DENSE MOLECULAR CLUMPS ASSOCIATED WITH THE LARGE MAGELLANIC CLOUD SUPERGIANT SHELLS LMC 4 AND LMC 5

Observations of the ¹²CO(J = 3–2) transition toward the N48 and N49 regions were made with the ASTE 10 m telescope at Pampa la Bola in Chile (Ezawa et al. 2004), in 2006 September and 2011 September, respectively. The half-power beam width was measured to be 22'' at 345 GHz by observing the planets. Both observations were performed in the on-the-fly (OTF) mapping mode.

In 2006, we observed the giant molecular cloud LMC/M5263–6606 (Mizuno et al. 2001) in the N49 region. The size of the ¹²CO(J = 3–2) mapping area was 30 × 55 (45 × 83 pc), which included the whole cloud. Along each row of an OTF field, individual spectra were recorded every 15 and the spacing between the rows was 6'', so that the 22'' beam of the telescope was oversampled. The OTF mapping was performed along two orthogonal directions, i.e., scans along the right ascension and declination directions. In this period, we used a single cartridge-type double-sideband (DSB) SIS receiver, SC345 (Kohno 2005). The spectrometer was an XF-type digital autocorrelator MAC (Sorai et al. 2000) and was used in the wide-band mode, which has a bandwidth of 512 MHz with 1024 channels. The spectrometer provided velocity coverage and channel spacing of 450 and 0.44 km s⁻¹ at 345 GHz, respectively. The chopper-wheel technique was employed to calibrate the antenna temperature $T_{\rm a}^{\ast }$. The typical system noise temperature during the observation was ∼500 K in DSB including atmospheric effects. The pointing error was measured to be within 5'' (peak to peak) by observing the CO point sources R Dor (α_B1950 = 4^h36^m4584, δ_B1950 = −62°04'3570) or o Cet (α_B1950 = 2^h19^m208, δ_B1950 = −2°58'4070) every 2 hr during the observing period. We observed N159W (α_B1950 = 5^h40^m37, δ_B1950 = −69°47'000) every 2 hr to check the stability of the intensity calibration. The average and the standard deviation of the antenna temperature $T_{\rm a}^{\ast }$ of N159W was 5.48 ± 1.42 K, i.e., the intensity variation during these observations was estimated to be less than 26%. We scaled the observational data taken in 2006 to T_mb scale with a scaling factor of 2.538 so that the average value of observed $T_{\rm a}^{\ast }$ of N159W is consistent with the main beam temperature of N159W T_mb = 13.9 ± 0.7 K, which was estimated by Minamidani et al. (2011).

In 2011, we observed the giant molecular cloud LMC/M5253–6618 (Mizuno et al. 2001) in the N48 region. The cloud was covered by two 7' × 7' (105 × 105 pc) square OTF maps. Along each row of an OTF field, individual spectra were recorded every 16 and the spacing between the rows is 75, so that the 22'' beam of the telescope is oversampled. The OTF mapping was performed both along the right ascension and declination directions, respectively. In this period, we used the waveguide-type sideband-separating SIS mixer receiver for single-sideband (SSB) operation, CATS345 (Ezawa et al. 2008; Inoue et al. 2008). The image rejection ratio at 345 GHz was estimated to be ∼10 dB. The spectrometer was an XF-type digital autocorrelator MAC (Sorai et al. 2000) and was used in the high-resolution mode, which has a bandwidth of 128 MHz with 1024 channels. The velocity coverage and the channel spacing at 345 GHz were 125 and 0.11 km s⁻¹, respectively. Typical system noise temperatures during the observation were 500 K in SSB including atmospheric effects. The pointing error was measured to be within 5'' by observing a CO point source R Dor every 2 hr. We observed N159W every 2 hr to check stability, and the average and standard deviation of the antenna temperature was found to be 7.58 ± 0.26 K on a $T_{\rm a}^{\ast }$ scale. The intensity variation was therefore estimated to be less than 3%. We scaled the observational data in 2011 to T_mb scale with scaling factor of 1.835.

The data reduction was made using the software package NOSTAR, which comprises tools for OTF data analysis, developed by the National Astronomical Observatory of Japan (Sawada et al. 2008). Linear baselines were subtracted from the spectra. The raw data were re-gridded to 10'' per pixel, giving an effective spatial resolution of approximately 27'', which corresponds to 7 pc at the distance of the LMC. The data sets taken along the right ascension and declination directions were co-added by the basket-weave method (Emerson & Graeve 1988) to remove any effects of scanning noise. A fifth-order polynomial function was then fitted to the baseline and subtracted in order to reduce the effects of baseline ripples. The 2011 data was binned to a channel spacing of 0.44 km s⁻¹, which corresponds to the channel spacing of the MAC wide-band mode used in the 2006 data.

We performed observations of the ¹²CO(J = 1–0) and ¹³CO(J = 1–0) transitions using the 22 m Australia Telescope National Facility (ATNF) Mopra telescope in 2012 June/July. We observed eleven 2' × 2' areas which covered the prominent parts of the ¹²CO(J = 3–2) clumps. The half-power beam width was 33'' at the 3 mm band, and the observations were performed in the OTF mapping mode. Individual spectra were recorded every 6'' and the spacing between rows is 9'', so that the 33'' (FWHM) telescope beam was oversampled.

We used the 3 mm MMIC receiver that can simultaneously record dual polarization data. The spectrometer was the Mopra Spectrometer (MOPS) digital filter bank and was used in the zoom-band mode, which can record up to sixteen 137.5 MHz zoom bands positioned within an 8 GHz window. The spectrometer provided a velocity coverage and channel spacing of 376 km s⁻¹ and 0.09 km s⁻¹ at 3 mm, respectively. Typical system noise temperatures during the observations were 600 K at the frequency of the¹²CO line, and 250 K at the frequency of the ¹³CO line, including atmospheric effects. The pointing error was measured to be within 10'' by observing the SiO maser toward R Dor (α_J2000 = 4^h36^m4561, δ_J2000 = −62°04'3792) every 1–2 hr during the observing period. We observed Orion KL (α_J2000 = 5^h35^m145, δ_J2000 = −5°22'2956) once a day to check the stability of the intensity calibration. The average and the standard deviation of Orion KL antenna temperatures was 47.9 ± 3.1 K in $T_{\rm a}^{\ast }$ scale, then the intensity variation during the observation was estimated to be less than 7%. The efficiencies we have assumed for CO are the "extended beam efficiency" η_xb discussed by Ladd et al. (2005), which includes the effect of coupling to the inner error beam for sources larger than ∼2'. We determined η_xb = 0.48 by dividing the observed the peak antenna temperature by 100 K, which is the corrected peak CO main beam temperature for Orion KL given by Ladd et al. (2005). In this paper, data reduction was performed using the ATNF's Livedata and Gridzilla software packages. Livedata performs bandpass calibration using off-source spectra, then fits and subtracts a linear baseline. Gridzilla takes the spectra and grids them onto a data cube using a Gaussian smoothing kernel of FWHM 33'', comparable to that of the Mopra primary beam. We weighted the spectra by the inverse of the system temperature when gridding. The resulting effective spatial resolution is approximately 45'', which corresponds to 11 pc at the distance of the LMC. Then we fit and subtracted a fifth-order polynomial function from the baseline to reduce effects of baseline ripples. The channel width was binned up to 0.44 km s⁻¹, which matches the channel spacing of ASTE data. Finally, we re-gridded to a right ascension and declination grid with a spacing of 10'' (to match as the ASTE data) using the ATNF's MIRIAD software. All parameters of the ASTE and Mopra observations are summarized in Table 1.

