The G332 molecular cloud ring: I. Morphology and physical characteristics

Abstract

1 INTRODUCTION

In our Galaxy, as in other galaxies, the space between the stars is filled by a faint medium that consists of ionized, atomic, and molecular gas, high-energy particles (cosmic rays), and dust grains at micron scales. Cosmic rays (CRs), magnetic fields, and the turbulent motion of diffuse gas and dust grains, perturb the interstellar medium (ISM) and contribute to it being inhomogeneously distributed at small scales. The molecular structures within the ISM span a wide range of sizes and masses, from Bok globules (Bok & Reilly 1947) of few solar masses on parsec scales (Clemens, Yun & Heyer 1991), where low-mass stars are born (Yun & Clemens 1990), to immense giant molecular clouds (GMCs) with masses of the order of 10⁴–10⁶M_⊙ spanning tens of parsecs (Bally 1986).

Molecular clouds are the densest and coldest regions of the ISM. Inside them, in regions where CRs, magnetic fields, and violent motions of gas are unable to provide support against gravitational collapse, the molecular gas condenses to form new stars. The ionization and turbulent feedback from the massive newborn stars photodissociates and perturbs the surrounding gas, dissolving the hosting molecular cloud as between 2 and 20 per cent of the original cloud mass is converted into stars (Molinari et al. 2014).

The gas processed by the thermonuclear reactions in the interior of the stars is recycled back into the ISM through mechanisms such as stellar winds and supernova explosions. The injection of matter and energy into the surrounding medium closes the Galactic ecology cycle between the ISM and the stars, continuously enriching the ISM gas reservoir with heavy elements and more complex molecules ready to form a new generation of stars.

The typical time-scale of this cycle is about 5 –10 × 10⁷ yr. With the mean star formation efficiency in GMCs being about 5 per cent, a molecule could exist in the ISM for 1 –2 × 10⁹ yr before being part of a star (Molinari et al. 2014).

The detection of molecular hydrogen (Carruthers 1970) in the far-UV spectrum of a hot star, and the detection of CO (Wilson, Jefferts & Penzias 1970) as the second most abundant molecule in space, started the era of systematic observation of the distribution of interstellar molecules. Although molecular hydrogen is ubiquitously present in the ISM, and is the primary component of GMCs, it is extremely difficult to directly observe. Its lowest transition energy (quadrupole J = 2 → 0 line at 28.22 |$\mu$|m) is at |$510\rm \, K$| above ground (Dabrowski 1984), very far from the typical GMCs temperatures (10–20 |$\rm \, K$|⁠; Goldsmith 1987). The carbon monoxide molecule, having an almost constant relative abundance with respect to H₂ (equal to ∼10⁻⁴), and a J = 1 → 0 dipole transition at |$5\rm \, K$| above the ground level, is a good proxy for H₂ in the environments of interstellar clouds (Scoville & Sanders 1987). The empirical link between the measured quantity, the ¹²CO line flux, and the H₂ column density is the X_co conversion factor (Dame, Hartmann & Thaddeus 2001; Shetty et al. 2011; Bolatto, Wolfire & Leroy 2013) which is typically X_co ≈ 2.0–2.7|$\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$| for |$\vert \mathit {b}\vert \lt 1{^{\circ }}$|⁠.

Many experimental observations of the ISM distribution across the Galactic Plane have been made and include the emission in various bands of CO (Heyer et al. 1998; Dame et al. 2001; Mizuno & Fukui 2004; Jackson et al. 2006; Burton et al. 2013; Jones et al. 2013; Barnes et al. 2015; Schuller et al. 2017; Umemoto et al. 2017), H i 21 cm (Gibson et al. 2000; McClure-Griffiths et al. 2001; Stil et al. 2006), methanol masers (Green et al. 2009), HCO⁺, N₂H⁺, NH₃, water maser, and other tracers (Jones et al. 2012; Jackson et al. 2013; Walsh et al. 2014), 1.3 mm continuum (Aguirre et al. 2011), |$870\,\hbox{$\mu $m}$| continuum (Schuller et al. 2009), 350–10|$^4\,\hbox{$\mu $m}$| Planck (Planck Collaboration 2014), 70–500 |$\hbox{$\mu $m}$| Herschel (Molinari et al. 2010), 65–160 |$\hbox{$\mu $m}$| Akari (Ishihara et al. 2010), 3.6–24 |$\hbox{$\mu $m}$| Spitzer (Benjamin et al. 2003; Carey et al. 2009; Gutermuth & Heyer 2015), 3.4–22 |$\hbox{$\mu $m}$| Wise (Wright et al. 2010), and 8.2–21.3 |$\hbox{$\mu $m}$| MSX (Price et al. 2001). The inventory of molecules detected in the ISM ranges from simple ones such as H₂O, to organic and more complex molecules such as aldehydes, fullerenes, and polycyclic aromatic hydrocarbon (PAH; Tielens 2013). The ISM is far from being a simple medium.

In the analysis performed in this paper, we will examine in detail the general physical characteristics and the morphology of a GMC recently discovered in the G332 region of the Galactic Plane, identified using data from the Mopra CO Southern Galactic Plane Survey (Braiding et al. 2018).

1.1 Paper structure

In this section, we will give a brief description of how we structured our study. We start with a short presentation of our CO data and the relative reduction pipeline (Section 2), followed by an overview of the CO emission in the G332 sector and the molecular ring cloud (Section 3.1), including how we inferred the cloud distance by solving the kinematic distance (KDA; Section 4). In Section 5.1, we describe the adopted data masking method for creating moment maps (Section 5.1), and the derivation of the CO column density (Section 6) assuming local thermodynamic equilibrium (LTE) conditions. The dust spectral energy distribution (SED) fit and its relative molecular hydrogen column density N(H₂), relative to the molecular ring cloud, are described in Section 6.5. In order to investigate the local variation of the X_co factor, in Section 7.1 we compare the ¹²CO integrated intensity map and N(H₂) column density derived from ¹³CO, at different T_ex. Taking advantage of the availability of [C i] line data, we define an X|$\mathrm{_{CI}^{809}}$| factor (Section 7.2) by comparing the molecular hydrogen column density traced by ¹³CO and ¹²CO to the [C i] integrated intensity. Moreover, we used the atomic carbon emission, in combination with the ¹³CO detection, to quantify the dark H₂ gas not traced by carbon monoxide (Section 7) and analyse how the [C i] /¹³CO ratio varies in the ring cloud (Section 8.3). To further inspect the different conditions occurring in the cloud, in Section 9 we describe how we identified clumps and report the C¹⁸O principal physical properties in a catalogue (Table A1). In the discussion section we analyse (Section 10.1) how we disentangled the dust emission across the Galaxy with a spiral arm model, while in Section 10.2 we examine the star formation activity ongoing in the cloud. The relation between the measured X_co variation and the physical conditions observed in the cloud are discussed in Section 10.3, and in Sections 10.4 and 10.5, we present ways in which the different physical conditions of the cloud can be inspected using other tracers from CO, such as [C i] and |$8\,\hbox{$\mu $m}$| emission. We end our study (Section 11), presenting new methods based upon virtual reality (VR) and augmented reality (AR) techniques to visualize radio astronomy data, we present three different VR/AR representations in Appendix  A.

To help localize the regions of the cloud, single objects, or specific features, we will use the convention shown in Fig. 1. All the global cloud quantities reported in the paper are derived from the region resulting by the union of [SE-R], [SW-R], [W-R], [NW-R], [N-R], and [C-R], if not otherwise stated.

Figure 1.

Summary of the molecular ring divisions, superimposed on a ¹³CO moment zero map, used here to show morphology. We divided the cloud into parts to examine their typical physical conditions: [N-R] light red, [NW-R] in light orange, [W-R] in blue, [SW-R] in yellow, [SE-R] in violet, and [C-R] in green, here the colours are only for identification purpose. In addition, we use the Galactic coordinate grid as reference: from A to F for Galactic longitude and from 1 to 6 for Galactic latitude, both in steps of 0.15°, for a total of 6 × 6 sub-regions. We use the standard astronomical coordinates convention to localize objects inside each sub-sector; for example, the H ii region in [N-R] will be indicated as [C2-West]. All the figures in this paper use the Galactic coordinate convention as in the figure above, unless otherwise stated. The region in the upper left, in grey, is a superposition between the North–East extremity of the molecular ring and gas that seems to come from the direction of the RCW106 cloud and is not directly connected to the CO ring. The astrophysical objects are plotted using data from simbad (Wenger et al. 2000) database and ds9 software (http://ds9.si.edu/site/Home.html).

Open in new tabDownload slide

2 OBSERVATIONS AND DATA REDUCTION

2.1 CO, [C i], and H i

In this paper five spectral lines will be considered: the three CO isotopologues ¹²CO, ¹³CO, C¹⁸O J = 1 → 0 transition (hereinafter called simply ¹²CO, ¹³CO, and C¹⁸O), the atomic carbon [C i] ³P₂ → ³P₁ transition line at 809.342 GHz (hereinafter [C i]), and the H i 21 cm line.

The CO lines come from the isotopologue survey of ¹²CO, ¹³CO, and C¹⁸O, J = 1 → 0 transition lines (30 arcsec pixel size and |${\sim }0.1 \rm \, km\, s^{-1}$| spectral resolution), obtained using the 22|$\rm \, m$| single dish Mopra millimeter wave telescope, part of the Mopra Southern Galactic Plane CO Survey of the Milky Way (Burton et al. 2013, 2014; Braiding et al. 2015; Rebolledo et al. 2016; Blackwell, Burton & Rowell 2017; Braiding et al. 2018). In this survey we used the fast-on-the-fly mapping that maps the four CO isotopologues at the same time, taking 8 × 256 ms samples each cycle (2.048 s). In Galactic coordinates it covers |$\vert \mathit {b}\vert \lt 1^{\circ}$| and 265° < ℓ < 10°, each square degree is divided into 20 rectangles of 60 arcmin × 6 arcmin, 10 aligned along lines of Galactic longitude, and 10 aligned with Galactic latitude. Each stripe required ∼1 h to be completed, at a scanning speed of ∼30 arcsec s⁻¹. The CO data presented in this paper, relative to the G332 region, was taken during the first half of 2012 June. The C¹⁷O line was included in the survey but did not show any detectable emission and will be not considered further.

