FIRST DETECTION OF AMMONIA IN THE LARGE MAGELLANIC CLOUD: THE KINETIC TEMPERATURE OF DENSE MOLECULAR CORES IN N 159 W

Assuming the optically thin case (see above), the column densities of the upper ammonia levels of the metastable (J, J) inversion doublets, N_u(J, J) (in units of cm⁻²), can be derived using

$$\begin{equation} N_{u}(J,J)=\frac{7.77\times 10^{13}}{\nu }\frac{J(J+1)}{J^{2}}\,\,\,\int T_{\rm mb}\,\, dv \end{equation} \tag{ 1 }$$

(e.g., Henkel et al. 2000) with T_mb denoting the main beam brightness temperature of the (J, J) inversion line in Kelvin, v being the velocity in km s⁻¹, and ν representing the frequency in GHz. Our values for N 159 W are listed in Table 2. For a single ortho- or para-ammonia species (the observed lines in N 159 W all belong to para-ammonia) the Boltzmann distribution for a rotational temperature $T_{\rm rot,\it JJ^{\prime }}$ (in K) is described by

$$\begin{equation} \frac{N_{u}(J^{\prime },J^{\prime })}{N_{u}(J,J)}=\frac{2J^{\prime }+1}{2J+1}\,\exp \left(\frac{-\Delta E}{T_{{\rm rot},JJ^{\prime }}}\right). \end{equation} \tag{ 2 }$$

Here, the energy difference between the (J, J) and (J', J') levels is given by ΔE (41.2 K between NH₃ (1,1) and (2,2)). We derive a rotational temperature T_rot,12 of (17 ±  2) K for the high-velocity NH₃ component and a very similar (15 ± 5) K for the low-velocity component (uncertainties are based on the ±1σ values for N_u derived in Equation (1), not quadratically propagated but using the extremes). Rotational temperatures are always lower limits to the kinetic gas temperatures. However, for temperatures ≲20 K the differences between T_kin and T_rot,12 are less than 10% and therefore negligible (e.g., Walmsley & Ungerechts 1983; Danby et al. 1988). We can safely assume T_kin = T_rot,12 for both components of N 159 W.

Thermal line widths of gas at such cold kinetic temperatures are ∼0.2 km s⁻¹. The measured NH₃ line widths are at least an order of magnitude larger than this figure and deconvolving the lines by the instrumental velocity resolution and the applied boxcar smoothing is not sufficient to reach the predicted thermal widths. The gas kinematics must thus be dominated by other processes such as turbulence or by systematic motions of spatially unresolved cloud components. This is supported by the line widths of other molecules which are similar to those of NH₃ (e.g., Heikkilä et al. 1999).

The kinetic temperature of the gas traced by NH₃ is colder than the 30–40 K dust temperature that is measured in N 159 W (Heikkilä et al. 1999; Bolatto et al. 2000). In their LVG analysis of high excitation molecular gas, Bolatto et al. (2005) establish a two-component model for N 159 W, which comprises a gas phase with ∼20 K kinetic temperature and a volume density of ∼10⁵ cm⁻³, as well as a second phase with ∼100 K and ∼10² cm⁻². Likely, the detected ammonia traces the low temperature component, albeit at larger volume densities (for volume density estimates based on NH₃, see Section 4.2). This could be in the form of cold cores surrounded by a warmer envelope. As mentioned above, dust temperatures in the region have been estimated to be ∼2–3 times that of NH₃ and it is likely that the dust is not only present in the cold cores but also occupies the interface to the warm envelope, covering a rather wide range of densities.

The detailed dust and gas morphologies have implications for the shielding against UV radiation. The NH₃ molecule is quite susceptible to photo-dissociation (Suto & Lee 1983). N 159 W itself is a rather strong source of ionizing photons (e.g., Jones et al. 2005, also see Section 4.3) and a smaller contribution comes from the most luminous star-forming region in the Local Group, 30 Doradus, which is about ∼600 pc away from N 159 W (see e.g., Pineda et al. 2009). The flux of energetic UV photons may penetrate into the envelope and destroy ammonia. It requires dense cores to provide enough shielding to keep ammonia intact and it is therefore likely that the molecule resides within such dense, cold, compact cores. In addition, unlike CO, nitrogen-bearing species such as NH₃ and N₂H⁺ are less prone to depletion on dust grains at densities typical for cold dense cores such as protostellar cores (>10⁵ cm⁻³; e.g., Tafalla et al. 2004; Flower et al. 2005; for a density estimate of N 159 W see Section 4.2) and NH₃ is expected to be abundant in such objects.