Table 1. Observation Parameters

Parameters	Values
	12CO(J = 3–2)		12CO(J = 1–0),13CO(J = 1–0)
Observation period	2006 Nov	2011 Nov	2012 Jul
Telescope	ASTE	ASTE	Mopra
Diameter	10 m	10 m	22 m
Rest frequency	345.79599 GHz	345.79599 GHz	115.27120 GHz (12CO)
			110.20135 GHz (13CO)
Beam size (HPBW)	22''	22''	33''
Spectrometer	MAC	MAC	MOPS
Channel number	1024	1024	4096
Bandwidth	512 MHz (440 km s−1)	128 MHz (125 km s−1)	138 MHz (376 km s−1)
Channel spacing	0.5 MHz (0.44 km s−1)	0.125 MHz (0.1 km s−1)	0.03 MHz (0.09 km s−1)
Map description
Mapping mode	On-the-fly	On-the-fly	On-the-fly
Target region	N49	N48	N48, N49/each clump
Field coverage	3' × 55	7' × 7' × 2 fields	2' × 2' × 11 fields
Effective beam width	27''	27''	45''
rms noise level (Δv = 0.44 km s−1)	0.4 K	0.3 K	0.1 K(12CO)
			0.04 K(13CO)

Download table as:  ASCIITypeset image

Figure 2(a) shows a ¹²CO(J = 3–2) integrated intensity map of the N48 and N49 regions obtained with ASTE. Typical rms noise levels are 0.68 K km s⁻¹ and 0.40 K km s⁻¹ in the N49 and N48 region respectively. Minamidani et al. (2008) have also performed similar resolution and sensitivity observations with ASTE toward GMCs in the LMC, which are located in the molecular ridge extending southward from 30 Doradus, and the "CO Arc" along the southeastern optical edge. Compared to these GMCs, the distribution of ¹²CO(J = 3–2) clouds in the N48 and N49 regions tends to be more clumpy; that is each dense molecular clump tends to be independently distributed and not be surrounded by a diffuse gas envelope.

Zoom In Zoom Out Reset image size

Figure 2(b) shows a ¹²CO(J = 1–0) integrated intensity map of the N48 and N49 regions obtained with Mopra. Typical rms noise levels are 0.24 K km s⁻¹ in the N49 region, and 0.15 K km s⁻¹ (¹²CO) in the N48 region. Figure 2(c) shows the equivalent ¹³CO(J = 1–0) map. Typical rms noise levels are 0.16 K km s⁻¹ in the N49 region, and 0.068 K km s⁻¹ in the N48 region. Although the spatial resolution is lower than the ASTE observations, the ¹³CO(J = 1–0) distribution traces well the bright features seen in ¹²CO(J = 3–2), whereas the ¹²CO(J = 1–0) distribution is more extended.

In Figure 3, typical peak ¹²CO(J = 3–2), ¹²CO(J = 1–0) and ¹³CO(J = 1–0) line profiles are presented. The positions selected are local ¹²CO(J = 3–2) integrated intensity peaks (see Section 3.3 and Table 2). The upper panels, Figures 3(a)–(c), show the ¹²CO(J = 3–2) profiles. These illustrate the typical noise level of the data, ∼0.4 K rms at 0.44 km s⁻¹ channel width. The ¹²CO(J = 3–2) intensity is brightest at N48-1 (T_mb ∼ 5.5 K). The line profiles are in general single peaked, but some of them show two clearly separated components, as illustrated in Figure 3(c). Figures 3(d)–(f) show the ¹²CO(J = 1–0) and ¹³CO(J = 1–0) profiles toward the same positions. The peak velocities and line widths of ¹³CO(J = 1–0) are similar to those of ¹²CO(J = 1–0). Figures 3(g)–(i) show the ¹²CO(J = 1–0) and ¹²CO(J = 3–2) spectra together, in which the latter have been extracted from data smoothed to the same effective resolution as the Mopra data (∼45''). The shapes of the ¹²CO(J = 3–2) profiles are similar to ¹²CO(J = 1–0), and nearly the same as ¹³CO(J = 1–0). This indicates that these lines trace similar parts of the clouds.

Zoom In Zoom Out Reset image size

Figures 4 and 5 show position–velocity diagrams of the ¹²CO(J = 3–2) emission in both right ascension and declination, i.e., the cube collapsed along each spatial axis. In these diagrams, the velocity pixel width were bound up to 0.88 km s⁻¹.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the N49 region, ¹²CO(J = 3–2) emission is distributed between ∼283 and 290 km s⁻¹. The overall velocity structure of the cloud shows moderate velocity gradients. There is a compact second component around (α_J2000, δ_J2000) ∼ (5^h26^m20^s, −66°2'40''), and V_LSR ∼ 292 km s⁻¹, which is defined as clump N49-3 (see Section 3.3). In the N48 region, ¹²CO(J = 3–2) emission is distributed between ∼277 and 300 km s⁻¹, and shows relatively clumpy structure. The overall velocity structure of the cloud has velocity gradients in both right ascension and declination. There is an isolated component around (α_J2000, δ_J2000) ∼ (5^h26^m26^s, −66°10'24''), and V_lsr ∼ 298 km s⁻¹ (N48-3). The south clump located around δ_J2000 ∼ −66°16'30'', is broken into several velocity components (N48-6, 8, and 9).

To discuss the physical properties of dense molecular clumps, we define ¹²CO(J = 3–2) clumps in the way described in Minamidani et al. (2008). The clump identification criteria are as follows: (1) identify local peaks in the integrated intensity map that are greater than the 10σ noise level (⩾4.0 K km s⁻¹ for the N48 region, and ⩾6.8 K km s⁻¹ for the N49 region). (2) Select those local peaks that have a peak brightness temperatures greater than the 3σ noise level of the spectrum (typically ∼0.90 K for the N48 region, and ∼1.0 K for the N49 region), then draw a contour at one-half of the peak integrated intensity level and identify it as a clump unless it contains other local peaks. (3) When there are other local peaks inside the contour, draw new contours at the 70% level of each integrated intensity peak. Then, identify clumps separately if their contours do not contain another local peaks (the boundary is taken at the minimum integrated intensity between the peaks), or else identify a clump using the contour of the highest peak as a clump boundary. (4) If a spectrum has multiple velocity components with a separation of more than 5 km s⁻¹, identify those components as being associated with different clumps.

Using these criteria, we identify 18 clumps in the N48 region and 3 clumps in the N49 region. The observed line parameters at the local peaks are shown in Table 2. The parameters of the identified clumps are listed in Table 3. Deconvolved clump sizes, R_deconv, are defined as $[R^2 _{\rm nodeconv} - (\theta _{\rm HPBW}/2) ^2 ]^{0.5}$. R_nodeconv is the effective radius defined as (A/π)^0.5, where A is the observed total cloud surface area. V_LSR and the composite line width, ΔV_clump, are derived using a single Gaussian fit to the spectrum obtained by averaging all spectra within a single clump.