The [C i] data (72 arcsec beam FWHM, pixel size of 2 arcmin and |$0.5 \rm \, km\, s^{-1}$| spectral resolution) comes from the High Elevation Antarctic Telescope HEAT¹ (Kulesa 2011; Burton et al. 2015) located at Ridge A, a deep-field site close to the highest dome on the Antarctic plateau, |${\sim }1000\rm \, km$| from the South Pole Station.

The H i data come from the Parkes+ATCA Southern Galactic Plane Survey (SGPS, McClure-Griffiths et al. 2005) and has a voxel size of 40 arcsec and a spectral resolution of |$0.8\rm \, km\, s^{-1}$|⁠. Table 1 summarizes the characteristics of the surveys we have used.

2.2 CO data reduction

Here is given a brief description of the pipeline used to reduce CO data. The raw data in rpfits² format are processed using livedata³ to subtract the reference position emission. An idl⁴ routine then flags bad data values above a given threshold (usually 5σ). Data are then gridded using the gridzilla³ package and cleaned again using routines in idl. The original data reduction pipeline, described in Burton et al. (2013), is used in this work with the only addition of the flagging routine between the livedata and gridzilla steps. The final output is a square degree fits datacube with angular dimensions of 131 × 131 pixels, and a velocity range of (⁠|$-582 \rm \, km\, s^{-1}$| to |$500 \rm \, km\, s^{-1}$|⁠), (⁠|$-510 \rm \, km\, s^{-1}$| to |$245 \rm \, km\, s^{-1}$|⁠) and (⁠|$-515 \rm \, km\, s^{-1}$| to |$215 \rm \, km\, s^{-1}$|⁠) for ¹²CO, ¹³CO, and C¹⁸O, respectively. All the CO datacubes were then regridded, using the miriad⁵ command regrid⁶, to the C¹⁸O grid, which is (marginally) the coarser spectral resolution compared to ¹²CO. The final carbon monoxide datacubes have is composed by voxels an angular voxel dimension of 30 arcsec and a spectral resolution of |$0.092\rm \, km\, s^{-1}$|⁠.

2.3 Near and far-infrared emission

To correlate our CO data to dust emission we used two surveys of the near-infrared emission coming from the G332 sector: the Galactic Legacy Infrared Midplane Survey ExtraordinaireGLIMPSE (Benjamin et al. 2003; Churchwell et al. 2009) and the |$24\,\hbox{$\mu $m}$| band MIPSGAL (Carey et al. 2009; Gutermuth & Heyer 2015) observed by the Spitzer space telescope.⁷ On the longer wavelength side, to derive the cold dust emission temperature profile, we considered the Herschel Infrared Galactic Plane Survey (Hi-GAL, Molinari et al. 2010), which is an Open Time Key Project of the Herschel Space Telescope covering the Galactic Plane in a 2° wide strip (⁠|$\vert \mathit {b}\vert \lt 1^{\circ}$|⁠) in five photometric bands centred at |$70, 160, 250, 350,\ \text{and}\ 500\,\hbox{$\mu $m}$|⁠.

3 THE G332 REGION

The H i and ¹³CO average emission spectrum coming from the whole G332 square-degree sector is shown in Fig. 2, together with Gaussian curves used to fit the CO emission along the line of sight (see Table 2 for line parameters). The Gaussian curves were obtained using the scouse⁸ software (Henshaw et al. 2016).

Figure 2.

Left-hand image: Average spectral profile plot along the V_LSR interval |$[-105,-10]\rm \, km\, s^{-1}$| of |$\mathrm{T}_{\mathrm{A}}^*$|¹³CO (black curve) and H i (blue dashed curve) for the entire G332 square-degree sector, which encloses the region shown in Fig. 1. H i is in T_mb units and reduced by a factor of 200. The red curve is the sum of seven Gaussian curves (displayed in different colours) derived from the multi-Gaussian fit of the ¹³CO average emission. See Table 2 for the parameters of each fit. Right-hand image: 2D heat map scatter plot comparing the second moment map (velocity dispersion) versus the first moment map (velocity centroid) of each line of sight belonging to the ¹³CO datacube, binned by four channels along the velocity axis in the same velocity range as considered in the left-hand panel.

Open in new tabDownload slide

The right-hand panel in Fig. 2 is a 2D heat map scatter plot⁹ between the gas velocity dispersion and the first moment map, computed along the interval |$[-115,-10]\rm \, km\, s^{-1}$|⁠. The pixel distribution shows a branching pattern that is visible at coordinates around |$[22,-70]\rm \, km\, s^{-1}$| and has nodes at |$[12.5, -85]\rm \, km\, s^{-1}$| and |$[10,-50]\rm \, km\, s^{-1}$|⁠.

Few pixels have the largest velocity dispersion with velocity centroid in the interval −60 <V_LSR|$\lt -80\rm \, km\, s^{-1}$|⁠, while most of the emission has a second moment below |$10\rm \, km\, s^{-1}$| and a first moment constrained to −55 <V_LSR|$\lt -40\rm \, km\, s^{-1}$|⁠. Small localized ¹²CO and ¹³CO emissions, missing in the square-degree average profile (Fig. 3), were detected at positive velocities |${\sim }5\rm \, km\, s^{-1}$| and at |${\sim }-20\rm \, , -7\rm \,$|⁠, and |$0\rm \, km\, s^{-1}$|⁠, but this will not be considered in our analysis.

Figure 3.

Average CO, H i, and [C i] spectral profiles for the ring region shown in Fig. 1. This differs from Fig. 2 in considering the sky region 332.1° < ℓ < 332.8° and −0.30° < b < 0.45°. The two vertical red lines indicate the spectral interval in which the molecular ring is identified. H i and ¹²CO are plotted reduced by a factor of 100 and 3, respectively, to allow a clear comparison with other lines. The CO and [C i] peak is at V_LSR|${\approx }-50\rm \, km\, s^{-1}$|⁠, while, at the same velocity, H i presents a local minimum.

Open in new tabDownload slide

3.1 The G332 giant molecular ring morphology

The average spectral profiles of the CO lines, and [C i], show a peak (left-hand panel of Figs 2 and 3) at V_LSR = |$-50\rm \, km\, s^{-1}$|⁠, where a molecular cloud is seen in both atomic and molecular carbon. In order to identify the feature linked to it, we investigated the emission distribution and produced a 3D model¹⁰ of the ¹³CO datacube between |$-55\lt \text{$V_{\mathrm{LSR}}$ }\lt -30\rm \, km\, s^{-1}$|⁠.

Inspection of the CO spectral line and the sigma threshold masks (see Section 5.1) led us to focus our study on Galactic coordinates 332.0 < ℓ < 332.8, −0.3 < b < 0.4, within which we identified the principal emission structure (the ring), and on the velocity range |$-55 \lt \text{$V_{\mathrm{LSR}}$ }\lt -44\rm \, km\, s^{-1}$| (which we refer to as the ring velocity range). This choice was made to prevent contamination coming from gas not evidently connected to the ring or with a very low signal-to-noise ratio.¹¹ The extent defined above is the same used to compute moment maps (as the moment zero in Fig. 4) and all the plots belonging to the ring, unless otherwise specified.

Figure 4.

Integrated intensity maps of CO lines (⁠|$\mathrm{T}_{\mathrm{A}}^*$|⁠) and C i (T_mb) over the velocity range |$-55\, \mathrm{km\, s}^{-1} \lt$|V_LSR|$\lt -44\, \mathrm{km\, s}^{-1}$|⁠. Axes are Galactic with the x and y indicating longitude and latitude, respectively.

Open in new tabDownload slide

The region delimited by coordinate ℓ > 332.6° and |$0.4^{\circ }\lt \mathit {b}\lt 0.15^{\circ}$| (see Fig. 1), where the North–East extremity of the ring and another cloud extremity coming from the East fall in the same ring V_LSR range, is excluded from our study, since the two clouds cannot be clearly separated.

Taking into account the above mentioned constraints, the ring angular dimensions are about 0.8° along both South–North and East–West directions, showing mean velocity FWHMs approximately equal to 9, 5, 4, and |$8.5\rm \, km\, s^{-1}$| for the ¹²CO, ¹³CO, C¹⁸O, and [C i] average line profiles, respectively.

4 THE KINEMATIC DISTANCE AMBIGUITY

We estimated the distance of the ring by using the Galactic rotation model from McClure-Griffiths & Dickey (2007), resolving the KDA by comparing the SGPS HI survey data to the Mopra CO and HEAT surveys. Velocities profiles in Fig. 3 show a local H i minimum in correspondence of CO and [C i] peaks at around |$-50\rm \, km\, s^{-1}$|⁠. Inspecting the CO and H i spectral lines we found a clear decrease in H i where CO peaks, for all of line of sights passing through the ring, indicating the presence of a cold atomic component at the same velocity of the CO peak. This strengthens the interpretation of the ring as a unique connected structure. The near distance solution for V_LSR = |$-50\rm \, km\, s^{-1}$|⁠, using a ring geometrical centre (ℓ, b) = (332.3°, 0.0°), is |${\sim }3.73\rm \, kpc$| from the Sun. Assuming the Galaxy model of Vallée (2014), this distance places the molecular ring inside the Scutum-Crux arm (Fig. 5). For comparison we also used the Bayesian approach¹² of Reid et al. (2016) that gave a distance of |$3.22\pm 0.44\rm \, kpc$|⁠. However, in our analysis we will adopt the 3.73 kpc value.

Figure 5.

Colour representation of the Galactic velocity field for the 330°–335° longitude range, based on the rotation curve and Galaxy model described in Section 4. The numbers along the line of sight indicate the distance from the Sun in kpc. The Solar circle is indicated by the dashed circle while the orange arc indicates the tangent points to the spiral arms.

Open in new tabDownload slide

In Fig. 6 we present the Galactic spiral arms overplotted with the Position Velocity (PV) image of the ¹²CO average emission from the region shown in Fig. 1. It is seen that most of the CO is constrained inside the Norma near, Norma far, and Scutum-Crux near arms, while in the Sagittarius near and Scutum-Crux arms, there are only a few small clouds. The position ambiguity between Scutum-Crux near and Norma far is clearly visible as they are superimposed.

Figure 6.

