Opening the Treasure Chest in Carina

TL;DR

This study maps and analyzes the physical conditions of the G287.84-0.82 globule in Carina, revealing how stellar radiation and winds shape its structure and gas properties, with detailed multi-line observations and modeling.

Contribution

It provides a comprehensive multi-wavelength analysis of the globule, combining new observations with PDR and non-LTE models to characterize its morphology, kinematics, and physical conditions.

Findings

Identification of two PDR gas components with different densities.

Determination of optical depths and excitation temperatures of [CII].

Estimation of the globule's total mass (~1000 solar masses).

Abstract

We have mapped the G287.84-0.82 cometary globule (with the Treasure Chest cluster embedded in it) in the South Pillars region of Carina (i) in [CII], 63micron [OI], and CO(11-10) using upGREAT on SOFIA and (ii) in J=2-1 transitions of CO, 13CO, C18O and J=3-2 transitions of H2CO using the APEX telescope in Chile. We probe the morphology, kinematics, and physical conditions of the molecular gas and the photon dominated regions (PDRs) in G287.84-0.82. The [CII] and [OI] emission suggest that the overall structure of the pillar (with red-shifted photo evaporating tails) is consistent with the effect of FUV radiation and winds from eta-Car and O stars in Trumpler 16. The gas in the head of the pillar is strongly influenced by the embedded cluster, whose brightest member is an O9.5V star, CPD-59 2661. The emission of the [CII] and [OI] lines peak at a position close to the embedded star, while all other tracers peak at another position lying to the north-east consistent with gas being compressed by the expanding PDR created by the embedded cluster. The molecular gas inside the globule is probed with the J=2-1 transitions of CO and isotopologues as well as H2CO, and analyzed using a non-LTE model (escape-probability approach), while we use PDR models to derive the physical conditions of the PDR. We identify at least two PDR gas components; the diffuse part (~10^4 cm^-3) is traced by [CII], while the dense (n~ 2-8x10^5 cm^-3) part is traced by [CII], [OI], CO(11-10). Using the F=2-1 transition of [13CII] detected at 50 positions in the region, we derive optical depths (0.9-5), excitation temperatures of [CII] (80-255 K), and N(C+) of 0.3-1x10^19 cm^-2. The total mass of the globule is ~1000 Msun, about half of which is traced by [CII]. The dense PDR gas has a thermal pressure of 10^7-10^8 K cm^-3, which is similar to the values observed in other regions.

Tables

Table 1. Table 1: Observation Details: Frequencies, Receivers, Beam sizes, and beam efficiencies

Telescope	Receiver	Transition	Frequency	${\theta }_{HPBW}$	${\eta }_{mb}$
			[GHz]	[ ″ ]
SOFIA	GREAT L1	CO(11–10)	1267014.49	21.0	0.67
	LFA	[C ii] ${}_{}{}^{2}P_{3/2}^{}{\to }^2{P}_{1/2}$	1900536.90	14.0	0.67a
	LFA	[13C ii] ($F$=2–1)	1900466.10	14.0	0.67a
	HFA	[O i] ${}_{}{}^{3}P_0^{}{\to }^3{P}_1$	4744777.49	6.3	0.65a
APEX	PI230	CO(2–1)	230538.000	27.3	0.68
		13CO(2–1)	220398.684	28.5	0.68
		C18O(2–1)	219560.354	28.6	0.68
		H2CO(303–202)	218222.192	28.8	0.68
		H2CO(322–221)	218475.630	28.8	0.68
		H2CO(321–220)	218760.066	28.8	0.68
		SO (65–54)	219949.44	28.6	0.68

Table 2. Table 2: Results of single (or double for position #8) component Gaussian fitting of observed spectra

