A&A 383, 591-597 (2002)
DOI: 10.1051/0004-6361:20011806

Gravitationally bound cores in a molecular cirrus cloud

A. Heithausen1 - F. Bertoldi2 - F. Bensch 3,4

1 - Radioastronomisches Institut der Universität Bonn, Auf dem Hügel 71, 53121 Bonn, Germany
2 - Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
3 - I. Physikalisches Institut der Universität zu Köln, Zülpicher Str. 77, 50937 Köln, Germany
4 - Harvard-Smithonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138, USA

Received 23 November 2001 / Accepted 18 December 2001

Using MAMBO at the IRAM 30m telescope we have observed a dense core in the cirrus cloud MCLD 123.5+24.9 in the dust continuum emission at 250GHz. The core is detected as an elongated filament with an extent of $4.\!'2\times 0.\!'7$, corresponding to 0.18pc$\times$0.03pc at an adopted distance of 150pc. We find a close correlation between the continuum emission and previously observed C18O (1$\to $0) line emission. Using standard dust models we derive hydrogen column densities of up to 1022cm-2. The total mass for the filament is about 0.66$M_{\odot}$.
We also present observations of the HC3N (3$\to $2), (4$\to $3), and (10$\to $9) emission lines obtained with the MPIfR 100m and the IRAM 30m telescopes. The distribution is very different from the dust continuum and the C18O (1$\to $0) line emission. HC3N is concentrated in two distinct clumps located at the ends of the filament seen in the other tracers. Based on a LVG analysis of the HC3N transitions we derive column densities of $N({\rm HC_3N})/\Delta v\approx10^{13}$cm-2/kms^-1kms-1 and volume densities of $n({\rm H}_2)\approx10^5$cm-3. We find that the HC3N clumps have masses of 0.13 and 0.19$M_{\odot}$. Our data demonstrate that the cirrus cloud cores are gravitationally bound, and that they show chemical structure indicating different evolutionary stages within the cloud.

Key words: stars: formation - ISM: abundances - ISM: clouds - ISM: individual objects: MCLD 123.5+24.9 - ISM: molecules - ISM: dust

1 Introduction

Stars form in dense cores of molecular clouds. One of the questions of current interest focuses on how dense cores form and how they eventually evolve into the protostellar Class 0 stage (e.g. Ward-Thompson et al. 1999). Most of the studies available concentrate on dense cores in regions with known star-forming activities (e.g. Motte et al. 1998). Galactic cirrus clouds have long been thought to be unable of forming any stars, because of their strong turbulent support against self-gravity (Magnani et al. 1987; Heithausen 1996). It did not surprise then that searches for T Tauri stars associated with high-latitude clouds remained unsuccessful (Martin & Kun 1996; Hearty et al. 1999), except maybe L1457, where several T Tauri stars were found (Hearty et al. 2000), but for which an actual spatial coincidence between the stars and the cloud is not well established.

The discovery of T Tauri stars far from actively star forming regions (e.g., Neuhäuser 1999) raises the question of their origin. Are they runaway stars that were ejected from the parent molecular cloud, or were they born in small cloudlets (Feigelson 1996) which then dispersed? In this context the star-forming abilities of molecular cirrus clouds, which are mostly found at high galactic latitudes, is of great interest. Recently, CS observations of the cirrus cloud MCLD123.5+24.9 (Heithausen 1999) showed spectroscopic signatures of infall motion (cf. Myers et al. 1996; Mardones et al. 1997) in one of its three CS cores, suggesting that star-formation is indeed possible in this type of clouds.