PV spiral arms plot, based on the Galactic rotation curve and Galaxy model as described in Section 4, over the ¹²CO average intensity PV plot image. The solid lines represent arm mean positions, while segmented lines represent their outer edges. The colour codes for the spiral arms are the same as in Fig. 5.

Open in new tabDownload slide

The analysis of the ¹²CO spectral lines revealed emission peaks between |$-20\rm \, km\, s^{-1}\lt$|V_LSR|$\lt 0\rm \, km\, s^{-1}$|⁠, coming from small weak clouds apparently localized in the Scutum-Crux far arm and in the Sagittarius near arm. Nevertheless, no local H i minimums are present at the same V_LSR of these ¹²CO peaks, suggesting a far distance solution for their KDA of the order of |${\sim }14\rm \, kpc$|⁠.

5 MOMENT MAPS

5.1 Mask for the moment maps

To improve the detection of the faintest CO emission (as in the case of C¹⁸O), we used the miriad task imbin¹³ to bin the original datacubes along the velocity axis by a factor of 4, consequently degrading the spectral resolution to |${\sim }0.35\rm \, km\, s^{-1}$|⁠. To separate real emission from noise we applied, on the binned cubes, a custom idl routine based on the method described by T. M. Dame.¹⁴ Selecting the velocity range, and the spatial region, the extracted cube is 3D smoothed by a boxcar average of a certain width.¹⁵ For CO and [C i] we used a boxcar kernel width of [3,3,2] voxels: the average is computed considering three voxels in each angular direction and two along the spectral axis. The rms noise of the smoothed cube, σ_s, is used to select only voxels above a σ_det = σ_s × ϕ_det threshold. The selection is then expanded including all the surrounding voxels above the σ_ext = σ_s × ϕ_ext level. This growing mask procedure is similar to the masking method used in the cprops algorithm (Rosolowsky & Leroy 2006). Both the selection and the expansion steps are done on the smoothed cube. All voxels that pass these two criteria form the mask applied to the original cube to compute the moment maps. For ¹²CO, ¹³CO, and C¹⁸O the chosen masking threshold values are [6,3], written in the form [ϕ_det, ϕ_ext]. In this paper we will always refer to this kind of 3D mask coming from a four-channel binned datacube (specifying from which line originates), if not otherwise stated. The same values of smoothing and threshold are used for [C i] and HI. In Fig. 4 are shown the resulting moment maps for CO and [CI]. We opted a boxcar instead of a Gaussian smooth after performing a series of tests applying a custom genetic algorithm (Holland 1992) to find the best combination of the two threshold and two smoothing parameters.

To simplify the description of this algorithm we will take advantage of terms usually used in genetics. The algorithm basically creates a certain number (population) of moment map individuals from a different choice of the above mentioned four parameters (four genes or genotype), randomly selected from a defined range. To each map (individual) is assigned a score through a custom defined function (fitness function). In our definition the best individuals have the higher fitness function. This function takes into account the number of voxels with negative values included in the 3D mask, the number of voxels of the mask, the number of negative and positive pixels in the final zeroth moment map. In detail, the fitness function is defined as f_fit = (P⁺/1000 − P⁻) + (V⁺ − V⁻)/V_N, where P⁺ and P⁻ are, respectively, the number of positive and negative value pixels in the resulting 2D zeroth moment map, while V⁺, V⁻, and V_N are, respectively, the number of positive, negative, and total voxels included in the 3D mask. The major weight given to P⁻, with respect to P⁺, is to greatly decrease the fitness function for individuals having negative value pixels in the final zeroth moment map. As a consequence, these individuals will have minimal chances of passing their genotype to the next generation. The parameters (genotype) of the elements of the population with the best score are mixed together (crossover) to create a new set of parameters (sons genotype) introducing a small mutation (mutation), thus creating a second generation of the parameter set. When the maximum value of the fitness function does not change for a certain number of iterations, prior to stopping the process, the algorithm increases the variation in the sons parameters (mutation amplification) for a maximum number of iterations, in order to prevent the algorithm stopping on a local maximum, exploring a wider parameter space. If a new individual is found the process continues with the mutation rate set to the initial value.

The integrated flux coming from the masked cubes was found to be the |$51$|⁠, |$70$|⁠, and |$55{{\ \rm per\ cent}}$| of the integrated flux coming from the non-masked cubes, respectively, for the ¹²CO, ¹³CO, and C¹⁸O lines.

Comparing the ¹²CO and the ¹³CO integrated intensity maps, it is possible to see regions in which there is no ¹²CO counterpart of ¹³CO emission, in particular around (ℓ, b)∼(332.3°, 0.2°). This apparent cavity in [NW-R] is close to PLW2012 G332.277+00.189-048.9 (Purcell et al. 2012), a dense core at (ℓ, b,) ≈ (332.274°, 0.187°). This is a threshold effect of the 3D mask: in this region ¹²CO emission in the smoothed cube is below the masking threshold while ¹³CO is above. Examining the line profile (Fig. 7) this behaviour can be explained as ¹²CO self-absorption, since the local minimum of the ¹²CO spectral lines is close to the ¹³CO and C¹⁸O local peaks. However, a deeper inspection of all the ring ¹²CO spectra reveals that this effect is limited to few lines of sight, mainly distributed in the Northern part of the ring, besides the above mentioned region.

Figure 7.

Single pixel spectral line profile of the line of sight (ℓ, b) ≈ (332.3°, 0.2°). ¹²CO shows self-absorption, while the ¹³CO and C¹⁸O show a peak close to a ¹²CO minimum. In black is ¹²CO while in red and blue are ¹³CO and C¹⁸O, respectively. The two dotted horizontal lines represent the 3σ threshold (black for ¹²CO and red for ¹³CO) used to derive the mask for the moment maps. The ¹²CO is thresholded out even though its emission is stronger than ¹³CO.

Open in new tabDownload slide

5.2 Integrated intensity

Integrated intensity maps of [C i] and all CO lines are shown in Fig. 4. All the lines show the ring shape, with ¹²CO, ¹³CO, and [C i] being most similar in shape. Thicker structures seen in C¹⁸O are distributed along an arc-like structure interrupted by a few cavities. Here we will briefly describe the main features:

• ¹²CO, ¹³CO, and [C i] are higher in [C-R], [SE-R], and [SW-R]. The first region covers a system of multiple bubbles [CPA2006] S52 that encloses [CPA2006] S53, and [SPK2012] MWP1G332437+000547 (Churchwell et al. 2006; Simpson et al. 2012), presenting a bright |$8\,\hbox{$\mu $m}$| emission enclosing diffuse |$24\,\hbox{$\mu $m}$| emission and no significant C¹⁸O detection.

• [S-R] and [SW-R] host H ii regions (Urquhart et al. 2014; see Fig. 1) with velocities compatible with the ring velocity interval: AGAL332.544-00.124 at V_LSR = −46.7 km s⁻¹, close to the highest ¹²CO integrated emission and AGAL332.296-00.094 at V_LSR = −48.2  km s⁻¹, near to a ¹³CO and a C¹⁸O integrated emission peak.

• ¹³CO and C¹⁸O maximum integrated emission is at (ℓ, b)∼(332.40°, −0.09°), close to a radio sub-mm source AGAL G332.389-00.091 (Contreras et al. 2013) and a Young Stellar Object (YSO) candidate SSTGLMC G332.3810-00.0992 (Robitaille et al. 2008).

• C¹⁸O distribution presents a number of dense spots in the southern part, in which there are prestellar and protostellar cores ([SE-R]) as well as dense cores ([SW-R]). The [N-R] shows lower emission, but as in the southern regions there are some YSO candidates (see Fig 1).

The major contribution to the overall ring emission comes from the southern ring that seems to hosts more sites of star formation activity (see Section 10.2) compared to [N-R], [NW-R], and [W-R] regions. As expected, among all the CO lines, ¹²CO emission is the most widespread. The apparent absence of ¹²CO emission in [NW-R] where ¹³CO is present is explained in Section 5.1.

5.3 Ring velocity field

The ring velocity field is shown in Fig. 8. The Southern ring emission comes mainly from V_LSR|$=-50\rm \, km\, s^{-1}$|⁠, and is |$3\, \mathrm{ to} \, -4\rm \, km\, s^{-1}$| wide (see Fig. 9). The Eastern part of [SE-R] has velocity centroid shifts of about |$+5\rm \, km\, s^{-1}$| reaching V_LSR|$=-44\rm \, km\, s^{-1}$| for all the lines, equal to a velocity gradient of |$0.30\rm \, km\, s^{-1}\, \mathrm{pc}^{ -1}$|⁠. In the same way, the [C1] region shear is about |$0.28\rm \, km\, s^{-1}\, \mathrm{pc}^{ -1}$|⁠.

Figure 8.

Velocity centroid maps of the emission for each line derived from the first moment map.

Open in new tabDownload slide

Figure 9.

Velocity dispersion distribution of the molecular ring for the four emission lines derived from second moment maps.

Open in new tabDownload slide

[N-R] and [W-R] are sheared by about |${\sim }2\rm \, km\, s^{-1}$| with respect to the Southern part. Differently from [W-R], in which the velocity field is almost constant at V_LSR|${\approx }-50\rm \, km\, s^{-1}$|⁠, [SW-R] has a more complex velocity field, which can be related to the objects in the regions (see Fig. 1).

In [C-R] the first moment map shows a rapid shift of about |$5\rm \, km\, s^{-1}$|⁠; as said in Section 5.2, this region is associated with many bubbles. Here, ¹²CO and ¹³CO have broader emissions (see Fig. 9) and show spectral lines in which it is possible to identify two peaks separated by a ΔV_LSR|${\approx }5.5\rm \, km\, s^{-1}$|⁠.

At coordinates (ℓ, b) = (332.4°, 0.16°) a velocity gradient ΔV_LSR = 1.5 |$\rm \, km\, s^{-1}$|⁠, 3 arcmin in diameter, is only visible in ¹²CO and ¹³CO emissions, and the [C i] resolution is insufficient to distinguish it. In the literature (Peretto & Fuller 2009) are reported two dark clouds close to this position SDC G332.383+0.182 and G332.419+0.174 and Beichman et al. (1988) an IR source IRAS 16111-5032. In the same position, ¹²CO and ¹³CO spectral lines indicate a dip in the ¹²CO emission in correspondence of a ¹³CO peak, a scenario compatible with ¹²CO self-absorption.