($\Delta \alpha$, $\Delta \delta$)	Transition	$I$0000	${\upsilon }_{\mathrm{LSR}}$000	$\Delta \upsilon$000
		K km s-1	km s-100	km s-1
(0″,38″)	[C ii]	288.4$\pm$6.2	-14.7$\pm$0.1	3.8$\pm$0.1
	CO(2–1)	137.4$\pm$0.3	-14.8$\pm$0.1	3.4$\pm$0.1
	13CO(2–1)	49.8$\pm$0.2	-14.7$\pm$0.1	2.7$\pm$0.1
	C18O(2–1)	7.7$\pm$0.1	-14.8$\pm$0.1	2.3$\pm$0.1
	CO(11–10)	86.9$\pm$3.3	-14.8$\pm$0.1	2.9$\pm$0.1
	[O i] 63$\mu$m	125.7$\pm$2.7	-15.0$\pm$0.1	4.9$\pm$0.1
(-15″,30″)	[C ii]	511.0$\pm$6.0	-15.0$\pm$0.1	4.2$\pm$0.1
	CO(2–1)	127.7$\pm$0.3	-14.8$\pm$0.1	3.2$\pm$0.1
	13CO(2–1)	33.9$\pm$0.1	-14.7$\pm$0.1	2.4$\pm$0.1
	C18O(2–1)	4.2$\pm$0.1	-14.7$\pm$0.1	2.1$\pm$0.1
	CO(11–10)	64.0$\pm$3.9	-14.5$\pm$0.1	2.5$\pm$0.2
	[O i] 63$\mu$m	210.0$\pm$3.8	-15.0$\pm$0.1	4.4$\pm$0.1
(10″,9″)	[C ii]	435.5$\pm$6.0	-14.1$\pm$0.1	4.2$\pm$0.1
	CO(2–1)	107.6$\pm$0.3	-14.2$\pm$0.1	3.5$\pm$0.1
	13CO(2–1)	32.8$\pm$0.2	-13.9$\pm$0.1	2.7$\pm$0.1
	C18O(2–1)	4.3$\pm$0.1	-14.0$\pm$0.1	2.9$\pm$0.1
	CO(11–10)	31.4$\pm$3.5	-13.7$\pm$0.2	2.6$\pm$0.3
	[O i] 63$\mu$m	103.3$\pm$3.9	-13.5$\pm$0.1	4.3$\pm$0.2
(-12″,5″)	[C ii]	303.1$\pm$4.8	-14.7$\pm$0.1	3.8$\pm$0.1
	CO(2–1)	88.8$\pm$0.3	-14.6$\pm$0.1	2.9$\pm$0.1
	13CO(2–1)	20.2$\pm$0.1	-14.5$\pm$0.1	2.1$\pm$0.1
	C18O(2–1)	2.7$\pm$0.1	-14.4$\pm$0.1	2.0$\pm$0.1
	CO(11–10)	28.4$\pm$2.8	-14.5$\pm$0.1	2.4$\pm$0.3