The virial mass is estimated as

$$\begin{equation} M_{\rm vir} ({M}_{\odot }) = 190\; \Delta V_{\rm clump}^{2}({\rm km\;s^{-1}})\; R({\rm pc}), \end{equation} \tag{ 1 }$$

which assumes that the clumps are spherical with density profiles of ρ∝r⁻¹, where ρ is the number density and r is the distance from the cloud center (MacLaren et al. 1988).

The ¹²CO(J = 3–2) luminosity of the cloud L_{CO(J = 3 − 2)} is the integrated flux scaled by the square of the distance,

$$\begin{equation} L_{\rm CO}\; ({\rm K\;km\;s}^{-1}\;{\rm pc}^2) = D^2 \left(\sum T_{i} \right) \delta v \delta x \delta y, \end{equation} \tag{ 2 }$$

where T_i is the ¹²CO(J = 3–2) brightness temperature of an individual voxel, D is the distance to the LMC in parsecs (taken to be 5 × 10⁴), δx and δy are the angular pixel dimensions in radians, and δv is the width of one channel in km s⁻¹.

Histograms of the ¹²CO(J = 3–2) clump physical properties are presented in Figure 6. Mean values in the N48 and N49 regions are clump sizes of R_deconv ∼ 4.7 pc, line widths of ΔV_clump ∼ 4.3 km s⁻¹, virial masses of M_vir ∼ 2.0 × 10⁴ M_☉, and ¹²CO(J = 3–2) luminosities of L_{CO(J = 3 − 2)} ∼ 1.2 × 10³ K km s⁻¹ pc². Note that the smallest clump identified by the procedure above is detected in seven spatial pixels, and its deconvolved radius is 1.2 pc. For comparison, the physical properties of ¹²CO(J = 3–2) clumps identified in Minamidani et al. (2008) are also shown. These clumps are identified in GMCs located in the more active star forming regions of the LMC (30 Dor, N159, N166, N171, N206, N206D, GMC225; in the molecular ridge and the CO Arc), and the mass of these GMCs ranges between 0.4 and 10 × 10⁶ M_☉ (30 Dor ∼0.7 × 10⁶ M_☉, N159 and N171 ∼10 × 10⁶ M_☉, N166 ∼2 × 10⁶ M_☉, N206 ∼0.9 × 10⁶ M_☉, N206D ∼0.4 × 10⁶ M_☉, GMC225 ∼1 × 10⁶ M_☉; Fukui et al. 2008). Although the mass of the GMC in the N48/N49 region (∼1.5 × 10⁶ M_☉) is comparable to these GMCs, the physical properties of the clumps in the N48/N49 region (size, linewidth, virial mass and luminosity) all tend to be smaller, and the number of identified clumps in the N48 region is notably large (18 for N48, 3 for N49, 5 for 30 Dor, 16 for N159 and N171, 5 for N166, 2 for N206, 1 for N206D, and 3 for GMC225). This suggests that the GMC in the N48/N49 region consists of a concentration of compact clumps—a fact which was also indicated by the very clumpy distribution of the ¹²CO(J = 3–2) emission.

Zoom In Zoom Out Reset image size

Since ¹³CO is assumed to be optically thin for a typical molecular cloud, column density can be obtained from ¹³CO luminosity under the assumption of local thermodynamic equilibrium. This gives a better estimate of clump mass than with ¹²CO, which is typically optically thick. In order to compare luminosity-based mass with the virial mass, we also derive the physical properties of the clumps using the ¹³CO(J = 1–0) transition. However, the signal-to-noise ratio of ¹³CO(J = 1–0) is not sufficient to identify all clumps via step 1 of the clump identification method described above, i.e., the peak integrated intensity of several clumps is weaker than the 10σ noise level. We instead use the ¹²CO(J = 3–2) clump boundaries to derive the ¹³CO(J = 1–0) clump properties, supposing that the ¹³CO(J = 1–0) distribution is similar to that of ¹²CO(J = 3–2). Since the spatial distributions and the line profiles of the two transitions are similar (Figures 2(a), (b), and 3), this is likely to be a reasonable assumption. We identify those clumps with ¹³CO(J = 1–0) spectral peaks greater than the 3σ noise level at the ¹²CO(J = 3–2) local peak positions. The selected clumps are N48-1, 2, 3, 4, 5, 6, 7, 8, 10, 14, N49-1, 2, 3, and their physical properties are listed in Table 4. T_mb, V_LSR, ΔV were derived in the same way as for the ¹²CO(J = 3–2) clumps. Virial masses M_vir were estimated from Equation (1), by substituting the ¹²CO(J = 3–2) R_deconv for the clump radius.

¹³CO(J = 1–0) column densities, N(¹³CO) (cm⁻²), are obtained from the following relations under the assumption of local thermodynamic equilibrium:

$$\begin{eqnarray} &&N(^{13}{\rm CO})\;({\rm cm}^{-2})=2.42 \times 10^{14} \left[ \frac{T_{\rm ex}}{1 - {\rm exp}(-5.29/T_{\rm ex}) }\right. \nonumber\\ &&\left. \quad\times \frac{\tau (^{13}{\rm CO})}{1-e^{-\tau (^{13}{\rm CO})}} \frac{1}{f(T_{\rm ex}) - f(T_{\rm bg})} \right] \sum T_{\rm mb}(^{13}{\rm CO}) \delta v,\quad \end{eqnarray} \tag{ 3 }$$

$$\begin{equation} f(T) = \frac{h\nu }{k} \frac{1}{{\rm exp}(h\nu / kT)-1}, \end{equation} \tag{ 4 }$$

where T_ex is the excitation temperature of ¹³CO(J = 1–0), T_bg is the brightness temperature of the background radiation (∼2.7 K), τ(¹³CO) is the optical depth and T_mb(¹³CO) is the observed main-beam temperature of the line. The optical depth is computed as

$$\begin{equation} \tau (^{13}{\rm CO}) = - {\rm ln} \left[ 1 - \frac{ T_{\rm mb}(^{13}{\rm CO}) }{ 5.29 [{\rm exp}(5.29/T_{\rm ex}) - 0.164)] } \right]. \end{equation} \tag{ 5 }$$

∑T_mb(¹³CO)δv in Equation (3) corresponds to the integrated intensity of the ¹³CO(J = 1–0) spectrum. Since the beam-filling factors of our targets are unknown, ¹²CO(J = 1–0) intensity is useless for estimating the excitation temperature of the clumps. We here supposed T_ex = 10 K, which corresponds to the excitation temperature of typical molecular clouds in our galaxy (e.g., Yonekura et al. 1997; Kawamura et al. 1998).