Comparing the first (Fig. 8) and the second moment maps (Fig. 9) it is possible to identify, for all the spectral lines except C¹⁸O, sharp value changes in [D4] and [D6]. This can be explained as a partial superimposition of cloud fragments detected at a V_LSR greater than the main ring V_LSR . The second moment maps (Fig. 9) show a pattern in [SW-R] that confirms this interpretation: an increase in the velocity dispersion where the first moment maps are suggesting the superposition between the fragments and the ring emission. The same explanation applies to ¹²CO in [D1], where both the first and second moment maps show a sharp change in value. In this case, the ¹²CO emission of the ring cloud is extended towards lower velocities.

To better understand the physical connection between these extensions and the main ring, we made use of VR and AR representations¹⁰ (described in Appendix  A), which allows us to consider them as part of the ring complex.

6 CO AND [C i] COLUMN DENSITY AND OPTICAL DEPTH

In order to take in account the response of the Mopra radio telescope, the measured signal (⁠|$T_{A}^{*}$|⁠) was converted into the corrected brightness temperature T_MB, using the extended beam efficiency η_xb = 0.55 at 115 GHz (Ladd et al. 2005), since the sources considered in this work are more extended than the Mopra main beam. The HEAT datacube is in T_mb. In Fig. A1 we report the column density maps derived for ¹³CO, C¹⁸O, and [C i] using different T_ex values. Our approach to derive ¹²CO, ¹³CO, C¹⁸O, and [C i] column densities is based on assuming LTE conditions, and same LTE temperature for all the CO isotopologues. It is important to note that the latter assumption is not verified in the presence of temperature gradients.

Besides LTE conditions, we considered a beam filling factor equal to unity, optically thick ¹²CO, and optically thin ¹³CO and C¹⁸O emissions. The last hypothesis was verified by inspecting voxel-by-voxel the C¹⁸O/¹³CO ratio, which was found to be in the range [0.2–0.25] in regions where the C¹⁸O is bright enough to yield a significant ratio value within the noise level. It is important to note that this ratio depends also on the oxygen ¹⁸O abundance relative to ¹⁶O as well.

In the next paragraphs we describe in detail all the steps followed to compute optical depth and column densities.

6.1 ¹³CO and C¹⁸O

Figure 10.

Map of T_ex as derived from the ¹²CO peak emission for each line of sight. The maximum value is ∼35 K.

Open in new tabDownload slide

Figure 11.

Dust temperature map derived from the HiGAL data fit described in Section 10.1. The black contour represents the edge of the ¹³CO integrated intensity as plotted in the ¹³CO panel of Fig. 4.

Open in new tabDownload slide

We applied the same approach used for ¹³CO to the C¹⁸O column density and optical depth computation. To derive the relative column density, we adopted the value from Frerking et al. (1982) of |$\mathrm{[H_{2}/C{^{18}}O]}=5.88\times 10^6$|⁠, an order of magnitude less abundant than ¹³CO.

6.2 ¹²CO to H₂ column density

6.3 [CI] ³P₂ → ³P₁

Figure 12.

Map of [C i] optical depth peak values for each line of sight, derived for a T_ex = 30 K. The highest values are in correspondence of the highest T_ex derived from the ¹²CO peak emission, as shown in Fig. 10, and of the highest dust temperature as plotted in Fig. 11.

Open in new tabDownload slide

We followed the same method used in the CO optical depth calculation shown in equation (4). Differently to CO we choose two a priori excitation temperatures: T_ex = 30 and 50 K, both of them higher than the typical CO T_mb to take into account that [C i] exists in warmer physical conditions (e.g. the surface regions of molecular clouds, or in the neighbourhood of intense radiation sources). We find comparing to T_ex = 50 K, N(CI) increases by a factor of ∼4 and decreases by |${\sim }70{{\ \rm per\ cent}}$|⁠, when using T_ex = 20 and 80 K, respectively.

Inspecting the [C i] data we found that the emission is coming from both the external layers around the ring and from the inner part, in a web pattern, in particular in the southern ring. This indicates that localized sources of Far-UV (FUV) radiation are present in the cloud interior. In Section 10.2 we will discuss further the behaviour and interpretation of the [C i] emission.

6.4 Column density sensitivity

We derived the CO and [C i] column density 1σ sensitivity by applying the same approach described in Section 6 to an emission-free region |$5\rm \, km\, s^{-1}$| wide (equal to 14 channels in the four-channel binned CO datacube), coherently with the typical spectral width of the lines along the ring. We used a different velocity interval [−130 to |$-120\rm \, km\, s^{-1}]$| for atomic carbon, since the HEAT datacube has a smaller spectral range than the CO data (see Table 1). The resulting column densities and relative N(H₂) 1σ levels are shown in Table 3 for different T_ex values; in Table 4 we report N(H₂) quantities derived from different tracers.

6.5 Dust spectral energy distribution

6.6 Dust to H₂ column density

The map of N(H₂) dust is shown in Fig. 13, overlaid by the ¹³CO moment zero map, computed considering the V_LSR range [−115 to 10 |$\rm \, km\, s^{-1}$|].

Figure 13.

H₂ column density map derived from the modified blackbody fit to the Hi-GAL data. The black contours represent the ¹³CO integrated intensity relative to the V_LSR interval |$[-115\ \mathrm{to}\ 10\rm \, km\, s^{-1}]$|⁠. The contours steps are 25, 50, 75, and 100 per cent of the maximum value of |$30\rm \, K\, km\, s^{-1}$|⁠. Scale bar shows the column density in units of cm⁻².

Open in new tabDownload slide

In order to seek a correlation between N(H₂) dust and total CO emission along the whole line of sight, in Fig. 14 we shown a scatter plot between these two quantities. Depending on the correlation between N(H₂) dust and ¹³CO integrated intensity, two groups of points were identified: the red ellipse group (A) possesses an higher correlation compared to the black ellipse group (B). The linear fit overlaid on the scatter plot is computed on group A points only. In doing so, we are assuming N(H₂) traced by ¹³CO is proportional to ¹³CO integrated intensity. We used a 2D heat map representation to show pixel values for density distribution, as shown in the scale bar in Fig. 14 (further discussed in Section 10.1).

Figure 14.

2D heat map scatter plot between N(H₂) dust and ¹³CO integrated intensity along the whole V_LSR emission range |$[-115\ \mathrm{to}\ 10\rm \, km\, s^{-1}]$|⁠. We distinguished two groups of points. Group A (red ellipse) showing a greater correlation between N(H₂) dust and ¹³CO integrated intensity compared to the group B (black ellipse). They will be further discussed in Section 10.1. The fit line in black is computed only on group A points, fit parameters are shown at the top, |$\mathrm{N(H_2)_{dust}=m\cdot F_{^{13}CO}+b}$|⁠.

Open in new tabDownload slide

6.7 Column densities distribution

Figure 15.

N(H₂) histograms (x axis are in logarithmic scale) of N(H₂) column densities traced by ¹³CO for different T_ex as indicated on each graph. Histograms are derived from voxels belonging to the 3D mask resulting from the intersection between ¹²CO and ¹³CO relative masks. The red dashed lines indicate the column density sensitivity limits in units of Log[cm⁻²]. Their values are 21.2 and 20.9, respectively, for T_ex = |$30$| and |$10\rm \, K$|⁠.

Open in new tabDownload slide

7 [C i] AND Dark MOLECULAR GAS

Examining the images, in Fig. 16 further we see that, in general, the dark gas fraction possesses lower values on lines of sights passing through the innermost part of the molecular cloud, while increasing towards the edges of the CO ring. In particular, the lowest values are located in [NW-R]. This is consistent with an increased fraction of dark gas in molecular cloud surfaces where CO is photodissociated.

Figure 16.

Dark molecular gas fraction map across the molecular ring. We used a fixed T_ex = 30 K for the [C i] column density while for CO the used two T_ex values as indicated above each panel. The scale is the same for both pictures.

Open in new tabDownload slide

Since column density increases almost linearly with the assumed T_ex (for excitation temperatures above the energy of the considered emission line), we found lower f_DG values for N_CO computed at T_ex = 20 K than at T_ex = 10 K. In particular the average f_DG at T_ex = 20 K is 16 per cent, while at T_ex = 10 K it is 18 per cent. The latter is similar to the f_DG computed assuming a T_ex derived from the ¹²CO peak temperature map (Fig. 10). The molecular dark gas fraction decreases if we instead assume [C i] T_ex = |$50\rm \, K$|⁠. In this case the average f_DG values are equal to |$10{{\ \rm per\ cent}}$| for a CO T_ex = |$10,20\rm \, K$|⁠, or derived from the ¹²CO peak temperature map. Comparing with other estimates, our average values are nearly half of the value estimated in Burton et al. (2015), which gave a value of |${\sim }30{{\ \rm per\ cent}}$| for the average dark molecular gas fraction, for a cold quiescent molecular cloud in G328.

7.1 X_co factor

We inferred the X_co factor by comparing the ¹²CO integrated intensity map with the N(H₂) map obtained from ¹³CO computed at different T_ex. In the derivation of these maps, we adopt for the 3D mask the intersection of the single 3D masks coming from ¹²CO and ¹³CO (see Section 5.1), in order to ensure comparison of the emission from the same voxels.

In Fig. 17 is shown the ratio between the X_co derived from different T_ex and the X_co computed at T_ex = 10 K, indicated as X_CO(T₁₀). In doing so, we used the T_ex values of 20, 30, |$50\rm \, K$|⁠, and the T_ex derived from the ¹²CO peak emission map for each line of sight, indicated in the plot as T_ex = T_map. We note that the X_co factor is nearly inversely dependent on T_ex. In particular, taking as reference the X_co derived at T_ex = |$10\rm \, K$| (for which the median value is |${\approx }2.1\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠), the X_co at T_ex = |$50\rm \, K$| is three times higher at lower X_co values while for higher X_co the ratio decreases to 2. Further, the scatter increases increasing X_co. A similar trend is found for all assumed values of T_ex.

Figure 17.