Table 1: Parameters for the HC3N observations.
Molecule transition Frequency Telescope FWHM $\eta_{\rm mb}$ $\delta v$ Positions
    (GHz)   ('')   ( kms^-1kms-1)  
HC3N $J=3\to2; F=3\to2$ 27.294295 MPIfR 100m 32 0.42 0.054 52
HC3N $J=4\to3; F=4\to3$ 36.392332 MPIfR 100m 26 0.42 0.060 2
HC3N $J=10\to9$ 90.978993 IRAM 30m 27 0.81 0.064 82

\par\includegraphics[angle=-90,width=17.8cm,clip]{h3322f1.eps} \end{figure} Figure 1: Comparison of the distribution of C18O (taken from Falgarone et al. 1998), dust continuum and HC3N. Contours are every 0.25K  kms^-1kms-1 starting at 0.5K  kms^-1kms-1 for the C18O (1$\to $0) line, every 2 mJy/beam starting at 0 mJy/beam for the dust continuum map, every 0.05K  kms^-1kms-1 starting at 0.05K  kms^-1kms-1 for the HC3N (10$\to $9) line, and every 0.1K  kms^-1kms-1 starting at 0.1K  kms^-1kms-1 for the HC3N (3$\to $2) line. Beam sizes are indicated in the lower left corner of the maps. Observed HC3N positions are marked in the respective maps. The arrows point towards the centres of HC3N-A and B, which are also the positions, for which the HC3N (4$\to $3) spectra have been obtained. The "X''s in the left two maps mark the centres of HC3N-A and B.
Open with DEXTER

MCLD 123.5+24.9 is a dense molecular cloud located in the Polaris Flare, a large molecular cirrus cloud in the direction of the north celestial pole (Heithausen & Thaddeus 1990). Its distance is between 130pc and 240pc (cf. Heithausen et al. 1993), and we shall here adopt a value of 150pc. The first molecules detected towards MCLD 123.5+24.9 were CO, 13CO, H2CO, and OH (Großmann et al. 1990), which shows an enhanced abundance. Later observations found NH3, HCO+, HCN, and HNC, with abundances typical of dark clouds (Großmann & Heithausen 1992), and implied kinetic temperatures between 6K and 15K. Further molecules detected here are C3H2, SO, CS, CCH (Gerin et al. 1997; Heithausen et al. 1995, 1999). Dust continuum observations made by IRAS, ISO, and PRONAOS (a balloon-borne submm-photometer) show steep spectrum emission from cold ( $T_{\rm d}=13$ K) dust (Bernard et al. 1999).

Here we present 1.2 mm (250 GHz) radio continuum observations of MCLD 123.5+24.9 obtained with the 37-channel Max-Planck Millimeter Bolometer (MAMBO) array at the IRAM 30m telescope. We also show observations of three HC3N transitions obtained with the MPIfR 100m and the IRAM 30m telescopes. Our data demonstrate that the cirrus cloud cores are gravitationally bound, and that they show chemical structure indicating different evolutionary stages within the cloud.

2 Observations

Table 1 summarizes the basic parameters of the spectroscopic observations. The HC3N (3$\to $2) and (4$\to $3) spectra were obtained in June 2000 with the MPIfR 100m radio telescope. For the lower transition we obtained a complete map with 52 positions. In the direction of the most intense (3$\to $2) line emission we also obtained two spectra in the (4$\to $3) transition. The velocity resolution of the autocorrelation spectrometer was 0.06kms^-1kms-1. The rms noise level in the final spectra is below 0.16K for the (3$\to $2), and 0.26K for the (4$\to $3) data. The HC3N (10$\to $9) transition was observed with the IRAM 30m telescope in September 2000 over a full map with 82 positions on a 15'' rectangular grid, with an rms noise in the final spectra below 0.05K.

Millimeter continuum observations were made on four different dates in March 2000 with the 37-channel bolometer MAMBO at the IRAM 30m telescope. MAMBO is sensitive between about 210 and 290 GHz, with an effective frequency of 250 GHz for steep thermal spectra. Eight maps covering about $4'\times4'$ each were co-added for the final map. The observations were taken in double-beam on-the-fly mode, i.e., chopping the secondary mirror in azimuth by 50'' to 70'' at 2Hz, and scanning the sky in azimuth at a speed of 4 or 5''s-1, then moving in elevation by 4''. The total observing time was about 8 hours. The maps were taken under variable winter weather conditions, with line of sight opacities between 0.2 and 0.7. The effective beam FWHM is $\approx$11''.

To our knowledge this is the first millimeter bolometer map of a galactic cirrus core. In Fig. 1 we compare the spectroscopic and continuum data to a complete C18O (1$\to $0) map of that cloud, which was previously obtained by Falgarone et al. (1998).