The column density of molecular hydrogen, $N _{\rm H_2}$ can be obtained from N(¹³CO) via the molecular abundance ratios [¹²CO/H₂] and [¹²CO/¹³CO]. We have used same abundance ratios as the N159 in the LMC (Mizuno et al. 2010), taking [¹²CO/H₂] as 1.6 × 10⁻⁵ and [¹²CO/¹³CO] as 50. Thus, we assumed the following relation between $N _{\rm H_2}$ and N(¹³CO):

$$\begin{equation} N({\rm H}_{2}) = \frac{50}{1.6 \times 10^{-5}}N(^{13}{\rm CO}). \end{equation} \tag{ 6 }$$

Finally, the masses M_clump of the ¹³CO clumps are estimated from

$$\begin{equation} M_{\rm clump} = \mu\,m({\rm H}_{2})\sum [D^{2} \Omega N({\rm H}_{2})], \end{equation} \tag{ 7 }$$

where μ is the mean molecular weight per hydrogen molecule, assumed to be 2.8 by taking into account a relative abundance of 71% hydrogen, 27% helium, and 2% metals in mass. m(H₂) is the H₂ molecular mass, D is the distance to the object (in the case of LMC, D ∼ 50 kpc), and Ω is the solid angle subtended by the clumps ($D^2 \Omega \sim \pi R ^{2}_{\rm deconv}$). We summarize these physical properties in Table 4. Mean values in the N48 and N49 regions are M_vir ∼ 1.8 × 10⁴ M_☉, N(H₂) ∼ 2.3 × 10²¹ cm⁻², and M_clump ∼ 1.0 × 10⁴ M_☉.

3.5.1. Temperature and Density Estimation via LVG Modeling

To estimate the kinetic temperature and number density of the molecular gas, we have performed a large velocity gradient analysis (LVG analysis; Goldreich & Kwan 1974; Scoville & Solomon 1974) of the CO rotational transitions. The LVG radiative transfer code simulates a spherically symmetric cloud of uniform density and temperature with a spherically symmetric velocity gradient proportional to the radius, and employs a Castor's escape probability formalism (Castor 1970). The LVG model requires three independent parameters to computer emission line intensities: the kinetic temperature, T_kin, and the density of molecular hydrogen, n(H₂), and X(CO)/(dv/dr). X(CO)/(dv/dr) is the abundance ratio of CO to H₂ divided by the velocity gradient in the cloud. We adopt uniform fractional abundance ratios [¹²CO/H₂] of 1.6 × 10⁵, [¹²CO/¹³CO] of 50, as in Section 3.4. We estimate the mean velocity gradient dv/dr = ΔV_clump/2R_deconv, which is the same estimate with Minamidani et al. (2008). They checked that the change in dv/dr caused by the definition of clump boundary does not largely affect the LVG calculations.

We performed the LVG model calculations over a grid of temperatures in the range T_kin = 5–200 K and densities in the range n(H₂) = 10–10⁶ cm⁻³, with grid spacings of 10^0.02 K and 10^0.125 cm⁻³, respectively. This produced sets of line intensity ratios, $R ^{12}_{3-2/1-0}$ and $R^{12/13}_{1-0}$. The model includes the lowest 40 rotational levels of the ground vibrational state and uses the Einstein A and H₂ impact rate coefficients of Schöier et al. (2005). We do not include higher energy levels, which would require including populations in the lower vibrationally excited states. This imposes a limit of T_kin ∼ 200 K in the present study, and even higher temperatures are not in general excluded below.

To solve for the temperatures and densities that best reproduce the observed line intensity ratio, we calculate chi-squared as

$$\begin{equation} \chi ^{2} = \sum _{i=1}^{N-1} \sum _{j=i+1}^{N} \frac{[R _{\rm obs}(i,j) - R_{\rm LVG}(i,j)]^2 }{\sigma (i,j)}, \end{equation} \tag{ 8 }$$

where N is the number of transitions of the observed molecule (in this case N = 3), i and j refer to different molecular transitions, R_obs(i, j) is the observed line intensity ratio from transition i to transition j, and R_LVG(i, j) is the ratio between transitions i and j, estimated from the LVG calculations. The standard deviation, σ(i, j), for R_obs(i, j) is estimated from the noise level of the observations and the calibration uncertainties.

3.5.2. Results of the LVG Analysis

We derived kinetic temperatures and number densities for the clumps via the LVG analysis described above. The analysis was performed for clumps covered by the Mopra observations, whose ¹³CO(J = 1–0) line intensities T_mb are greater than 3σ noise level at the ¹²CO(J = 3–2) clump peak positions. The selected clumps are N48-1, 2, 3, 4, 5, 6, 7, 8, 14, N49-1, 2, 3 (hereafter N4849 clumps). $R ^{12}_{3-2/1-0}$ and $R^{12/13}_{1-0}$ were calculated observationally from the ratios of the peak main-beam temperature values, where the ¹²CO(J = 3–2) data has been smoothed to the resolution of the Mopra data. The uncertainties of the two ratios are estimated to be 8%–28% and 10%–31%, respectively, calculated by combining the calibration errors and the formal errors on the Gaussian model fits. We also performed the same analysis on the clumps of Minamidani et al. (2008, 2011). (These authors also performed their own LVG analysis, but we repeat the calculations using the method and parameters described above, in order that the results may be consistently compared.) The χ² method enables us to derive a best solution for the physical parameters, which had sometimes been difficult to estimate from the overlap of two ratios (see Appendix A of Minamidani et al. 2008). The selected clumps are 30 Dor–1, 3, 4, N159–1, 2, 4, N166–1, 3, 4, N206–1, 2, N206D–1, GMC225–1, 3 (see Table 2 of Minamidani et al. 2008, hereafter M0811 clumps). We summarize the parameters used in the LVG analysis in Table 5.