X_co factor distribution comparison (derived as described in Section 7.1), y axis is in the form X_co(T_ex)/X_co(T₁₀), where the excitation temperature |$\mathrm{T_{\mathrm{ex}}}$| is equal to 20, 30, and 50 K, respectively (T₁₀ is for T_ex = 10K), T₁₀ median value is |$2.2\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠. T_map means that column density calculation was based on the pixel-by-pixel T_ex map (Fig. 10), derived for each line of sight from the ¹²CO peak emission.

Open in new tabDownload slide

In Fig. 18 we plotted the angular distribution of X_co for T_ex = T_map, the median value is |$2.2\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠. Small scale (compared to cloud extension) X_co variations are present in almost all the cloud, up to a factor of 2. In general, there is no direct relation between X_co and integrated intensity (Fig. 4). An exception is in [C2], [D2], and [E2], where there are evidences of an inverse proportion between X_co and ¹²CO integrated intensity. This is not unexpected since ¹²CO self-absorption in these regions (see last part of Section 5.1).

Figure 18.

X_co factor local variation map derived from the ratio between the ¹²CO integrated intensity and the N(H₂) computed from ¹³CO for a T_ex derived from the ¹²CO peak temperature map.

Open in new tabDownload slide

In Fig. 19 is illustrated the corresponding histogram, showing that the values are mostly (⁠|${\sim }80{{\ \rm per\ cent}}$|⁠) in the range between 1.7 × 10²⁰ and |$3.5\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠.

Figure 19.

Histogram of the X_co distribution derived from a T_ex computed from ¹²CO peak emission map as in Fig. 18. The median value is |${\approx }2.2\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠.

Open in new tabDownload slide

A deeper discussion on X_co variation across the ring molecular cloud is presented in Section 10.3.

7.2 X|$\mathrm{_{CI}^{809}}$| factor

The linear fit m · F_CI, being F_CI the [C i] integrated intensity in |$\rm \, K\, km\, s^{-1}$|⁠, gives for X|$\mathrm{_{CI}^{809}}$| the values of (1.77 ± 0.03) × 10²¹, (1.83 ± 0.02) × 10²¹, and |$(1.97\pm 0.02)\times 10^{21}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠, respectively, for N(H₂) traced by ¹²CO, ¹³CO without adding dark gas and ¹³CO adding the dark gas fraction.

We report the value of |$X_{\mathrm{CI}}=1.10\times 10^{21}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$| by Offner et al. (2014) and |$X_{\mathrm{CI}}=1.01\times 10^{21}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$| obtained from the numerical simulation of Glover et al. (2015) that were derived for the CI emission at 609 μm, about half of the value we derive from the data.

8 CO AND [C i] LINE RATIOS

8.1 evans plot

In the following, we present ratios between the ¹²CO, ¹³CO, and [C i] lines using a signal-to-noise (SNR) visual discriminator type of graph known as an evans¹⁷ plot, where high SNR pixels are more evident than lower ones. Here we give a brief description on how to read evans plot taking as reference Fig. 20 (see also Burton et al. 2015). Each plot consists of two pictures, one showing data (left-hand panel) and one showing a Hue Vs Saturation colour table (right-hand panel). Each pixel in the data map has hue and saturation values, which are codified into SNR and data values (the ratio value in this case) using the colour table, according to the axes scales. In this way, those pixels with high SNR values are more evident with respect to lower SNR pixels, which fade into grey independently from their ratio value, giving them a lower visual weight.

Figure 20.

evans plot of the ¹²CO/¹³CO integrated intensity map over the ring V_LSR range. Left-hand panel – Each pixel in the ratio map possesses hue and saturation values that are decoded to SNR and ratio values using the colour table on the right-hand panel (see 8.1 for details on evans plot). In white is the contour representing ¹³CO integrated emission at |$3\rm \, K\, km\, s^{-1}$|⁠.

Open in new tabDownload slide

8.2 ¹²CO/¹³CO ratio

In Fig. 20 is shown the ratio between integrated intensity map of ¹²CO and ¹³CO, over the ring spectral channel range. Each zeroth moment map was derived applying the same ¹³CO 3D mask to both datacubes. The ratio exhibits almost everywhere a value >1, increasing towards the cloud edge respect to innermost regions, in which it assumes values in the range [2–4]. A small region located in [D2], around (ℓ, b) = (332.3°, 0.2°), shows a ratio <1; this could be explained by ¹²CO self-absorption (Fig. 7).

8.3 [C i]/¹³CO ratio

The C abundance is expected to increase in the external layers of GMCs where the FUV radiation dissociates the CO, so that the [CI]/CO ratio may be greater than in the innermost parts of the cloud. To investigate the [CI]/¹³CO in the ring we regrid the ¹³CO cube to the [C i] coarser resolution and applied to both the same [C i] 3D mask. In Fig. 21 we present the integrated intensity maps ratio between [C i] and ¹³CO.

Figure 21.

evans plot of the integrated intensity map [CI]/¹³CO ratio along the ring V_LSR velocity range. White contour is ¹³CO integrated intensity at |$2\rm \, K\, km\, s^{-1}$|⁠.

Open in new tabDownload slide

The ratio between the zeroth moment maps possess values <1 along almost all the lines of sight passing through the thickest region of the molecular ring, with lower peaks in [NW-R] and [N-R]. Ratio values are >1 at the edge of the cloud subregions [SE-R], [SW-R], [C-R], and [N-R], although here ¹³CO detection drastically decreases, lowering the signal-to-noise level and making pixels less distinguishable. Interestingly, some sections of the ring edge have a ratio <1, in [NW-R], in eastern [C2] and around (ℓ, b) = (332.25°, 0.0°).

To better characterize the ratio values dispersion, we explore the ratio voxels distribution along the ring velocity range considering only those included in the 3D [C i] mask (see Fig. 22, 2D heatmap scatter plot between ¹³CO and [CI]/¹³CO). The median of the ratio for the whole cloud is equal to 0.5, and its tendency is to decrease with the increase of ¹³CO emission, following a |$y_0 +\frac{a}{x-x_0}$| law, with a = 0.43 ± 5.2 × 10⁻³, y₀ = 0.27 ± 3.8 × 10⁻³ and x₀ = (8 ± 1) × 10⁻³. There are few ratio >1 voxels that come from low ¹³CO emission, the analysis of their position with respect to the molecular ring confirms that all of them are at the edge of the CO molecular ring.

Figure 22.

2D heat map between ¹³CO datacube regridded to [C i] resolution and [CI]/¹³CO voxels. The black line represents the fit in form: |$y_0+\frac{a}{x-x_0}$| (parameters are reported in Section 8.3).

Open in new tabDownload slide

9 C¹⁸O CLUMPS CLASSIFICATION

Since C¹⁸O emission is generally weak, due to the low isotopic abundance ratio compared to ¹²CO and ¹³CO, its detection is mostly likely to be optically thin. This makes the C¹⁸O line a probe through the entire molecular clouds column, sensitive in particular to the highest column density regions, when ¹³CO (and even more so ¹²CO), is optically thick. In order to investigate the highest column density clumps within the molecular ring, we decomposed it applying the clumpfind algorithm (Williams, de Geus & Blitz 1994) to the C¹⁸O emission cube, binned and smoothed in the same way as in moment maps computation (see Section 5.1). We limited our analysis to the ring velocity range; however an expansion of this work, performed on 10° of the Galactic Plane binned C¹⁸O datacube, is presented in Braiding et al. (2018). After a series of tests we selected the value of 0.35 K, equivalent to 3.5σ, for both input parameters inc and low. We set a threshold of 10 voxels on the minimum voxel number for a clump to be detected. We verified for false detection comparing C¹⁸O and ¹³CO spectral lines passing through each clump centroid. To avoid false detection of a single clump as two distinct clumps we set a minimum centroid distance equal to seven voxels; all the clumps passed this verification step. The final detections are summarized in Table A1 with their physical characteristics.

The resulting total virial mass is |$\mathrm{M^{tot}_{vir}}=4.9\times 10^4\,\mathrm{ M}_{\odot}$|⁠. In order to compare it to the total mass derived under LTE assumption |$(\mathrm{M^{20}_{LTE}}$|⁠) we applied the method described in Section 6 including only voxels belonging to a clump related to the ring. According to the different T_ex used, the total LTE clump mass traced by C¹⁸O is 0.8 × 10⁵, 1.1 × 10⁵, and 1.5 × 10⁵M_⊙, respectively, for T_ex = 10, 20, and 30 K. These values are similar to the LTE masses calculated for voxels included in the 3D smooth mask (Table 4). Comparing the total mass of the C¹⁸O clumps and the total mass of the ring derived from ¹³CO, under the LTE assumption and taking the same T_ex for both tracers, the fraction of gas in C¹⁸O clumps is 59 and |$76{{\ \rm per\ cent}}$|⁠, respectively, for T_ex = 10 and 30 K. Changing inc and low parameters in clumpfind, from 0.35 to 0.4 or 0.3 K, respectively, decreases or increases the LTE mass by |${\sim }20{{\ \rm per\ cent}}$| for all T_ex, as a consequence of smaller or bigger clumps being identified by the algorithm.

We note that the clumpfind algorithm found four clumps that are in a portion of the sky in which the C¹⁸O integrated intensity has values equal to zero. This discrepancy is due to the C¹⁸O 3D mask process creation. As described in Section 5.1, the 3D mask is derived from the smoothed cube. This operation decreases the intensity level of a weak emission extended for few voxels. As a consequence, this emission could result below the sigma threshold set on the smoothed cube in order to define the 3D mask extension. These off mask clumps are indicated with A, B, C, and D in Fig. 23 and, respectively, correspond to GMCC332.50+0.32, GMCC332.39+0.35, GMCC332.42+0.02, and GMCC332.29-0.01 in Table A1. Because of the exclusion from the 3D mask, the mass of these clumps is not present in the N(H₂) derivation as in Section 6 and reported in Table 4. Assuming a T_ex = |$20\rm \, K$| their total mass results to be 10⁴M_⊙, equal to |${\sim }8{{\ \rm per\ cent}}$| of the total C¹⁸O clump mass.

Figure 23.

Plot of Hi-GAL compact sources (see Section 10.2) with a near distance solution between [3,4] kpc (red) and C¹⁸O clump ellipsoids as reported in Table A1 (green). The black spline is the path along which are plotted the Hi-Gal dust clumps properties as in Fig. 27.