For a given single rotational temperature, total NH₃ column densities, $N({\rm NH_{3}}, \rm total)$, can be derived by extrapolating the level distribution over all energies. Ungerechts et al. (1986) provide this extrapolation in their Equation (A15), which only requires T_rot,12 and N_u(1, 1) to be determined. For N 159 W, we derive a total, 17'' beam-averaged, ammonia column density of (5.8 ± 0.6) × 10¹² cm⁻². Toward the same position, the integrated CO(1 → 0) intensity is ∼45 K km s⁻¹ over a 45'' beam (Mopra data taken from Ott et al. 2008; see also Figure 1). Assuming a CO to H₂ conversion factor of X_CO ∼ 4 × 10²⁰ cm⁻² (K km s⁻¹)⁻¹ for the LMC (Pineda et al. 2009), the H₂ column density can be estimated to be ∼1.8 × 10²² cm⁻². With this value, the fractional ammonia abundance in the N 159 W complex becomes ∼3.3 × 10⁻¹⁰. This number is based on a 45'' average of the H₂ column density. The H₂ distribution within that area, however, is most likely not smooth but exhibits some variation. To excite NH₃, a critical density needs to be exceeded. We do not observe NH₃ outside the smaller, 17'' ATCA beam, which indicates that the H₂ density is likely higher at the position of the NH₃ emission than further away from it. Consequently, the ammonia abundance of ∼3.3 × 10⁻¹⁰ may be considered an upper limit and any H₂ column exceeding the average over 45'' will dilute the ammonia abundance. Furthermore, larger X_CO factors, such as reported by Israel et al. (2003) or Fukui et al. (2008), would also drive the fractional ammonia abundance to smaller values.

The X_CO factor is just one way to estimate the total H₂ content of the molecular cloud. A different approach is to derive virial masses M_vir, assuming that the system is virialized and the masses of the clouds are dominated by H₂. For a cloud with a Gaussian density profile, the virial mass can be estimated using

$$\begin{equation} M_{\rm vir}=444\,r\,\Delta v_{\rm 1/2}^{2} \end{equation} \tag{ 3 }$$

(Protheroe et al. 2008) with the radius r in pc, the FWHM velocity Δv_1/2 in km s⁻¹, and the virial mass M_vir in M_☉ (note that the numerical factor changes with the shape of the radial density profile; e.g., it is ∼2 times lower for $\rho =\rm constant$, and ∼3.5 times lower for a ρ ∝ r⁻² profile; MacLaren et al. 1988). The NH₃ emission position angle may show some deviation from a pure point source but, given the faintness of the source, we use half of the FWHM ATCA beam size as an upper limit to the radius of the cloud, 8'' which corresponds to ∼2 pc. Together with Δv_1/2 ∼ 5 km s⁻¹, the virial mass amounts to an upper limit of <2.2 × 10⁴ M_☉ (for comparison, the virial mass based on the Mopra CO data over the much larger Mopra beam would be ∼10⁵ M_☉). This limit also accounts for the boxcar smoothing of the data; a more narrow line would lower the upper limit and our estimate is therefore a more conservative one. In addition, this estimate also accounts for material other than H₂ that is contributing to the virial mass, justifying the upper limit nature of the H₂ mass based on the virial method.

Using a spherical geometry of the NH₃ emission and assuming that all the virial mass is due to H₂, the mean volume density of the molecular gas then can be estimated to be ≲1.3 × 10⁴ cm⁻³. This value is fairly typical for dense gas in Galactic star-forming regions, spread over the physical size of the beam, ∼4 pc (e.g., Foster et al. 2009). Finally, if we take the column density of ammonia of (5.8 ± 0.6) × 10¹² cm⁻² (see above), we derive a beam averaged ammonia mass of ∼1 × 10⁻⁵ M_☉, or ∼3.3 M_⊕. With H₂ masses and densities being upper limits, the fractional abundance of ammonia (molecule number density ratio) then follows as a lower limit of ≳4.5 × 10⁻¹⁰. This value is in agreement with the upper limit derived via X_CO above.

We cannot infer any robust variation in fractional ammonia abundance between the two velocity components. The CO(1→0) emission is about 2 orders of magnitude brighter (cf. Figure 2) at the velocity of the ammonia high-velocity component. The ratio appears to be somewhat smaller for the low-velocity component, which would indicate some variation in the abundance. But, given the uncertainties in beam filling factors for the CO and NH₃ observations (see also Table 2), better data are needed to reliably establish such a variation.