Table 5. Parameters for LVG Analysis

Region	Clump	Tmb				Line Ratio		dv/dr
		12CO(J = 3–2)	12CO(J = 1–0)	13CO(J = 1–0)		R3 − 2/1 − 0	R12/13
Name	No.	(K)	(K)	(K)				(km s−1)
(1)	(2)	(3)	(4)	(5)		(6)	(7)	(8)
N48	1	3.3 ± 0.2	3.2 ± 0.2	0.31 ± 0.04		1.0 ± 0.1	10 ± 1	0.7
	2	2.5 ± 0.2	3.3 ± 0.2	0.19 ± 0.03		0.76 ± 0.08	17 ± 3	0.7
	3	1.7 ± 0.2	1.8 ± 0.1	0.11 ± 0.03		0.94 ± 0.12	16 ± 5	1.0
	4	3.6 ± 0.3	2.7 ± 0.2	0.30 ± 0.04		1.3 ± 0.1	9.0 ± 1.4	0.4
	5	2.8 ± 0.2	2.1 ± 0.2	0.19 ± 0.03		1.3 ± 0.2	11 ± 2	0.4
	6	2.3 ± 0.2	2.0 ± 0.2	0.14 ± 0.04		1.2 ± 0.2	14 ± 4	0.7
	7	2.7 ± 0.2	2.8 ± 0.2	0.21 ± 0.04		0.96 ± 0.10	13 ± 3	0.2
	8	1.7 ± 0.2	1.7 ± 0.1	0.17 ± 0.03		1.0 ± 0.1	10 ± 2	0.3
	14	1.3 ± 0.2	2.2 ± 0.2	0.20 ± 0.04		0.59 ± 0.11	11 ± 2	0.4
N49	1	3.6 ± 0.3	4.8 ± 0.3	0.56 ± 0.06		0.75 ± 0.08	8.6 ± 1.1	0.3
	2	2.1 ± 0.2	3.2 ± 0.2	0.33 ± 0.06		0.66 ± 0.07	9.7 ± 1.9	0.2
	3	1.4 ± 0.2	1.6 ± 0.2	0.30 ± 0.05		0.88 ± 0.17	5.3 ± 1.1	0.3
30 Dor	1	3.7 ± 0.5	1.8 ± 0.4	0.13 ± 0.03		2.1 ± 0.5	14 ± 4	0.9
	3	2.0 ± 0.3	1.5 ± 0.3	0.13 ± 0.03		1.3 ± 0.3	12 ± 4	0.9
	4	2.3 ± 0.4	1.7 ± 0.3	0.13 ± 0.03		1.4 ± 0.3	13 ± 4	0.5
N159	1	8.6 ± 1.2	6.2 ± 1.2	0.80 ± 0.16		1.4 ± 0.3	7.8 ± 2.2	0.9
	2	6.1 ± 0.8	4.0 ± 0.8	0.44 ± 0.09		1.5 ± 0.4	9.1 ± 2.6	0.5
	4	3.8 ± 0.5	4.5 ± 0.9	0.72 ± 0.14		0.84 ± 0.20	6.3 ± 1.7	0.4
N166	1	3.3 ± 0.7	4.0 ± 0.8	...a		0.83 ± 0.17	11 ± 2a	0.3
	3	1.6 ± 0.3	2.3 ± 0.5	...a		0.70 ± 0.15	13 ± 3a	0.3
	4	0.98 ± 0.2	1.3 ± 0.3	...a		0.75 ± 0.17	16 ± 3a	0.4
N206	1	2.6 ± 0.5	3.1 ± 0.6	...a		0.84 ± 0.16	14 ± 3a	0.5
	2	1.3 ± 0.3	2.1 ± 0.4	...a		0.62 ± 0.12	9.8 ± 2.0a	0.4
N206D	1	3.3 ± 0.5	4.6 ± 0.9	0.85 ± 0.17		0.72 ± 0.18	5.4 ± 1.5	0.3
GMC225	1	1.5 ± 0.2	3.4 ± 0.7	0.55 ± 0.11		0.44 ± 0.11	6.2 ± 1.8	0.2
	3	1.2 ± 0.2	2.8 ± 0.6	...a		0.43 ± 0.09	6.7 ± 1.3a	0.3

Notes. Columns 1 and 2: names of the region and ID numbers of the clumps that we performed LVG analysis. Columns 3–5: peak main beam temperatures of ¹²CO(J = 3–2), ¹²CO(J = 1–0), ¹³CO(J = 1–0) lines. Columns 6 and 7: line intensity ratios that we have used in LVG analysis. Ratio of ¹²CO(J = 3–2) and ¹²CO(J = 1–0), R_{3 − 2/1 − 0}, and ratio of ¹²CO(J = 1–0) and ¹³CO(J = 1–0), R_12/13, are shown. Column 8: averaged velocity gradients of the clumps. ^aThe value quoted from Minamidani et al. (2008) as an intensity ratio R_12/13.

Download table as:  ASCIITypeset image

We regarded the lowest values of χ² as the best solution for the temperatures and the densities of the clumps, and define error bars as χ² < 3.84, which corresponds to the 5% confidence level of the χ² distribution with one degree of freedom. We omitted those clumps whose lowest χ² value is greater than 3.84. When the valid region of parameter space exceeds the boundary of the calculated range, T_kin = 5–200 K and n(H₂) =10–10⁶ cm⁻³, we treat the edge of the calculated range as the true boundary. Contour plots of LVG analysis on the n(H₂)–T_kin plane are summarized in the Appendix (Figures 15 and 16), and derived values are summarized in Table 6.

Table 6. Results of LVG Analysis

Region	Clump		n(H2) (cm−3)			Tkin (K)		min. χ2		F24 μm
Name	No.		χ2 < 3.84	Min. χ2		χ2 < 3.84	Min. χ2			10−12 (erg s−1 cm−2)
(1)	(2)		(3)	(4)		(5)	(6)	(7)		(8)
N48	1		103.3–105.2	105.0		>46	191	0.023		99
	2		102.8–103.3	103.0		46–183	91	0.053		8.4
	3		103.0–105.5	103.4		>34	72	0.035		12
	4		...			...	...	10		31
	5		103.3–105.8	104.0		>79	126	2.0		110
	6		103.1–105.2	104.1		>52	126	0.33		41
	7		102.7–104.6	103.1		>84	191	0.00030		26
	8		103.1–104.9	103.9		>61	120	0.0039		43
	14		102.6–103.1	102.9		23–96	50	0.070		11
N49	1		102.7–104.8	103.0		>27	55	0.010		15
	2		102.4–104.5	102.7		>19	72	0.0036		6.1
	3		102.8–105.3	103.5		>13	50	0.0054		16
30 Dor	1		...	...		...	...	4.2		5600
	3		102.9–104.7	104.5		>51	126	0.95		1900
	4		103.1–105.0	104.1		>61	151	1.1		1200
N159	1		103.5–105.6	105.3		>35	182	2.0		2000
	2		103.1–105.3	104.9		>38	191	1.7		2500
	4		102.9–105.4	103.4		>15	44	0.0041		23
N166	1		102.6–104.8	103.1		>38	91	0.011		11
	3		102.4–103.5	102.9		>32	100	0.058		8.0
	4		102.4–103.6	102.9		>46	132	0.0041		25
N206	1		102.7–105.0	103.1		>39	87	0.0087		140
	2		102.7–103.3	102.9		20–87	44	0.080		51
N206D	1		102.7–105.3	103.1		>12	32	0.0042		100
GMC225	1		102.5–102.8	102.7		13–48	19	0.24		2.8
	3		102.6–103.0	102.7		11–34	19	0.16		1.9

Notes. Columns 1 and 2: names of the regions and ID numbers of the clumps that we performed the LVG analysis. Columns 3–6: derived number density, n(H₂) cm⁻³, and kinetic temperature, T_kin, of the clumps are shown. The range of value for that χ² is less than 3.84, and the values at the point that minimum χ² are indicated. Column 7: the values of minimum χ². Column 8: flux densities of Spitzer 24 μm from each 45'' area toward the peak position of the clumps.

Download table as:  ASCIITypeset image

Although N48-4 and 30 Dor–1 are omitted from the calculation due to their high minimum χ² values (exceed 3.84), it is still possible to obtain an estimate of their properties. In Minamidani et al. (2011), the LVG results of 30 Dor–1 with different line ratios are quoted as $T_{\rm kin} = 134^{+66}_{-49}$ K and $n({\rm H_2}) = 4.6^{+2.4}_{-1.3} \times 10^3$ cm⁻³, making it one of the densest and warmest clumps in their sample. Since the LVG results of N48-4 are similar to those of 30 Dor–1, and $R^{12}_{3-2/1-0}$ is high in both clumps (≃1.3), it is possible that the temperatures and densities of N48-4 may be as high as that of 30 Dor–1.