Open in new tabDownload slide

The virial parameter, defined as |$\mathrm{\alpha _{vir}}=\mathrm{M_{vir}/M^{20}_{LTE}}$| and plotted in the Table A1 (column 11), gives an indication about the dynamical state of each clumps. Typical values are in the range [0.2–0.4] suggesting that they are not being supported against gravitational collapse. Though care is needed in this interpretation since masses derived under LTE assumptions applied depend on the T_ex and C¹⁸O abundance ratio choice.

10 DISCUSSION

In the following sections we will describe a method, based on the ¹³CO emission, to separate dust emission across the Galactic spiral arms, giving an estimate of the total mass traced by dust through the ring region. We will also discuss the different physical conditions in the cloud and how these suggest a scenario in which the molecular structure shows different evolutionary stages: quiescent clouds in which star formation is occurring at its earliest stage or not yet initiated, and active clouds in which star formation is ongoing at a later stage. In doing so, we will compare the C¹⁸O with Hi-Gal clumps associated with the G332 ring cloud.

10.1 Dust across the Galaxy

As discussed in Section 6.5, the overall N(H₂) derived from the dust emission is the sum of all contributions along the line of sight. To compare it to N(H₂) traced by the ¹³CO we first consider Fig. 14, relating dust column to ¹³CO line flux. There are two sets of point here, one (group A) showing a linear correlation, the other (B) not. The former demonstrates the link between flux and column density for most regions, which we now decompose into contributions coming from separate spiral arms along the line of sight.

We note that in the N(H₂) column density traced by dust (Fig. 13, which comprises all infrared emission from this region) it is possible to find, to first approximation, a coherency between the dust and ¹³CO and ¹²CO integrated flux (Fig. 4). In particular, as shown in Fig. 13, in [SE-R] and [SW-R] ¹³CO integrated flux peaks are close to dust column peaks. The West side of the ring, [W-R], and part of [NW-R], present low integrated CO emission while the dust is bright. Bright dust, in general, follows the ring, a likely consequence of it arising from the region too. However, focusing on the ring CO and dust emission, we need to separate dust related to different Galactic spiral arms. In the following we describe our approach.

According to the average CO spectral line profile and following the spiral arm scheme adopted in this paper (Fig. 5), we separated the ¹³CO emission into four velocity ranges (R1, R2, R3, and R4), choosing the near or far solution coherently with any H i emission local minimum (see Fig. 3):

• R1: [−115 to |$-75\rm \, km\, s^{-1}$|] [intra-arm Perseus/Norma far]

• R2: [−74 to |$-56\rm \, km\, s^{-1}$|] [Norma far]

• R3: [−55 to |$-44\rm \, km\, s^{-1}$|] [Scutum-Crux near] (ring)

• R4: [−44 to |$-30\rm \, km\, s^{-1}$|] [Scutum-Crux near]

This scheme follows the position velocity (PV) plot of Fig. 6. We excluded ¹³CO emission coming from −30 to |$+10\rm \, km\, s^{-1}$| since it is localized in small structures and its integrated intensity is negligible with respect to other velocity intervals. To separate the dust emission coming from each of the spiral arms we computed, following the method of Section 5.2, for each velocity interval, a ¹³CO pixel-by-pixel moment zero map normalized to the total integrated intensity relative to the V_LSR interval [|$-115,-30\rm \, km\, s^{-1}$|] (see Fig. 24), each of which is subsequently multiplied by the N(H₂) derived from dust SED fit (Fig. 13). In this way we created four different N(H₂) dust maps (see Fig. 25 one for each spiral arm). Our approach is based on two assumptions that N(H₂) traced by ¹³CO is proportional to ¹³CO integrated intensity and that the emission from dust coming from regions where there is no CO emission is negligible (see below for H i contributions). Thus, we are able to compare a directly measured quantity (¹³CO emission) to N(H₂) dust. We also excluded the molecular hydrogen not traced by CO in this analysis.

Figure 24.

Normalized ¹³CO integrated intensity map belonging to the defined velocity ranges as in Section 10.1. Each pixel represents the multiplication factor applied to the corresponding pixel in the dust derived N(H₂) map; thus it is possible to obtain a fractional molecular hydrogen column density traced by dust for each velocity range.

Open in new tabDownload slide

Figure 25.

N(H₂) dust maps relative to each velocity interval derived from the ¹³CO normalized moment zero maps plotted in Fig. 24 multiplied by the N(H₂) dust map derived from the dust SED fit (Fig. 13).

Open in new tabDownload slide

Scatter plots, between ¹³CO integrated intensity and N(H₂) dust for each spiral arm, are shown in Fig. 26. Higher column densities belong to [R1] and [R3]. In each plot we use different colours for those pixels having a normalized ¹³CO integrated intensity map equal to 1, while the other pixels are in grey. Both of them follow a direct proportional relation but have not the same offset: grey points starts at zero while coloured pixels begins at about 10²²cm⁻² for all the ranges. Pixels belonging to the offset group comes from regions in which the normalized ¹³CO moment zero map has values equal to 1, which means the total ¹³CO integrated intensity (along the whole line of sight) has contributions coming from a single velocity interval only, without contamination from other clouds at different distances. These are those pixels in black (value equal to 1) in the normalized moment zero maps (Fig. 24); for them dust emission can be constrained in a specific velocity range since there is no other CO detection on the same line of sight.

Figure 26.

Scatter plots between normalized moment zero ¹³CO (x axis) and fractional N(H₂) dust relative to all velocity intervals (as defined in Section 10.1). For each scatter plot we use colours for those pixels whose normalized ¹³CO integrated intensity map is equal to 1 (see Fig. 24). The grey points represent all the other pixels.

Open in new tabDownload slide

In Table 6 are reported linear fit parameters for each velocity interval, considering only coloured points as in Fig. 26. The offset value is almost the same for every range, around 10²² cm⁻², this value is similar to the mean H₂ column density traced by dust in regions where no CO emission is detected along the whole Galactic velocity range (considering the ¹²CO integrated intensity map in the [|$-110, +10\rm \, km\, s^{-1}$|] V_LSR range). This could be a consequence of the infrared background emission coming from this part of the sky, [R1] shows the largest background having the largest velocity width. The fit parameters (except for [R2], but this has the lowest contribution to the total emission), are comparable to the fit parameters reported in Fig. 14. The relative contributions of each velocity range with respect to total ¹³CO integrated intensity, along the whole line of sight, are 35, 8, 47, and 10 per cent, respectively, for R1, R2, R3, and R4. Although, R3 is only |$10\rm \, km\, s^{-1}$| wide, it accounts for almost half of the total ¹³CO emission in G332.

Taking into account a near or a far distance solution, the mass traced by the disentangled dust in each velocity interval is reported in Table 7, together with the mean neutral hydrogen column density, N(HI), which is the |$13,$||$13,$||$6,$| and |$8{{\ \rm per\ cent}}$| of N(H₂) column density (derived from dust) in [R1], [R2], [R3], and [R4], respectively.

Comparing HI and CO column densities, across the −110 <V_LSR|$\lt +10\rm \, km\, s^{-1}$| velocity range, we found that the amount of atomic gas is about |${\sim }10{{\ \rm per\ cent}}$| of the column density of the molecular gas, so that the dust column density is dominated by emission from dust within molecular clouds. The total mass traced by dust across the whole sky portion defined by the ring region in the Galaxy velocity range (−110 <V_LSR|$\lt +10\rm \, km\, s^{-1}$|⁠) is ∼2.15 × 10⁶M_⊙.

10.2 Star formation activity along the molecular ring

The overview image in Fig. 1 shows the Southern side of the ring hosting several signposts of star formation with activity (masers and H ii regions) counterpoised a less active Northern side. In order to investigate this difference we compared the C¹⁸O clumps distribution (reported in Table A1) to the compact sources distribution taken from the Hi-GAL catalogue (Elia et al. 2017) classified for different stages of star evolution.

The Hi-GAL sources (hereafter called dust clumps) selected for our analysis have coordinates that fall inside the C¹⁸O clumps ellipsoid extension (with semi-axis defined by Δℓ and Δb FWHMs derived by clumpfind), and a near distance solution between 3 and 4 kpc. In order to avoid the exclusion of sources not satisfying these criterion, but potentially linked to the molecular ring, we included all the dust clumps that totally (or partially) fall in the portion of sky covered by the angular extension of the C¹⁸O integrated intensity map (see Fig. 23). The total mass of all the dust clumps considered is ≈6.8 × 10⁴ M_⊙, equal to the |$35{{\ \rm per\ cent}}$| of the ring mass traced by ¹³CO at T_ex = |$30\rm \, K$|⁠.

The L_bol/M_clump ratio is a useful indicator of the star formation activity ongoing in the hosting clump (Molinari et al. 2008; Elia & Pezzuto 2016). In order to examine the star formation activity along the molecular ring, in Fig. 27 we plot the L_bol/M_clump and T_clump values of the dust clumps following the path defined from the [E,SE-R] to the northern part of [C-R] (see Fig. 23).

Figure 27.

Plot of Hi-Gal T_clump (left-hand panel) and of ln[L_bol/M_clump] (right-hand panel) ratio derived from the bolometric luminosity of the clump and its mass (in M_⊙). The x axis is the offset distance from the initial point of the path (at coordinates ℓ ∼ 332.75° and b ∼ −0.25°) defined in Fig. 23 by the black spline.

Open in new tabDownload slide

The highest dust clump temperatures are in the Southern ring, as well as the highest L_bol/M_clump ratios, both decrease following the path. The C¹⁸O clumps corresponding to the Southern ring (as reported in Table A1) exhibit in mean, higher values in ¹³CO, C¹⁸O average temperatures and in velocity dispersion, while those relative to the Northern cloud show lower values for the same quantities. These factors suggest a difference in the star formation activity along the cloud: the dust cores in the Southern ring exhibits (on average) hotter temperatures suggesting that star formation activity is going on, while the Northern cores, colder, indicate that the star formation has not yet started, or at an earlier stage.

10.3 Local X_co variation

¹²CO (J = 1 → 0) emission is often used in the determination of the N(H₂), and total molecular mass, through the use of the X_co factor, especially for extra-Galactic sources. Thus, it is important to understand how the physical conditions of the emitting gas contribute to any variation in the X factor. Considering the cloud distance and the pixel dimension of 30 arcsec we are able to examine X_co variations on parsec scales (at |${\sim }0.5\rm \, pc$|⁠).

