transition is found at the position of (R.A. offset, decl. offset) = (0'', 0''), while the CO 6 − 5 total integrated intensity map peaks at the position near L1521F-IRS, i.e., (R.A. offset, decl. offset) = (0'', −10'') owing to a broad line width measured at the position (observed line width V_FWHM of 2.46 ± 0.46 km s⁻¹).

Zoom In Zoom Out Reset image size

We considered four candidate mechanisms for the excitation of WEDG. These are radiative excitation from L1521F-IRS, infrared pumping from L1521F-IRS, heating by UV radiation generated from a possible accretion disk, like the one that may explain narrow submillimeter CO line components associated with class 0 objects (Spaans et al. 1995), and heating due to shocks. Only the last one provides a convincing interpretation of the data. Radiative excitation from L1521F-IRS is dismissed as L1521F-IRS's luminosity is not high enough to pump CO molecules up to such a high J level over ≳2400 AU region. In fact, CO 7 − 6 emission is not detected near the L1521F-IRS position. Infrared pumping is ruled out due to the large vibrational level spacing of CO (∼3000 K; Carroll & Goldsmith 1981). The protostellar core's temperature does not become higher than 3000 K as molecular hydrogen begins to dissociate at ∼2000 K. It is implausible that UV radiation from the accretion disk heats the observed CO gas as the peak T_mb values of both transitions are not found toward the position of the IPS.

On the other hand, shocks would be able to pump the gas up to ∼100 K over a large region (Neufeld et al. 1995). In fact, a systematic velocity pattern is not measured in WEDG, indicating that WEDG may be excited in shock-heated regions generated by the collision between the cold dynamically collapsing components and outflowing/rotating components near the center. Considering the presence of L1521F-IRS, it is highly likely that some outflow activities have already been initiated. The cavity of the bipolar nebula also indicates the existence of a small jet associated with the protostar. Furthermore, L1521F-IRS may have a large rotating circumstellar disk with a size of the order of 10³ AU. The north–south elongated feature seen in the 850 μm dust continuum map may indicate the existence of the north–south elongated disk that is perpendicular to the cavity axis, i.e., perpendicular to the axis of the bipolar nebula. Based on the measured velocity gradient of 15 km s⁻¹ pc⁻¹ over 0.01 pc with a clump mass of 0.1 M_☉ by using the N₂H⁺ J = 1 − 0 transition at the center of the dense core (Shinnaga et al. 2004), the centrifugal radius of the collapsing medium is estimated to be about 10³ AU. In addition, the asymmetric distribution of WEDG supports the view that WEDG comes from the shocks but not from the small jet itself. If the two CO lines are in the LTE condition, the lower limit of the mass of WEDG becomes an order of 10⁻² M_☉.

Figure 3 illustrates a magnified view of the center of the dense core that we propose. The cavity with blue color may be in front of the disk, while the cavity with green color is behind the disk, along the line of sight. The cavity in front is opening toward the west from L1521F-IRS, which may channel part of the outflowing gas, making the shock easier to observe on the western side. The north–south extension of WEDG may be due to the circumstellar disk elongated along the north–south direction.

Zoom In Zoom Out Reset image size

We made survey observations using the two CO transitions in submillimeter bands to search for the dark cloud dense cores which harbor newly born protostellar cores and to investigate the physical properties of the objects. This study identified a transient stage of star formation between the starless core and class 0 stages. As the starless core stage progresses, the cold collapsing dense material starts warming up at the center due to the formation of a protostellar core and forms WEDG in the central region. We name this new stage "warm-in-cold core stage (WICCS)." One should search for WEDG using warm and dense gas tracers such as the CO 6 − 5 and 7 − 6 transitions to identify the objects in WICCS. This object would constitute a missing link in evolution between a starless core and a protostar, yielding an important step toward understanding of the formation mechanism of a protostellar core and a protostar. Survey observations of CO transitions in submillimeter bands toward dense cores in early evolutionary stages are necessary to add more samples in WICCS in order to obtain a complete picture of WICCS.

This research was performed at the Caltech Submillimeter Observatory, supported by NSF grants AST-05-40882 and AST-0838261. This work was also supported by Grant-in-Aids from the Ministry of Education, Culture, Sports, Science and Technology of Japan (No. 19204020 and No. 20740113). H.S. is grateful to Richard Chamberlin, Brian Force, and Hiroshige Yoshida for their support on the observations in 2009 January.