The output of the LVG calculations are shown as n(H₂)–T_kin plots in Figures 7(a) and (b). The N4849 clumps are typically warm (greater than 50 K), with a wide distribution of densities (10³–10⁵ cm⁻³). n(H₂) and T_kin of the N48 clumps tend to be higher than those of the N49 clumps. In Figure 7(b), the M0811 clumps are colored according to their parent GMC types defined in Kawamura et al. (2009): "starless" molecular clouds in the sense that they are not associated with H ii regions or young clusters (Type I); molecular clouds with H ii regions (Type II); and molecular clouds with H ii regions and young clusters (Type III). The clumps in active star forming GMCs (Type III: 30 Dor, N159) tend to be warm and dense, and the clumps in starless GMCs (Type I: GMC225) tend to be colder and less dense. These tendencies suggest an evolutionary sequence in terms of increasing density leading to star formation and increased star formation activity leading to intense FUV photons that heat the clouds, as discussed in Minamidani et al. (2008, 2011). The n(H₂)–T_kin plots of the N4849 clumps are distributed among those of M0811 clumps hosted by Type II and Type III GMCs. This suggests that the N48 region GMC may be just at the stage of evolving from Type II to Type III. This is consistent with the suggestion of the GMCs in the N48/N49 region are in the early stage of a cluster-forming cloud (Mizuno et al. 2001).

Zoom In Zoom Out Reset image size

We note that here uniform fractional abundance ratios [¹²CO/H₂] of 1.6 × 10⁵ have been adopted, which was used in Mizuno et al. (2010) and Minamidani et al. (2011). Different abundance ratios give different solutions (higher temperature and lower density for a higher abundance ratio), but the differences are not significant (Minamidani et al. 2008). We consider the assumption of uniform fractional abundance to be reasonable, and to not affect the relative tendencies of the results.

Maps of the ¹²CO(J = 3–2) to ¹²CO(J = 1–0) integrated intensity ratio (hereafter $R_{3-2/1-0}^{\rm int}$) are shown in Figure 8(a). Note that the ASTE data is smoothed to the same effective resolution as the Mopra maps (45'') before this ratio is taken. These are a useful means of showing relationships between the physical conditions of the clumps and the spatial distribution of star formation tracers—Hα nebulae, YSOs the IR dust emission. Mean values of $R_{3-2/1-0}^{\rm int}$ are 0.9 for the N48 region and 0.5 for the N49 region, with errors of 0.085 and 0.14, respectively (combination of the noise rms and calibration uncertainties). The clumps in the N48 region show significantly higher $R_{3-2/1-0}^{\rm int}$ than those in the N49 region, even taking into account the relatively large calibration error of the N49 data. In the N48 region, $R_{3-2/1-0}^{\rm int}$ is typically higher than the unity around the peak position of each clump, indicating that the CO molecules are highly excited due to high density or temperature of the clumps.

Zoom In Zoom Out Reset image size

Figure 8(b) shows the Hα flux distribution and positions of the Spitzer YSO candidates. H ii regions and an SNR identified by Henize (1956) and Davies et al. (1976) are indicated in the figure and also listed in Table 7. The Hα emission nebulae take spatial offset from the clumps, and the clumps have no prominent Hα emission inside them. This indicates that prominent H ii regions have not formed yet inside the molecular clumps. However, since Hα emission is strongly affected by dust extinction, it is possible that H ii regions are embedded within or hidden behind the molecular gas. We compared the ¹²CO(J = 3–2) distribution with 1.4 GHz radio continuum data (Filipovic et al. 1995; Filipović & Staveley-Smith 1998; Hughes et al. 2007) and confirm that there is no prominent 1.4 GHz emission where Hα emission is absent, and hence that there are no such hidden H ii regions behind or within the molecular clumps.

Figure 8(c) shows a 24 and 8.0 μm two-color image, with ¹²CO(J = 3–2) contours and Spitzer YSO candidates overlaid. These mid-infrared bands are a useful probe of embedded star formation activity (Calzetti et al. 2007), since 24 μm emission is primarily due to thermal emission from hot dust heated by nearby stars, and diffuse 8.0 μm emission mainly arises from polycyclic aromatic hydrocarbon (PAH) molecules (Li & Draine 2001), which are fluorescing in the local UV radiation in the neutral zones of the PDRs (e.g., Calzetti et al. 2007). The 24 and 8.0 μm emission is well correlated with the molecular clumps, especially in the N48 region. A prominent 24 μm feature is seen in the southern part of the N48 region (around Clump-1 and 5), indicating that the most active star formation is occurring around this area. The 8.0 μm emission shows quasi-linear bar-like distribution running from north to south. As discussed in Cohen et al. (2003), this might be indicative of a region of high-density gas swept up by the expansion of LMC 4 and LMC 5, producing a dense PDR in the dense molecular clouds. PAH abundances are believed to be enhanced by ISM shocks through shock fragmentation of graphite grains (e.g., Jones et al. 1996). So it is also possible that the PAH molecules in these regions were produced by the large-scale shock of the SGS collision and have survived only in and around the molecular clumps with high enough column densities to shield them from local UV radiation.

The YSO candidates are distributed in and around the molecular clumps, suggesting that recent star formation has occurred in both the N48 and N49 regions. Massive YSO candidates are only found in the N48 region, where the prominent 24 μm emission is located. Figure 9 shows a zoomed-in view of Figure 8(a) around the N48 Hα nebulae with additional plots of previously identified OB stars (Will et al. 1996). Note that the sample consists of 11 targets that are luminous in the V band (brighter than 15 mag). These OB stars are located away from the molecular clumps, which indicates that their parental clouds have already been dispersed by UV radiation or stellar winds. On the other hand the YSO candidates are preferentially located inside the clumps. The ages of these OB stars are estimated to be 5–10 Myr, and the age of the YSOs can be estimated to be ∼1 Myr, assuming a YSO formation timescale of 0.2 Myr (Whitney et al. 2008). These features are well interpreted as a time sequence or age gradient of the star formation activity in this area.

Zoom In Zoom Out Reset image size

The distribution of $R_{3-2/1-0}^{\rm int}$ shows minimal correlation with that of Hα emission. For those N48 clumps that are located beside H ii regions (Clump 1, 5, 6, 7, 9, 12), the ratios are relatively high around the center of the clumps and decrease toward the rims of the clumps. An increase of the ratio toward the H ii regions is only seen around Clump 12, the only example in which the effect of UV radiation on the ratio is clearly seen. On the other hand, $R_{3-2/1-0}^{\rm int}$ shows a moderate spatial correlation with the YSOs and the infrared emission, especially the 8.0 μm flux. These indicate that $R_{3-2/1-0}^{\rm int}$ is more (effectively) enhanced by internal heating sources such as the YSOs and the PDR than by H ii regions.