3 Dust continuum

Broadband continuum emission at mm and submm wavelengths yields information on the distribution of dust in molecular clouds. The dust emission provides an independent measure for the total mass surface density, which is otherwise hard to estimate, especially in high-latitude clouds.

As seen in Fig. 1, the core is clearly detected as an elongated filament with an extent of $4.\!'2\times 0.\!'7$, corresponding to 0.18pc $\times$ 0.03pc at the adopted distance of 150pc. The flux density, $S_\nu$, in a Gaussian beam with a half-power beamwidth $\Theta$, of thermal emission by dust with a temperature $T_{\rm d}$ and an optical depth $\tau_\lambda$ is

S_\nu={\pi\over4\ln2}~ \Theta^2~ (1-{\rm e}^{-\tau_\lambda})~ B_\nu(T_{\rm d}).
\end{displaymath} (1)

Following Mezger et al. (1995), the dust optical depth through a hydrogen nuclei column density $N_{\rm H}$ is

\tau_\lambda=b~ C~ \Big({\lambda\over{\rm mm}}
\Big)^m \Big({Z\over Z_0}\Big) ~N_{\rm H},
\end{displaymath} (2)

where ${Z\over Z_0}$ is the gas metallicity relative to the solar value, where $C=7\times10^{-27}\rm cm^{2}$ is the Draine & Lee (1984) dust absorption cross section per H nucleon at wavelength 1mm, and b is an empirical correction factor accounting for differences to the Draine & Lee diffuse cloud grains; $b\approx
1$-2 for grains in dark clouds, and we adopt b=1.9 (Mezger et al. 1995). For m=2 and $T_{\rm d}=13$K (Bernard et al. 1999), in the case of $\tau_\lambda\ll 1$ we find a linear relation between the H column density and the dust continuum emission:

N({\rm H_2})\approx N_{\rm H}/2 = 1.1\times10^{21}
\left({S_\nu \over {\rm mJy\ beam^{-1}}}\right) {\rm cm^{-2}},
\end{displaymath} (3)

assuming solar metallicity. Thus the mean 250GHz flux density of 5.2 mJy beam-1averaged over the whole filament corresponds to a column $N({\rm
H}_2)=5.7\times 10^{21}\,{\rm cm}^{-2}$, and the peak flux density of 10.6mJy beam-1 yields $N({\rm H}_2)=11.7\times 10^{21}\,{\rm
cm}^{-2}$. Assuming the core to be a cylindrical filament and not an edge-on sheet, we derive an average volume density $n({\rm
H}_2)=6.2\times 10^4\,{\rm cm}^{-2}$. The total mass of the filament (corrected for the contribution of helium) is 0.66$M_{\odot}$.

Figure 1 illustrates the good correspondence between the dust continuum emission and the integrated C18O line flux over most of the filament. Adopting an excitation temperature of $T_{\rm ex}=10$K the C18O emission peak flux yields a column density $N({\rm
C^{18}O})=1.2\times10^{15}$cm-2. With the previously derived H2 column density, we find an abundance $\rm [C^{18}O]/[H_2]\approx
1\times 10^{-7}$, which is in good agreement with the value previously found by Großmann & Heithausen (1992) in this cloud.

\par\includegraphics[angle=-90,width=8.8cm,clip]{h3322f2.eps}\end{figure} Figure 2: Correlation of the velocity-integrated line intensities of the HC3N (3$\to $2) and (10$\to $9) transitions (in K kms^-1kms-1). The values for HC3N-A are coded as filled squares, those of HC3N-B as open squares. The lines drawn are eye-fits to the data. The error bar represents a typical value for all the data.
Open with DEXTER

\par\includegraphics[width=8.8cm,clip]{h3322f3.eps}\end{figure} Figure 3: Spectra of the HC3N ($J=3\to 2$), (4$\to $3) and (10$\to $9) transitions towards the centre of HC3N-B. In the (3$\to $2) spectrum three hyperfine components are visible: the $F=(3\to 2)$ component at $v_{\rm LSR}=-4.19$ kms^-1kms-1, the $(2\to 1)$ component to the right and the $4\to 3$ to the left. The J=(4$\to $3) transition also shows three hyperfine components: the $F=4\to 3$ component at $v_{\rm LSR}=-4.19$ kms^-1kms-1, the $3\to 2$ component to the right and and the $5\to 4$ to the left.
Open with DEXTER

4 Cyanoacetylene

Because of its low rotational constant cyanoacetylene (HC3N) has many rotational lines which are easily accessible with ground-based radio telescopes. Therefore this molecule has widely been used to determine the excitation conditions in molecular clouds cores. We will use our HC3N observations to derive the gas density in MCLD 123.5+24.9.