The approach of our X_co analysis uses equation (7), where N(H₂) is traced by ¹³CO under different T_ex assumptions, or N(H₂) traced by dust. As shown in Section 7.1, the average X_co is found to increase with the assumed T_ex used to derive the column density.

A pixel-by-pixel map of the cloud X_co values is shown in Fig. 18, the highest values are in [NW-R], and in part of [N-R] (at b ∼ 0.2° and 332.45° < ℓ < 332.15°), in [A6], in [B6], and around the position (ℓ, b) = (332.3°, −0.1°) of typically |${\sim }5\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠.

¹²CO spectral lines, where the X_co is higher, show lower ¹²CO intensity compared to ring average, as does ¹³CO as well, and in general present a non-Gaussian spectrum profile. In many cases, ¹²CO is self-absorbed, while ¹³CO and C¹⁸O peak at the ¹²CO local minimum (as shown in Fig. 7).

Local decrease in intensity of the ¹²CO/¹³CO ratio (Fig. 20) corresponds to higher X_co factor (Fig. 18), where the ¹²CO/¹³CO ratio is lowest, the CO is most optically thick and the X_co is highest.

This is not unexpected since, by definition, our X_co map derives from the ratio between N(H₂) traced by ¹³CO (proportional to ¹³CO integrated intensity through the optical depth integration; see equations 4 and 5) and ¹²CO integrated intensity. These optically thickest regions show significant decrease in ¹²CO emission with respect to ¹³CO.

10.3.1 X_co and T_ex

To further investigate the physical conditions underlying the observed X_co distribution (Fig. 19), we looked at the variation of T_ex around the molecular ring. In Fig. 28 are shown the scatter plots of T_ex versus X_co for each sub-region of the cloud. The T_ex value was derived from the ¹²CO peak temperature for each sight line, while the X_co was computed by using the N(H₂) traced by ¹³CO, assuming in the ¹³CO column density a T_ex derived from the dust temperature computed by the SED fit (see Section 6.5). In Table 8 are summarized the X_co and T_ex mean values for each region.

Figure 28.

X_co factor plots for each ring sub-region as per Fig. 1. The range of x and y axes are the same for each plot. The plot [SE-R] comprises only pixels coming from [SE-R], excluding those with ℓ > 332.65°, which are reported in the sub-plot indicated with [E,SE-R] (Eastern-South East, Region).

Open in new tabDownload slide

Examining Fig. 28 we note a general decrease in the dispersion of X_co values, moving from [N-R] towards [SE-R], while T_ex increases. This behaviour in T_ex is similar to the temperatures and L_bol/M_clump ratio trend observed in the dust clumps (Fig. 23) and is consistent with a stellar evolution gradient scenario: the Southern cloud hosts active and more evolved star formation sites, which heat the surrounding gas so increasing its temperature. On the contrary, the Northern cloud hosts less and, on average, earlier stage star formation sites, the gas is colder compared to the South and the peak emissions are the lowest, as well as the velocity dispersion and the column densities.

Supporting a scenario in which the Southern cloud is more active than the Northern, as illustrated in Fig. 1, the literature reports many indicators of star formation activity, as H ii regions (see Section 5.2) and masers, localized in the Southern ring: 6 GHz methanol maser (Caswell et al. 2010), water maser (Walsh et al. 2014), class I methanol maser at 36 and 49 GHz (Voronkov et al. 2014), and 95 GHz maser (Yang et al. 2017).

Recently Kong et al. (2015) reported for the California Molecular Cloud high X_co values dispersion for cold quiescent regions and an inverse correlation between X_co and T_ex. They argued that this correlation could explain the range of X_co values, found for different GMCs, as a consequence of the different amount of warm gas heated by OB stars in the cloud. The G332 molecular cloud presents similar behaviour, where the coldest parts of the cloud show the most scattering and highest X_co values and the lowest T_ex, while the most active regions show the opposite behaviour.

10.4 [C i]

The UV radiation coming from the massive young stars photodissociates the surrounding gas, creating a PhotoDissociation Region (PDR), where the carbon is mainly in atomic form, in neutral or singly ionized state (Tielens & Hollenbach 1985; van Dishoeck & Black 1988); similarly, the external layer of molecular clouds are exposed to the diffuse far-UV radiation, photodissociating the CO before it reaches a column density high enough to enable self-shielding. To inspect the atomic carbon distribution around the cloud, in Fig. 21 is illustrated the ratio between the integrated intensities of [C i] and ¹³CO. As expected, the outskirts of the clouds possess the highest ratio values, though with the lowest SNR. Interestingly, [NW-R] and [W-R] present an exception: low ratio values at the edge facing the North–West direction, while the edge towards South–East presents higher values. Looking at the sightlines passing through the innermost part of the cloud, the regions with the lowest [CI]/¹³CO ratio values are [N-R], [NW-R], [E,SE-R], and part of [W-R]. Profiles showing the ratio as a function of velocity are in Fig. 29. They were derived from the [C i] and ¹³CO (regridded to the [C i] resolution) datacubes, both masked using the same [C i] 3D mask, the resulting profiles were then plotted applying a Hanning smooth of order 3. The resulting picture is of a northern ring, [N-R] and [NW-R], with mean ratio values of 0.3 centred at V_LSR = |$-50\rm \, km\, s^{-1}$| (we consider the second peak at V_LSR = |$-42\rm \, km\, s^{-1}$| belonging to another cloud; see Fig. 3), while the southern ring has a mean ratio value of 0.5, peaking at 0.6 in the external layer of the cloud. In [SW-R] (aperture D) the mean ratio spectral profile shows a constant value going deeper in the cloud, a possible indication of photodissociation in the cloud interior caused by star formation activity. Apertures A and C, belonging to the southern cloud, show a small dip in the innermost part of the ring, where the ¹³CO emission increases. Regarding aperture E, we note that the emission velocity centroids of the [C i] and the ¹³CO emissions are close to V_LSR = |$-50\rm \, km\, s^{-1}$| (Fig. 8), while we consider the second peak at V_LSR = |$-42\rm \, km\, s^{-1}$| belonging to another cloud.

Figure 29.

Plots of the mean spectral profiles of [CI]/¹³CO ratio right for the different apertures indicated with different letters in the left-hand image. The image map represents the ¹³CO integrated intensity map, with the green contours its extent.

Open in new tabDownload slide

We conclude that the fraction of atomic carbon, with respect to its molecular form, is lower in the Northern cloud than in the Southern cloud. The latter possesses the highest photodissociation levels due to increased radiation fields. We speculate that [NW-R] is the most quiescent part of the cloud, and the difference between North and South is driven by a star formation activity ongoing at a later stage in [SW-R] and [SE-R], compared to [N-R], [NW-R], and [E,SE-R]. This further strengthens the hypothesis of a difference in the star formation activity between northern and southern ring.

10.5 CO and 8 |$\mu$|m emission

IRDCs are regions in which the gas is cold and dense, absorbing background IR emission. They provide suitable targets to look for signposts of early star formation activity (Teyssier, Hennebelle & Pérault 2002; Rathborne, Jackson & Simon 2006; Simon et al. 2006). The dust temperature map reported in Fig. 11 suggests that the North–Western ring is the coldest region, thus it is likely to host IRDCs. In Fig. 30 we plot the angular positions and equivalent radius size of the IRDCs (red circles) identified by Peretto et al. (2016), overlaid on the X_co map. Despite the coarser angular resolution of the X_co, compared to the 16 arcsec Hi-GAL maps, it is possible to see a spatial correlation between their positions and the angular extent of the molecular cloud.

Figure 30.

Image of the X_co factor (see Section 7.1) calculated assuming as T_ex the ¹²CO peak temperature map. The red circles represent the angular position and the equivalent radius of the IRDCs (as reported by Peretto et al. 2016) .

Open in new tabDownload slide

In Fig. 31 is shown the Spitzer|$8\,\hbox{$\mu $m}$| image, in green contour is the angular extension of the ¹³CO integrated intensity map, while the red contours represents the dust temperature as per Fig. 11. We note that the distribution of cores with lowest dust temperatures precisely follow the extinction features seen in |$8\,\hbox{$\mu $m}$|⁠. Moreover, between 332.3° < ℓ < 332.1° and −0.1° < b < 0.3°, an arc-like extinction feature is clearly seen, following the CO emission in the sky portion defined by northern [SW-R], [W-R], and [NW-R] and southern [N-R]. The same correspondence between lowest dust temperatures values and |$8\,\hbox{$\mu $m}$| extinction is seen in [E,SE-R]. On the other hand, the highest dust temperatures match the highest |$8\,\hbox{$\mu $m}$| emission. We found correspondence between the distribution of N(H₂) traced by C¹⁸O and the extinction features (see Fig. 32), also in [E,SE-R], and in small localized regions of the southern cloud, i.e. (ℓ, b) = (332.4°, −0.1°), suggesting a strong correlation between the IRDC position, the gas in the molecular cloud, and the coldest dust.

Figure 31.

Spitzer |$8\,\hbox{$\mu $m}$| map, overlaid contours in green are the extent of the ¹³CO while in red is the dust temperature as per Fig. 11. The temperature contours are in the range [15,|$27\rm \, K$|] in steps of |$1\rm \, K$|⁠. The red square delimits where the CO data are used.

Open in new tabDownload slide

Figure 32.

Spitzer |$8\,\hbox{$\mu $m}$| overlaid with the extent of the ring defined by ¹³CO (green contours) and in the cyan contours N(H₂) as traced by C¹⁸O. Contour levels are from 5 × 10²⁰ to 6 × 10²² in steps of 7 × 10²¹ cm⁻².

Open in new tabDownload slide

Carbon monoxide depletion has been observed (Alves, Lada & Lada 1999; Kramer et al. 1999; Kong et al. 2015) in dense and cold regions. Thus, the cold and dense environment of IRDCs could favour CO depletion caused by CO removal from gas phase due to freeze out on to dust grains. It is important to note that could affect the CO to dust comparison method presented in Section 10.1, which was used to decompose the dust emission along the line of sight across the Galaxy spiral arms, since we assumed that the dust and CO emission are proportional. A consequence of the CO depletion would be an underestimate of the dust content.