	[O i] 63$\mu$m	90.4$\pm$3.9	-14.7$\pm$0.1	3.8$\pm$0.2
(-4″,-32″)	[C ii]	293.4$\pm$8.9	-14.6$\pm$0.1	3.7$\pm$0.1
	CO(2–1)	96.0$\pm$0.5	-14.6$\pm$0.1	2.7$\pm$0.1
	13CO(2–1)	22.4$\pm$0.2	-14.6$\pm$0.1	2.2$\pm$0.1
	C18O(2–1)	1.9$\pm$0.1	-14.8$\pm$0.1	1.8$\pm$0.1
	CO(11–10)	7.0$\pm$2.1	-14.5$\pm$0.1	0.9$\pm$0.3
	[O i] 63$\mu$m	67.9$\pm$4.0	-15.0$\pm$0.1	3.9$\pm$0.3
(-81″,-44″)	[C ii]	95.9$\pm$3.9	-13.9$\pm$0.1	2.3$\pm$0.1
	CO(2–1)	15.1$\pm$0.3	-13.8$\pm$0.1	1.6$\pm$0.1
	13CO(2–1)	3.8$\pm$0.1	-13.6$\pm$0.1	1.3$\pm$0.1
	C18O(2–1)	0.5$\pm$0.1	-13.5$\pm$0.2	2.0$\pm$0.5
	CO(11–10)	7.0$\pm$1.4	-13.7$\pm$0.1	1.0$\pm$0.2
	[O i] 63$\mu$m	19.5$\pm$3.5	-13.7$\pm$0.2	1.9$\pm$0.5
(-112″,-25″)	[C ii]	92.3$\pm$3.9	-14.2$\pm$0.1	2.6$\pm$0.1
	CO(2–1)	19.4$\pm$0.5	-14.3$\pm$0.1	2.3$\pm$0.1
	13CO(2–1)	4.6$\pm$0.2	-13.9$\pm$0.1	1.8$\pm$0.1
	CO(11–10)	17.1$\pm$2.8	-14.4$\pm$0.1	1.7$\pm$0.3
	[O i] 63$\mu$m	31.2$\pm$5.5	-12.3$\pm$0.3	3.9$\pm$0.7
(74″,-70″)	[C ii]	70.1$\pm$2.3	-14.3$\pm$0.1	2.8$\pm$0.1
		85.2$\pm$2.2	-11.9$\pm$0.1	2.3$\pm$0.1
	CO(2–1)	31.9$\pm$1.0	-15.0$\pm$0.1	2.7$\pm$0.1
		33.3$\pm$0.9	-12.0$\pm$0.1	2.5$\pm$0.1
	13CO(2–1)	3.7$\pm$0.3	-14.9$\pm$0.1	1.8$\pm$0.1
		6.8$\pm$0.3	-12.0$\pm$0.1	2.3$\pm$0.1
	[O i] 63$\mu$m	7.6$\pm$2.7	-14.7$\pm$0.5	2.4$\pm$1.0
		21.0$\pm$2.6	-11.8$\pm$0.1	1.9$\pm$0.2