Figure 1 compares the intensity distributions of the dust continuum, and the C18O, HC3N (10$\to $9) and (3$\to $2) line emission. For the HC3N (3$\to $2) map the fluxes of the $F=2\to1, 3\to2$ and $4\to 3$ hyperfine components were coadded. The HC3N emission looks very different from that of dust and C18O. HC3N is concentrated in two distinct clumps located at the ends of the filament traced by CO and dust. We shall denote the clump at higher galactic latitudes as HC3N-A, the other one as HC3N-B.

The positions and sizes of the two C18O clumps were determined through a 3-dimensional Gaussian fit (using GAUSSCLUMPS by Stutzki & Güsten 1989) to each hyperfine component. The two clumps are well represented by the Gaussian intensity distribution, suggesting that they are centrally condensed. In Table 2 we list the average properties of the two clumps.

Figures 1 and 4 show that in between both clumps both transitions show only weak emission, lower by a factor 3 to 5 from the flux seen towards the clumps (see Table 3).

In Fig. 2 we show a direct comparison of the HC3N (3$\to $2) and (10$\to $9) integrated intensities. The ratio of the line temperatures is clearly different in the two clumps. An eye-fit to the data points yields approximately $R_{10/3}\equiv W(10\to9)/W(3\to2)=1/5$ in HC3N-A, and R10/3=1/2 in HC3N-B.

Toward the peak intensity positions of the two clumps we also observed the HC3N (4$\to $3) transition, and for HC3N-B we show this spectrum in Fig. 3, and fluxes are listed in Table 3. The three main hyperfine components are clearly detected toward both clumps. The expected ratio is for the components is $(F=5\to4)$:$(F=4\to3)$: $(F=3\to2)=1.00$:0.77:0.58 (Lafferty & Lovas 1978), the observed ratio is 1.0: $(0.8\pm0.1)$: $(0.6\pm0.1)$, i.e. consistent with optically thin emission.

We have computed a large velocity gradient model to match the observed line intensities of the three transitions. Because the kinetic temperature as derived from ammonia observations (Großmann & Heithausen 1992) is below 15K we discuss the results for kinetic temperatures of 10 and 15K. The range of physical parameters reflects the 1$\sigma$ uncertainties of the observed intensities. For $T_{\rm kin}=10$K we find $n_{\rm H_2}\approx (1.1\pm0.5)\times 10^5 \rm
cm^{-3}$ toward HC3N-A, and a factor 2.5 higher density for HC3N-B. The peak column density $N({\rm HC_3N})/\Delta v\approx (1.6\pm0.8)\times
10^{13}\rm cm^{-2}$/kms^-1kms-1 is similar in both clumps. For $T_{\rm kin}=15$K the densities decreases by a factor of 2 but the column density is insensitive to the kinetic temperature.


Table 2: Properties of the HC3N clumps.
Core l b $v_{\rm LSR}$ $\Delta v$ $\Delta l\times \Delta b$ r $n({\rm H_2})$ $N({\rm HC_3N})\over\Delta v$ $M_{\rm cl}$ $M_{\rm vir}$
  [deg] [deg] [ kms^-1kms-1] [ kms^-1kms-1] [arcsec] [pc] [105cm-3] [ $10^{13}\,{{\rm cm}^{-2}\over\ifmmode{{\rm\thinspace km\thinspace s}^{-1}}\else{\thinspace km\thinspace s$^{-1}$ }\fi}$] [$M_{\odot}$] [$M_{\odot}$]
HC3N-A 123.691 24.931 -4.59 0.19 $35\times 33$ 0.025 0.6-1.6 $1.6\pm0.8$ 0.13 0.19
HC3N-B 123.680 24.886 -4.19 0.19 $20\times42$ 0.021 1.6-4.0 $1.6\pm0.8$ 0.19 0.16

Remarks: values are averages of the three hyperfine components. $\Delta v$, $\Delta l$, and $\Delta b$ are deconvolved from the instrumental resolution.