Here the gravitational state of the clumps based on the clump mass derived from the ¹³CO(J = 1–0) line is discussed. A plot of M_vir against M_clump is displayed in Figure 10, in which the clumps are classified according to their spatial correlation with Spitzer YSO candidates. Here M_vir may be thought of as an estimate of the mass required for a clump to be self-gravitating, and M_clump as an estimate of the luminosity mass within an assumption of local thermodynamical equilibrium. It is notable that the absolute value of both mass have errors due to the definition of the clump boundary and the estimate of relative molecular abundances, but these errors are less significant when we compare the relative differences between the clumps. The values of M_vir/M_clump are shown in Table 4. The clumps with YSOs tend to be closer to M_vir/M_clump ∼ 1 than the clumps without YSOs, indicating that the clumps with YSOs are relatively gravitationally relaxed than the clumps without YSOs. The best-fit line to the full sample is $M _{\rm vir} = 40 M _{\rm clump}^{0.65}$ with a correlation coefficient 0.68. This corresponds to $M _{\rm vir}/M _{\rm clump} \propto M _{\rm clump}^{-0.35}$. This trend of decreasing M_vir/M_clump with increasing M_clump is also reported in previous Galactic and LMC works (e.g., Orion: Ikeda et al. 2007, 2009; Cygnus: Dobashi et al. 1996; Cepheus and Casseopeia: Yonekura et al. 1997; entire the LMC: Wong et al. 2011; and 30 Dor: Indebetouw et al. 2013). The power-law index of −0.35 for the N48/N49 clumps is smaller than the theoretical value of −2/3 for pressure-confined and magnetized clumps ($M _{\rm vir}/M _{\rm clump} \propto M _{\rm clump} ^{-2/3}$; Bertoldi & McKee 1992) and is rather similar to that of Orion (−0.33) and Cepheus and Casseopeia (−0.38). This suggests that the ratio of the internal kinetic energy to the gravitational energy in the N48 and N49 clumps is relatively low and the N48 and N49 clumps are gravitationally relaxed. However, this is a tentative result because the uncertainty of the power-law index is large due both to the small number of data points and to the uncertainties in both masses, mainly arising from the estimates of clump radius. There is no significant difference in the masses of the clumps with massive YSOs and those with intermediate mass YSOs. Although observations at higher spatial resolution (∼2 pc) with higher density tracers (HCO⁺ and HCN) in the LMC have revealed that massive YSO bearing clumps tend to be larger and more massive than those without signs of star formation (Seale et al. 2012), the correlation between the mass of the ¹³CO clumps and the YSOs cannot be seen in our data at a spatial resolution of 11 pc. Note that here we have assumed a uniform excitation temperature of 10 K in the derivation of M_clump, and this might be a lower limit for the N48/N49 region since it corresponds to typical excitation temperatures of Galactic molecular clouds with fewer signs of star formation (e.g., the Gemini and Auriga regions; Kawamura et al. 1998). However, this does not significantly affect the trends seen in the mass ratio, for instance, if we assume T_ex = 20 K, M_clump increases by only a factor of ∼1.4.

Zoom In Zoom Out Reset image size

Minamidani et al. (2008, 2011) discussed the relationship between n(H₂) and T_kin in molecular clumps with their parent GMC types, as defined by Kawamura et al. (2009) (see Section 3.5.2). However, GMC types are defined by star formation activity on the size scale of entire GMCs, which have typical radii of 30–50 pc (Fukui et al. 2008). Here we instead assign as similar "type" to the N4849 and M0811 clumps based on the star formation activity in and around them on ∼10 pc size scales. We sort the clumps into three categories: Type I: clumps without O stars capable of ionizing an H ii region, which does not exclude the possibility of associated young stars later than B type; Type II: clumps with small H ii regions only; Type III: clumps with stellar clusters and H ii regions. We used the catalog of Henize (1956) and Davies et al. (1976) for H ii regions and require an overlap of a clump with diffuse Hα flux from these regions in order for it to be assigned to Type II or Type III. We used the catalog of Bica et al. (1996) for young clusters (<10 Myr; SWB 0). Figure 11 shows the distribution of n(H₂) and T_kin sorted by the clump type. Type I clumps tend to be cooler and less dense, whereas clumps with star formation activity tend to be warmer and denser. In particular, only dense, hot clumps are identified as young star-cluster-forming clumps (Type III). This is demonstrates that n(H₂) and T_kin in molecular clumps are related to the star formation activity within them.

Zoom In Zoom Out Reset image size

We also compare n(H₂) and T_kin with the integrated 24 μm flux contained within the half power width of the 45'' beam toward each peak position (Figure 12). The derived values are listed in Table 6. As displayed in Figure 12, 24 μm flux shows a positive correlation with n(H₂) and T_kin. Emission at 24 μm originates from small dust grains mainly heated by UV photons from young stars, and it has been shown to directly relate to ongoing star formation over a timescale of τ₂₄ ∼ 10 Myr (Calzetti et al. 2005; Pérez-González et al. 2006; Calzetti et al. 2007). The positive correlation with n(H₂) indicates that star formation activity is higher in denser molecular clumps, and the positive correlation with T_kin indicates that the temperatures of the clumps are higher where the UV radiation is more intense. These positive correlations, including the trend for increased temperature and density with higher levels of star formation activity (Figure 11), suggest that n(H₂) and T_kin trace an evolutionary sequence of the clumps. Compared with the M0811 clumps, the N4849 clumps, particularly those in the N48, show high densities and temperatures even though 24 μm flux is not prominent. If density is indeed an effective probe of molecular clump evolutionary state, then these high densities with low 24 μm flux indicate that the clumps are about to form stars, and star formation will be more active in the near future. High temperatures with low 24 μm flux suggest that the N4849 clumps are heated effectively due to the lower mass and smaller size, as seen in Section 3.3. There may also be an alternative heating source, potentially something that cannot be traced by 24 μm emission, such as low-mass stars.

Zoom In Zoom Out Reset image size

H i gas is diffuse and extended compared with the molecular clouds traced by CO, and it is the better tracer of the global kinematics and structure of the SGSs. The H i gas distribution therefore allows us to examine the relation between the physical properties of the clumps and the structure of the SGSs. We use H i survey data taken with the Australia Telescope Compact Array (Kim et al. 1998) and the Parkes single-dish telescope (Staveley-Smith et al. 2003), combined by an image feathering (linear merging) approach (Kim et al. 2003). The data cube covers the entire LMC at a spatial resolution of 1' and a pixel size of 20'', and has a 1σ noise level of 2.4 K in a 1.65 km s⁻¹ velocity channel.

Figure 13 shows a color image of the H i column density with ¹²CO(J = 3–2) contours overlaid. The H i forms a large ridge, which extends northeast to southwest for several hundred parsecs. The SGSs can be seen as giant cavities in the H i gas (east: LMC 4, west: LMC 5), and the ridge is located between the two shells. The molecular clumps are distributed along the ridge, and the H i column density is high (up to N(H₂) ∼ 6 × 10²¹ cm⁻²) around the N48 clumps. Figure 14 shows velocity channel maps of H i brightness temperature and ¹²CO(J = 3–2) intensity. The two tracers show good spatial correlation and the clumps are distributed along the H i ridge. Then the spatial coincidences in the N48/N49 regions on the size-scales of clumps suggest that the clumps are associated with the H i envelopes and are formed in the high column density H i ridge.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