Equations

$$\frac{{\mathrm{I}}_{12}}{{\mathrm{I}}_{13}}=\frac{{}^{12}\mathrm{C}}{{}^{13}\mathrm{C}}\left[\frac{1-\exp (-{\tau }_{12})}{{\tau }_{12}}\right],$$

$${\mathrm{I}}_{12}^{\mathrm{corr}}=\frac{{\mathrm{I}}_{12}{\tau }_{12}}{1-\exp (-{\tau }_{12})}$$

$$\frac{{\mathrm{I}}_{12}^{\mathrm{corr}}}{\int {\tau }_{12}\mathrm{d}\upsilon }=\frac{92\exp (-91/{\mathrm{T}}_{\mathrm{ex}}^{\mathrm{CII}})}{1-\exp (-91/{\mathrm{T}}_{\mathrm{ex}}^{\mathrm{CII}})}$$

$$\int {\tau }_{12}\mathrm{d}\upsilon =7.15\times 10^{-18}{\mathrm{N}}_{{\mathrm{C}}^{+}}\frac{32.9\text{leavevmode}\mathrm{K}}{{\mathrm{T}}_{\mathrm{ex}}}$$

Full text

1 Introduction

Some of the most spectacular structures in the molecular interstellar medium (ISM) are observed in the vicinity of hot O/B stars or associations in the forms of cometary globules and pillars. The cometary globules arise when the expanding H ii region due to the O/B star overruns and compresses an isolated nearby cloud causing it to collapse gravitationally. As the cloud contracts the internal thermal pressure increases causing the cloud to re-expand, so that a cometary tail is formed pointing away from the exciting star. A high density core develops at the head of the globule, which could collapse to form one or more stars. The pillars arise when an ionization front expands into a dense region embedded in the surrounding molecular cloud, so that a cometary tail or pillar forms due to the shadowing behind the dense clump. The pillar so formed is supported against radial collapse by magnetic field but undergoes longitudinal erosion by stellar winds and radiation. There exist multiple analytical (Bertoldi, 1989; Bertoldi & McKee, 1990) and numerical models (Lefloch & Lazareff, 1994; Bisbas et al., 2011), which explain and reproduce many of the observed features of these structures sculpted by the radiation and wind from massive O/B stars starting from either an isolated or an embedded pre-existing dense clump or core. Additionally, there are observations which find the presence of sequential star formation in such structures consistent with triggered formation of stars due to the photoionization-induced shocks (Sugitani et al., 2002; Ikeda et al., 2008). However, numerical simulations are yet to provide any strong evidence supporting the enhanced formation of stars due to triggering by the ionization fronts. Physically the scenario is complex due to the simultaneous influence of radiation, magnetic field and turbulence on the evolution of such clouds. Since the pillars and globules are regions in which the gas is strongly influenced by radiation, observation of the emission from the photon dominated regions (PDRs) as traced by the [C ii] at 158 $\mu$m, [O i] at 63 $\mu$m and mid- and high-$J$ CO lines are ideal to constrain the physical conditions of these regions.

The Carina Nebula Cluster (NGC 3372) is the nearest ($d$=2.3 kpc) massive star-forming region in the southern hemisphere with more than 65 O stars (Smith et al., 2010). Early surveys in molecular lines and far-infrared continuum of the central parts of the Carina Nebula suggested that the radiation and stellar winds from hot massive stars in the region are clearing off the remains of their natal cloud (Harvey et al., 1979; de Graauw et al., 1981; Ghosh et al., 1988). The south eastern part of the nebula is particularly rich in large elongated bright pillars, all of which seem to point towards $\eta$ Carina and Trumpler 16, a massive open cluster with four O3-type stars, as well as numerous late O and B stars. Studies of this region, called the South Pillars, in the thermal infrared and radio suggest that it is a site of ongoing and possibly triggered star formation (Smith et al., 2000; Rathborne et al., 2004; Smith et al., 2005). Since the pillars in this region have the bright parts of their heads pointing towards $\eta$ Car and their extended tails pointing away from it, it seems clear that they are sculpted by the radiation and winds from the massive stars associated with Trumpler 16 and $\eta$ Car.

One of the cometary globules in the South Pillars, G287.84-0.82, is known to have a dense compact cluster embedded in the head of the globule. This cluster, called the Treasure Chest (Smith et al., 2005), contains more than 69 young stars. Oliveira, Bica & Bonatto (2018) estimate a cluster age of 1.3 Myr. The most massive cluster member is CPD -59∘2661 (at $\alpha_{\rm J2000}$= 10h 45m 53${}^{s}\mskip-7.6mu.\,$713, $\delta_{\rm J2000}$ = -59∘57′3${}^{\prime\prime}\mskip-7.6mu.\,$8), an O9.5 star (Walsh, 1984; Hägele et al., 2004), which illuminates a compact H ii region seen in H$\alpha$ and in PDR emission (Thackeray, 1950; Smith et al., 2005), see Fig. 1 & 2. To the north and east of the star the surrounding dense molecular cloud prevents the H ii region from expanding. Here the distance from the expanding H ii region to the star is only $\sim$ 12″, while it appears to be expanding freely to the west, where the gas densities are low (Fig 2). The bright star $\sim 30$″ northeast of CPD -59∘2661 (saturated in IRAC images) is probably a member of the embedded cluster. This star (Hen 3-485 = Wra 15-642) is Herbig Be star (Gagné et al., 2011). It is reddened by $\sim$ 2 mag and has strong infrared excess. It is associated with an H2O maser with a single maser feature at the same velocity of the globule (Breen et al., 2018). The vibrationally excited H2 observations of Treasure Chest by Hartigan et al. (2015) show the external PDR illuminated by $\eta$ Car. They also show an internal PDR excited by CPD -59∘2661, which expands into the head of the globule. Thus, while the cometary globule around the Treasure Chest cluster appears to have been shaped by the radiation from the massive stars, the embedded young stellar cluster illuminates its own PDR in the head of the globule and will eventually dissociate the surrounding molecular cloud.