Table 3: HC3N line fluxes (summed over the finestructure components) and their ratios.
Region $\big<W(3\to2)\big>$ $W(4\to3)^a$ $\big<W(10\to9)\big>$ $\Big<{W(10\to9)\over W(4\to3)}\Big>$ $\Big<{W(3\to2)\over W({\rm C^{18}O}\ 1\to0)}\Big>$ $\Big<{W(10\to9)\over W({\rm C^{18}O}\ 1\to0)}\Big>$ Identification
  [K kms^-1kms-1] [K kms^-1kms-1] [K kms^-1kms-1]        
HC3N-A $0.61\pm0.02$ $0.83\pm0.10$ $0.170\pm0.010$ 0.2 $0.69\pm0.05$ $0.20\pm0.02$ NH3-Ab, CS-Ac
HC3N-B $0.57\pm0.02$ $0.90\pm0.06$ $0.333\pm0.006$ 0.5 $0.39\pm0.05$ $0.23\pm0.02$ NH3-Bb, CS-Cc
Intercore $0.12\pm0.02$ - $0.060\pm0.004$ - $0.10\pm0.03$ $0.06\pm0.01$ CS-Bc

Remarks: a: single position; b: Großmann & Heithausen (1992); c: Heithausen (1999).

\par\includegraphics[angle=-90,width=8.6cm,clip]{h3322f4.eps}\end{figure} Figure 4: Cut along a line connecting HC3N-A and B in the integrated HC3N (3$\to $2) (solid line), the C18O (1$\to $0) (dotted line), and the CS (2$\to $1) (dashed line) transitions. Offsets are relative to HC3N-A. The location of the centres of the clumps CS-A, CS-B, and CS-C (Heithausen 1999) are marked by vertical lines.
Open with DEXTER

5 Discussion

5.1 Chemistry

It had been noted previously (Heithausen et al. 1999) that the dense core in MCLD 123.5+24.9 shows a different structure when traced by different molecular emission lines. The most striking difference apparent from our data is that C18O and the dust continuum emission trace a narrow filament, whereas the HC3N emission is concentrated in two clumps, HC3N-A and B, located on both ends of the filament. These clumps coincide with two ammonia emission clumps, NH3-A and B, detected by Großmann & Heithausen (1992). Although the NH3 (1,1) inversion transition was observed with lower (40'') angular resolution, there appears also an emission gap between the NH3 clumps, just as we found for HC3N.

The cross cut shown in Fig. 4 shows that the HC3N clumps somewhat overlap with the CS clumps A and C, but their peak intensity is clearly shifted towards the ends of the CS and C18O filament. The CS clump B has no corresponding HC3N feature.

For a quantitative comparison between the HC3N and C18O emission we computed average intensities for both species toward the HC3N clumps and in the area in between. The ratios listed in Table 3 show that the relative intensity drops by a factor 3 to 7 in between the HC3N clumps.

\par\includegraphics[width=12cm,clip]{h3322f5.eps}\end{figure} Figure 5: Correlation of the line width ( FWHM) and the peak line temperature of the HC3N (10$\to $9) line as derived from a Gaussian fit to the spectra for HC3N-A (right box) and B (left box). The solid line represents a running average for the values binned over $\Delta T=0.1$K. The dotted line marks the thermal line width for a gas temperature $T_{\rm gas}=T_{\rm dust}=13$K. Errorbars are $\pm 1\sigma $.
Open with DEXTER

The strong drop in HC3N intensity relative to C18O may be caused by a change in excitation or abundance. It is unlikely that the excitation conditions vary along the filament, since its gas column density, as traced by the C18O and dust emission, rather increases toward the middle of the filament. A near constant gas density is also implied by the CS (5$\to $4) observations (Heithausen 1999, and new unpublished observations). A drop in the kinetic temperature also seems unlikely, since this would also affect the C18O and dust emission.