There are some offsets in CO and H i peak positions, particularly obvious in the velocity range 288–292 km s⁻¹. These offsets are not surprising, and it has been pointed out that in general H i envelopes tend to be not perfectly correlated with CO emission, but show some offset in their peak positions, even on the size-scales of GMCs (Fukui et al. 2009). This can be interpreted as the conversion of H i into molecular H₂ such that the amount of H i gas is reduced around the molecular clumps. Another interpretation is that the H i envelopes around CO clouds are likely to be the dense and cold atomic gas from which molecular clumps are formed (Fukui et al. 2009). Such cold, optically thick H i typically makes only a small contribution to H i emission profiles, meaning that the apparent H i intensity may be relatively small at the velocities at which these components are present. It is likely that both explanations are partially responsible for the observed offsets. One-dimensional numerical simulations of the propagation of a shock wave into the atomic ISM find that thermal instabilities occur in the shock-compressed layer, which produces geometrically thin, dense layers of gas in which a large amount of hydrogen molecules are formed (e.g., Koyama & Inutsuka 2000; Hosokawa & Inutsuka 2007). Two-dimensional models of colliding supershells whose terminal sizes reach several hundreds of parsecs (Ntormousi et al. 2011), suggest that small (typical sizes of <1 pc), dense (up to 10⁴ cm⁻³, fulfilling the density requirements for molecule formation), cold (T < 100 K) gas clumps and filaments form naturally by thermal instabilities in the highly turbulent and compressed collisional area of the two supershells. It has also been pointed out from observational studies that the cool H i gas in the LMC is unusually abundant compared to the cold atomic phase of the Milky Way (Dickey et al. 1994; Marx-Zimmer et al. 2000). The spatial offset between CO and H i peaks in the N48/N49 region is consistent with a scenario in which the molecular gas was formed in cold neutral medium that embedded in more extended warm neutral medium. The current spatial resolution of H i data is not high enough for the investigation of the existence of such a small and cold gas, and higher resolution observations will be necessary in order to discuss it further.

Note that there are shell-like H i holes where the H ii regions N48-B, C, and the SNR N49 are located (Figure 14). The spatial agreement of the holes and the Hα nebulae can be clearly seen in the bottom right panel of Figure 14. In these holes, H i gas has presumably been dispersed or completely ionized by the UV radiation and the SNR shock. Around the N48 H ii regions, 8.0 μm flux is also suppressed (Figure 8(c)), indicating that PAH molecules are have also been dispersed or destroyed. These tendencies strongly suggest that the N48 H ii regions and the N49 SNR form local shells, which may be currently expanding (although the spatial resolution of the H i data is insufficient for expansion to be confirmed). These small shells may also have triggered star formation around them. It is notable that massive YSOs are only found in the clumps around the shells (Figure 8), as is the bright 24 μm point-like feature, which indicates young star formation. The clumps next to these holes (Clump 1, 4, 5, 6, 8, and 9) show higher n(H₂) and T_kin than those isolated from the H ii regions (Table 6, Figure 11). However, further investigation is necessary to confirm that these are signs of heating and/or compression of molecular clumps by the H ii regions, because $R_{3-2/1-0}^{\rm int}$ in these clumps does not appear to be affected by their presence (Figure 8(a)).

The H i distribution suggests that the N49 clumps are associated with only one SGS, LMC5, whereas the N48 clumps are located right in the boundary of the two SGSs. Comparisons of these two regions may therefore provide insight into the differences between the physical effects of a single shell and the interaction of two SGSs on the molecular clump properties and star formation activity. The excitation state of the molecular clumps, as traced by R_{3 − 2/1 − 0} and $R_{3-2/1-0}^{\rm int}$, is apparently higher for the entire population of molecular clumps in the N48 region (Figure 8(b)). The value is typically higher than unity around the clump peak positions, which indicates that the CO molecules are efficiently excited in high density or temperature conditions. n(H₂) and T_kin are higher in the N48 region, and some of the clumps show high densities that are comparable to massive cluster-forming clumps in the LMC molecular ridge and CO Arc (Figure 7(a)). Although there is no apparent difference in virial state between the two regions (Figure 10), the dense, warm, and highly excited states of the N48 clumps indicate that they have the more suitable conditions for the formation of massive stars or clusters. YSO candidates are found in both regions, but massive star formation is more active in the N48 region. There is no sign of massive star formation in the N49 clumps, whereas the N48 region contains several H ii regions, and massive YSOs and luminous 24 μm and 8.0 μm emission can be seen in the molecular clumps. The star formation in the N48 region seems to be more evolved compared to the N49 region. These differences of molecular clump properties and star formation evolutionary stage suggest that formation of dense molecular clumps and massive stars is enhanced in the region of high-density gas swept up by the two SGSs, and this SGS interaction has worked more efficiently to create stars from the ISM than the effect of only one SGS.

In the previous subsection, we argued that the high density molecular clumps have been formed at the interface of the two SGSs. The spatial correlation between H i and CO implies that the dense clumps are formed in the high column density H i ridge, but to conclude that the shells actually trigger formation of these clumps, we must check the consistency of the shell and GMC formation timescales. The dynamical ages of LMC 4 and LMC 5 have been estimated as 4–15 Myr, and 5–7 Myr, respectively, using the current observed expansion velocities of the H i gas (Dopita et al. 1985; Kim et al. 1999; Book et al. 2008). However, these dynamical ages give roughly lower limits for the shell formation timescales, because they are derived under the assumption that the SGS is formed as a single expanding shell by feedback from a single generation of star formation. There is alternative theory that large SGSs might be formed by several generations of star formation (Efremov & Elmegreen 1998), and age estimates based on stellar population studies in individual SGSs have indicated ages of typically 10–20 Myr (e.g., Dopita et al. 1985; Points et al. 1999; Glatt et al. 2010). Particularly in LMC 4, there is an extended stellar arc (∼600 pc) in Constellation 3, which is located in the central area of the H i hole, and it has been argued that this arc was formed in the gas swept up by the first generation stellar feedback within the triggering timescale of ∼14 Myr (Efremov & Elmegreen 1998). The arc itself is also an energy source via second generation stellar feedback. We here adopt ages of 10–20 Myr for the two SGSs, with a particularly strong constraint of >15 Myr implied for LMC 4. Typical timescales for the formation of GMCs from diffuse H i gas have been roughly estimated as ∼10 Myr in the LMC (Fukui et al. 2009), which corresponds to the crossing time of a typical GMC of ∼100 pc size assuming a typical H i velocity dispersion of ∼10 km s⁻¹ (Fukui & Kawamura 2010). The free-fall time of H i gas whose density is ∼10 cm⁻³ is also ∼10 Myr. The GMCs in the N48/N49 region are classified as Type II GMCs, which are roughly estimated to have ages of between 6 and 19 Myr (Kawamura et al. 2009). These timescales are on the same order of the estimated SGS age of 10–20 Myr, and are consistent with a scenario in which the GMCs in the N48/N49 region were formed by the SGSs.

The suggested evolutionary scenario for the N48 and N49 region can be summarized as follows: Dense, young, clumpy, star-forming molecular clumps were formed in the large-scale H i ridge accumulated by the two SGSs, LMC 4 and LMC 5. The densest molecular clumps (∼10⁴ cm⁻³), in which star formation is more advanced, were formed at the interaction zone of the two SGSs. Some massive stars have already dispersed their parental molecular gas and formed H ii regions. Thus, the large-scale structure of the SGSs affects the formation of the young, star-forming molecular clumps; in particular the interaction of the two SGSs plays an important role and triggers the active formation of dense molecular clumps and massive stars.