Using the above values for temperature and upper volume density limit for N 159 W, we estimate the pressure of the clumps in the N 159 W region to be P/k ≳ 2 × 10⁵ K cm⁻³. This value is likely to increase substantially when the average value for H₂ densities is resolved into individual, smaller clumps with relatively little H₂ in between. For comparison, Wong et al. (2009) calculate the midplane pressure for the positions where molecular gas is observed in the LMC. Their highest value is close to our lower limit, which may indicate that the dense NH₃ clumps are not in pressure equilibrium with their surroundings, a property that is typical for self-gravitating clouds. We like to note though that the volume density of our calculation is actually based on the virial theorem, i.e., a gravitational equilibrium of the molecular gas that is independent of its environment.

We can now revisit the upper limits of the ammonia survey toward the other six positions. If we assume the same line width and kinetic temperature that we find for the two components of N 159 W, the 3σ upper limits to the non-detections translate to ammonia columns N_u(1, 1) and $N ({\rm NH_{3}}, \rm total)$ of <5 × 10¹² cm⁻², and <1.5 × 10¹³ cm⁻² for a 9'' sized region, respectively.

If we take the upper limit to the virial mass of the ammonia emitting region of ∼2 × 10⁴ M_☉ for the mass of the molecular clump (see Section 4.2 above), the typical SF efficiency of a few percent (Evans et al. 2009) would convert the gas to stellar masses of ∼10³ M_☉. This falls within the mass range of open stellar clusters. To be independent of extinction effects (the very nearby ionizing source N 159 W AN is deeply enshrouded by dust as shown by Jones et al. 2005), we use the 1.4 GHz radio continuum map presented in Hughes et al. (2007) to estimate the current SF rate (SFR) via the conversions provided by Haarsma et al. (2000). For N 159 W AN, we measure a flux of ∼260 mJy which corresponds to a current SFR of ∼8 × 10⁻⁴ M_☉ yr⁻¹. The radio emission at that position, however, cannot be entirely separated from other nearby sources and we estimate an uncertainty to the flux and the SFR of up to 50%. If SF continues at that level, the available molecular gas will be fully converted into stars within the next ∼10 Myr.

The LMC has a lower mass and metallicity than the Galaxy. In addition, the N 159 W region is located in the molecular ridge south of the extremely active 30 Doradus star-forming region, at a location where the LMC's regular disk becomes disturbed by tidal and ram pressure forces that act as the galaxy moves through the Milky Way halo (e.g., Ott et al. 2008). Do these peculiarities change the properties of the molecular clumps formed at this location? In order to address this question, we compare our results to the properties of dense molecular cores in the Galaxy. A homogenized sample of molecular cores, drawn from Galactic star-forming regions such as Orion, Ophiuchus, Perseus, Taurus, Serpens, Cepheus, etc. is presented in Jijina et al. (1999). The authors have collected ammonia (1,1) and (2,2) observations for all of their sample and can therefore provide data that may act as a reference for comparison with clouds in other galaxies such as the LMC. More recently, Rosolowsky et al. (2008; see also Suzuki et al. 1992; Foster et al. 2009) published a comprehensive ammonia survey of 193 dense molecular cores toward the Perseus molecular cloud, a cloud that may also be representative for typical star-forming regions in the Galaxy. The studies agree that Galactic dense molecular cores exhibit typical kinetic gas temperatures of ∼10 K, with sizes of ∼0.1 pc, velocity widths of ∼0.1–0.7 km s⁻¹, ammonia column densities of ∼10^14.5 cm⁻², and virial masses of ∼5 M_☉. In most aspects these values are clearly different from what we measure toward the LMC cloud N 159 W. The difference can be explained by observing an ensemble of cores within the ATCA beam. Such a site could potentially be the birth of an entire stellar cluster (see also Sections 4.2 and 4.3). Along these lines, the N 159 W region may consist of individual, dense molecular cores that are at temperatures very similar to those found in the Galaxy along with some inter-clump material. Together, the line profiles of the clumps and the more diffuse gas blend to form the measured ammonia lines with a width of ∼5 km s⁻¹ (note that Rosolowsky et al. 2008 also largely attribute the upper end of their measured line widths to blending of a number of cores along the line of sight). We would like to note that the N 159 W SF timescales as derived in Section 4.3 also match those that are typical in Galactic SF regions (Evans et al. 2009).