Comparing N(H₂) traced by dust and N(H₂) traced by C¹⁸O to probe the densest regions of molecular cloud, we found the distribution of N(H₂) by C¹⁸O follows the IRDC position in the Northern part of [SW-R], in [W-R] and [NW-R], as well as in localized regions of the southern ring. In particular, in cold dust regions (Fig. 33), we found an offset between the C¹⁸O N(H₂) peaks and dust N(H₂) peaks, which could be an indication of CO depletion in the highest column density dust regions.

Figure 33.

Molecular hydrogen column density from C¹⁸O and in red contours N(H₂) traced by dust as per Fig. 13.

Open in new tabDownload slide

The emission at 8 μm comes mainly from PAHs (Tielens & Hollenbach 1985) excited by UV radiation from high-mass stars in PDRs. The highest |$8\,\hbox{$\mu $m}$| emission (in Fig. 31) are localized in [SW-R], [SE-R], and [C-R] regions, where the CO spectral lines show the highest emission peaks and the highest velocity dispersion. If the gas in the molecular cloud and the IRDC are part of the same structure, we expect to observe a decreasing gradient in the T_ex from [SE-R] to [N-R], similar to that observed from the Southern to Northern cloud (see Table 8).

Taking into account all these factors, we hypothesize that this IRDC arc-like structure, and its extension in the souther ring, is associated with the coldest and densest parts of the cloud, a unique structure along which the star formation activity is at a later stage in the southern ring than the northern ring. Considering [E,SE-R] and [N-R] as the extremities of the IRDC, its length is |${\sim }65\rm \, pc$|⁠.

11 VIRTUAL REALITY AND AUGMENTED REALITY: AN APPLICATION TO RADIO ASTRONOMY

In this section we present a new perspective for looking at astronomical data, developed during the analysis of the G332 molecular ring and applicable to other 3D surveys. The aim is to find a visualization method to better understand the distribution of CO data and to perform an efficient multiwavelength synergy with other surveys. This is not only a data-driven problem but also a data visualization issue, which we tried to solve by experimenting with the use of VR and AR environments and applications.

Currently, the astronomy community is experiencing a growing interest in AR/VR software and hardware to improve data visualization and analysis (Ferrand, English & Irani 2016; Fluke & Barnes 2018). However, most of the software used by astronomers does not offer advanced visualization tools, mainly because it was developed before the current available hardware, but also because its 3D plot capabilities have limited user interaction. On the other hand, the possibility given by VR/AR tools to literally fly through data sets dramatically improves the analysis process. Since our senses evolved in a 3D world, a more intuitive data interpretation is likely to occur in a 3D environment rather than in a 2D representation; this is why immersive techniques have the potential to boost data examination. We made use of VR/AR visualization in our data CO investigation¹⁰ to discriminate between emissions related to the main molecular ring structure (as in Section 5.3) and not. Standard 2D datacube plots (as moments maps or PV plots) could lead to data misinterpretations, because of projection effects, while a 3D immersive visualization could help showing data in a greater dimensional space. To solve the above ambiguity, we inspected ¹²CO, ¹³CO, and C¹⁸O emissions at different thresholds at the same time (2σ, 3σ, 3.5σ, 4σ, and 4.5σ). Such an amount of details is hard to clearly represent on a standard 3D plot, restricting analysis to a limited amount of information.

However, it is important to say that care is needed in PPV 3D datacube inspection since the velocity (or spectral) dimension could lead to misinterpretation in the analysis of the emission morphology (as discussed in Clarke et al. 2018).

A digression on the limits of VR/AR application in astronomy is beyond the purpose of this paper; they were analysed by Vogt & Shingles (2013). Here we note that the continuous increase of computing power, and the consequent increase in smartphone performance, makes it possible to use complex software capable of achieving advanced results without increasing its complexity.

In Appendix  A we describe in detail three examples of our VR and AR visualization techniques. All of them are based on the same initial step: convert the fits format into another 3D format, which can be easily read by VR (or AR) frameworks.

12 SUMMARY

We used data from the Mopra Southern Galactic Plane CO Survey of the ¹²CO, ¹³CO, and C¹⁸O J = 1 → 0 transition lines (Braiding et al. 2018) to perform a morphological and a physical characterization of a GMC (G332 molecular ring) sited in the G332 region of the Galactic plane. The CO emission peaks at V_LSR = |$-50\rm \, km\, s^{-1}$|⁠, which is equal to a distance of ∼3.3 kpc from the Sun (assuming correct a near distance solution for KDA) and to a Galactocentric distance of ∼5.5 kpc. The cloud shape is similar to a ring of |${\sim }42\rm \, pc$| in diameter. The total mass of the cloud is estimated in ∼2 × 10⁵ M_⊙, assuming [¹²CO]/[¹³CO] = 53, [H₂]/[¹²CO] = 1.1 × 10⁴, and uniform T_ex = |$30\rm \, K$| for all the subregions of the molecular structure. We found a strong spatial correlation between the C¹⁸O and an IRDC structure that was seen as an extinction feature at |$8\,\hbox{$\mu $m}$|⁠.

We derived the X_co variation along the ring for different T_ex values, ranging from X_co = 2.1 × 10²⁰ to |$2.3\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠, respectively, for a T_ex = |$10\rm \, K$| and for a T_ex derived from peak temperature map of ¹²CO.

We defined an X|$\mathrm{_{CI}^{809}}$| factor, using the [C i] ³P₂ → ³P₁ transition at 809 GHz and N(H₂) traced by ¹³CO. The derived values are (1.77 ± 0.03) × 10²¹, (1.83 ± 0.02) × 10²¹, and |$(1.97\pm 0.02)\times 10^{21}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$|⁠, respectively, for N(H₂) traced by the three assumptions of ¹²CO, ¹³CO without adding dark gas, and ¹³CO adding the dark gas fraction.

The median value of the ¹²CO/¹³CO ratio is ∼3.3, with a minimum value close to 1, mostly in the [NW-R], region that we concluded to be the most quiescent part of the cloud. The median [C i]/¹³CO ratio for the whole cloud is equal to 0.5, ranging in the interval [0.4–1], with the lowest values in [NW-R] and [W-R].

The [C i] emission is lowest in the [NW-R] and [W-R], while the maximum values are found to be in the Southern cloud regions.

The mean dark molecular gas mass fraction of N(H₂) not traced by CO results to be equal to |${\sim }16{{\ \rm per\ cent}}$|⁠, in average, considering a T_ex = |$20\rm \, K$| for ¹³CO, and a T_ex = |$30\rm \, K$| for [C i].

We compiled a catalogue of clumps decomposing the ring GMC C¹⁸O emission and reporting the main physical characteristics derived from CO line, mostly all possess a virial parameter <1.

HiGAL maps were used to derive the dust SED and the N(H₂) traced by dust. The derived dust temperature is in the range between 16 and |$27\rm \, K$|⁠. The linear fit between N(H₂) traced by dust and the ¹³CO integrated intensity along all the line of sight gives a coefficient m = |$5.9\times 10^{20}\,\mathrm{cm}^{-2}\, \mathrm{K}^{-1}\, \mathrm{km}^{-1}\, \mathrm{s}$| and b = 9.8 × 10²¹ cm⁻².

We presented a method to disentangle dust emission across the spiral arms from our adopted Galactic model (McClure-Griffiths & Dickey 2007; Vallée 2014), through the use of the ¹³CO emission along the velocity range |$[-115,-30]\rm \, km\, s^{-1}$| (Table 6).

We found a decreasing gradient in Hi-Gal clumps temperatures and L_bol/M_clump ratio going from [SE-R] towards [N-R], which corresponds to a T_ex decrease and to an increase of X_co median value and values dispersion.

The spatial distribution of cold dust and C¹⁸O follows an arc-like |$8\,\hbox{$\mu $m}$| extinction feature in northern [SW-R], [W-R], [NW-R], and [E,SE-R]. We found correspondences between the lowest |$8\,\hbox{$\mu $m}$| emission, cold dust, and C¹⁸O distribution also in the southern ring. We consider this extinction feature as an IRDC that is inside the molecular ring. Along the IRDC we found that, in general, the peak position of N(H₂) traced by C¹⁸O present an offset position with respect to N(H₂) traced by dust peaks.

We concluded that the observed gas properties are influenced by the stellar formation activity occurring along the identified IRDC, which is an unique structure around which the molecular cloud is wrapped. The path from [E,SE-R] to Southern [N-R] is |${\sim }65\rm \, pc$|⁠. We found evidence of a different star formation activity between the southern and northern ring.

We presented a new visualization approach for the 3D radio astronomy datacubes by using VR and AR environment, based on unity, javascript, and online platforms in combination with non-specifically designed equipment.

ACKNOWLEDGEMENTS

The Mopra radio telescope is part of the Australia Telescope National Facility. Operations support was provided by the University of New South Wales, the University of Adelaide, and Western Sydney University. Many staff of the ATNF have contributed to the success of the remote operations at Mopra. We particularly wish to acknowledge the contributions of Balthasar Indermuehle, David Brodrick, Philip Edwards, Brett Hiscock, and Peter Mirtschin.

Funding for the HEAT telescope is provided by the National Science Foundation under grant number PLR-0944335. PLATO-R was funded by Astronomy Australia Limited, as well as the University of New South Wales, as an initiative of the Australian Government being conducted as part of the Super Science Initiative and financed from the Education Investment Fund. Logistical support for HEAT and PLATO-R is provided by the United States Antarctic Program. We also acknowledge ARC support through Discovery Project DP120101585.

This research made use of astropy, a community-developed core python package for Astronomy (Astropy Collaboration 2013) and the idl astronomy library (Landsman 1995), the simbad data base (Wenger et al. 2000), and aladin sky atlas desktop program (Bonnarel et al. 1999, 2000; Chilingarian & Zolotukhin 2008), maintained by the CDS in Strasbourg, France, and the glue¹⁸ (Beaumont, Goodman & Greenfield 2015) visualization package. We thank the anonymous referee for the useful comments that helped us to enrich the work presented in this paper. We would also thank, for the stimulating discussions, Shaila Akhter, Nigel Maxted, Davide Elia, Matthew Freeman, Anant Tanna, and John Lopez.

Footnotes

REFERENCES