Could it therefore be that the relative intensity gradients are due to a lower HC3N and NH3 abundance compared with C18O and CS in the middle of the filament? The chemistry of HC3N has been studied in great detailed, finding that it is an early-time molecule which has its highest abundance after (0.3- $1)\times10^6$ years after the formation of the cloud (e.g. Nejad & Wagenblast 1999; Gwenlan et al. 2000); over the next (0.5- $1.0)\times10^6$ years the HC3N abundance decreases by several orders of magnitudes. CS is the fastest-formed molecule, followed by HC3N, and then by NH3, which is formed at times later than 106 years (Taylor et al. 1998). If we were to interpret the relative abundance gradients as an age effect, the clumps HC3N-A (= NH3-A and CS-A) and HC3N-B (= NH3-B and CS-C) would be the oldest, whereas CS-B must be the younger because it shows no ammonia emission and little cyanoacetylene.

5.2 Stability

The mass of the two HC3N clumps may be estimated from the peak volume density derived from our excitation analysis (Sect. 4, Table 2), and, in agreement with the parameters derived from Gaussclumps listed in Table 2, assuming a Gaussian density profile with a FWHM, r, which we take as the geometric mean of the minor and major axis of the Gaussian intensity distribution. We then find HC3N-A to contain 0.13 $M_{\odot}$, whereas HC3N-B has 0.19 $M_{\odot}$. These masses are similar to those derived from the column density estimate of our bolometer map integrated over the extent of the HC3N clumps.

To assess the clumps' stability we first analyse the HC3N (10$\to $9) line widths. In Fig. 5 for each position we plot the line width against its peak brightness temperature. Because the clumps are centrally concentrated (s. Gaussian analysis above), the peak brightness temperature is also a measure of the distance from the cloud centre. For HC3N-B we find a trend for the spectra closer to the clump center to show smaller line widths, but for HC3N-A no such correlation appears.

The clump mass may now be compared with the virial mass,

M_{\rm vir}={5\sigma^2 r ~/~ G}
\end{displaymath} (4)

(Bertoldi & McKee 1992), where $\sigma=\Delta
v/2.355$ is the one-dimensional gas velocity dispersion and G is the gravitational constant. With the values listed in Table 2, the virial masses turn out to be very close to the masses we computed from the line and continuum emission, indicating that the clumps are gravitationally bound.

6 Conclusions

We have imaged the dust emission and HC3N line emission in the cirrus cloud MCLD 123.5+24.9 and find that most of the HC3N emission arises from two condensations at the ends of an elongated filament which is traced by the dust continuum and by C18O and CS line emission. The HC3N emission does correlate with previously reported NH3emission, but does not correlate with the other tracers, suggesting either excitation or relative abundance gradients. We suggest that abundance gradients are likely to be responsible for the appearance, and that the abundance differences are due to chemical evolution. This would suggest that the HC3N clumps are older than the center of the filament, which would be less than $3\times 10^5$ years old and there has been no time for either the HC3N or the NH3 to achieve a large abundance. Based on our HC3N observations we find that the older clumps are gravitationally bound, and as signs for inward motion in one of them show (Heithausen 1999), they might be near gravitational collapse.

The detection of graviationally bound structures in cirrus clouds is unexpected, since on the large scale these clouds, unlike star-forming giant molecular clouds, are not gravitationally bound. Their turbulent kinetic energy is typically ten to one hundred times larger than their self-gravitational energy (Magnani et al. 1987; Heithausen 1996). As our observations show, even such cirrus clouds may contain substructures where gravity dominates, and which may be able to collapse to form stars.

We thank Peter Schilke for providing the LVG code used in this study. Many thanks to E. Kreysa and the MPIfR bolometer group for providing MAMBO, and to R. Zylka for the MOPSI data reduction package. This work was supported by the Deutsche Forschungsgemeinschaft grant SFB-494, and it made use of the IRAM 30 m telescope. IRAM is supported by INSU/CNRS (France), MPG (Germany), and IGN (Spain).



Copyright ESO 2002