The differences in ammonia column densities and fractional abundances between Galactic cores and the emission in N 159 W, however, cannot be reconciled by beam filling arguments. The NH₃ column densities in N 159 W are about 50 times lower (log(N[NH₃]) ∼ 12.8 for N 159 W, ∼14.5 for Galactic SF regions). If beam filling would be responsible for the difference, the ATCA beam would be diluted by about the same factor. This implies that of the 20 pc² ATCA beam area only 0.4 pc² is covered by clouds. If these clouds are similar to Galactic cores (with typical sizes of 0.1 pc and masses of ∼5 M_☉, see above), the coverage corresponds to ∼50 molecular cores adding up to a total mass of ∼250 M_☉. In Section 4.2, we derive an upper limit of ∼2 × 10⁴ M_☉ to the virial mass in N 159 W, which is about 2 orders of magnitude larger. This indicates that the basic assumption for the calculation, the equality of Galactic and N 159 W ammonia column density, is wrong.

The virial mass is an upper limit due to unknown mass components other than H₂ within the beam and due to the fact that the ATCA beam was taken as the virial radius of the unresolved emission. Can the parameters play together to create the 250 M_☉ mass as predicted from Galactic clumps? We measure a velocity width of 5 km s⁻¹ within N 159 W and, according to Equation (3), this determines the size of the gaseous emission for the 250 M_☉ to be 0.05 pc if the clouds are in virial equilibrium. Such a small value is clearly unrealistic and we conclude that it is not possible to reconcile the ammonia column densities with those of Galactic SF regions based on beam filling arguments alone. It is likely that intra-clump gas covers a larger region of the beam and/or a larger number of cores are present in N 159 W. But this implies that the ammonia column densities in N 159 W are intrinsically lower than those in Galactic cores.

Similarly, the low fractional abundances of ammonia in N 159 W cannot be fully explained by clumping of the gas within the beam (note that the abundances are independent of the beam size when applying the virial mass as in Section 4.2). Benson & Myers (1983) derive for Galactic dark clouds on average fractional ammonia abundances of ∼10⁻⁷ and Mauersberger et al. (1987) obtain ∼10⁻⁵ for a hot core. The resolved Galactic clumps as listed in the more recent Rosolowsky et al. (2008), Foster et al. (2009), and Jijina et al. (1999) studies have NH₃ abundances of order ∼10⁻⁸. In contrast, we estimate a fractional ammonia abundance in N 159 W of ∼4 × 10⁻¹⁰ (see Section 4.2). N 159 W thus exhibits at least 1.5 orders of magnitude lower fractional ammonia abundance than seen in individual cores embedded in Galactic clouds. But N 159 W is not an exception: the upper limits toward the other LMC star-forming regions in our survey exhibit a limit of at least 1 order of magnitude below the Galactic ammonia abundances. Such a difference cannot be an effect of the overall lower metallicity of the LMC alone, which is ∼45% solar (Luck et al. 1998; Rolleston et al. 2002, based mostly on iron, carbon, oxygen, and silicon). A combination of the ∼5 times lower abundance of nitrogen in the LMC, ∼10% solar (Hunter et al. 2007), and the high UV flux at the N 159 W location (see Section 4.1 and also Deharveng & Caplan 1992; Bolatto et al. 2000; Pineda et al. 2009) that can potentially photo-dissociate the fragile NH₃ (Suto & Lee 1983) must therefore contribute to its low fractional abundance.

We cannot exclude that the fraction of clump to inter-clump NH₃ emission might be very different in N 159 W as compared to Galactic SF regions. In such a scenario, hardly any of the molecular gas parameters can be reconciled between the two galaxies. Beam dilution effects can explain most of the differences, except for the ammonia abundance and column density, and is therefore a more natural interpretation. To fully resolve this issue, sensitive observations at better spatial resolution are required.

Among extragalactic objects, only M 82 is known to exhibit a similarly low fractional ammonia abundance like N 159 W. Weiß et al. (2001a) derive a value of ∼5 × 10⁻¹⁰ for this prominent starburst galaxy. The gas in M 82, at T_kin ≈ 60 K, is warmer than in N 159 W and the metallicity of M 82 may also exceed that of the LMC, in particular close to the central starburst region (see, e.g., McLeod et al. 1993; Martin 1997). The presence of nearby, strongly ionizing sources in both M 82 and N 159 W suggests that photo-dissociation plays an important role in regulating the ammonia abundances in both galaxies.