Here we present mapping observations of the [C ii] at 158 $\mu$m, [O i] at 63 $\mu$m, and CO emission of G287.84-0.82 in Carina to derive a better understanding of the physical conditions of the globule and the Treasure Chest cluster.

2 Observations

2.1 SOFIA

We have retrieved [C ii] and CO(11–10) observations of G287.84-0.82 (the Treasure Chest) from the data archive of the Stratospheric Observatory for Infrared Astronomy (SOFIA; Young et al., 2012). The observations (PI: X. Koenig) were part of a larger program (01_0064) on observing irradiated pillars in three massive star forming regions. The observations were done with the German REceiver for Astronomy at Terahertz frequencies (GREAT111The development of GREAT was financed by the participating institutes, by the Federal Ministry of Economics and Technology via the German Space Agency (DLR) under Grants 50 OK 1102, 50 OK 1103 and 50 OK 1104 and within the Collaborative Research Centre 956, sub-projects D2 and D3, funded by the Deutsche Forschungsgemeinschaft (DFG).; Heyminck et al., 2012) in the L1/L2 configuration on July 22nd 2013 during the New Zealand deployment. These observations were made on a 70 minute leg at an altitude of 11.9 km. The L1 mixer was tuned to CO(11–10), while the L2 mixer was tuned to the [C ii]  ${}^{2}{\rm P}_{3/2}\to\ ^{2}{\rm P}_{1/2}$ transition, see Table 1 for observation details. The backends were fast Fourier transform spectrometers (AFFTS) (Klein et al., 2012) with 8192 spectral channels covering a bandwidth of 1.5 GHz, providing a frequency resolution of 183.1 kHz. The Treasure Chest was mapped in total-power-on-the-fly mode with a 1 second integration time and sampled every 8″ for a map size of 248″ $\times$ 168″. The map was centered on ($\alpha$, $\delta$) = (10h 45m 55$\aas@@fstack{s}$05, -59° 57′41$\aas@@fstack{\prime\prime}$2) (J2000) with an off-position 5′ to the south. All offsets mentioned in the text and figure captions hereafter refer to this centre position. Here we only use the CO(11–10) data, because the [C ii] observations made with upGREAT, see below, are of superior quality.

The source G287.84-0.82 was also observed with upGREAT in consortium time from Christchurch on 2018 June 14 on a 65 minute leg at an altitude from 11.4–12.4 km. The upGREAT was in the Low Frequency Array/High Frequency Array (LFA/HFA) configuration with both arrays operating in parallel. The 14 beam LFA array (Risacher et al., 2016) was tuned to [C ii], while the 7 beam HFA array was tuned to the 3P1–3P2 63 $\mu$m  [O i] line. The spectrometer backends are the last generation of fast Fourier transform spectrometers (FFTS; Klein et al. (2012)). They cover an instantaneous intermediate frequency (IF) bandwidth of 4 GHz per pixel, with a spectral resolution of 142 kHz. To ensure good baseline stability for the [O i] line we mapped the Treasure Chest in Single Beam Switch (SBS) on-the-fly mode by chopping to the east of the cometary globule. The region we chopped to appears free of [C ii] emission, even though there is widespread faint foreground emission at velocities of $\sim$ -30 to -24 km s*-1* in the area around the globule. We split the region into two sub-maps. The northern part of the globule was done as a 123″ $\times$ 69″ map with a chop amplitude of 120″ at a position angle (p.a.) of 60°. The map was centered with an offset of 0″,+27″ relative to our track position and sampled every 3″ with an integration time of 0.4 seconds. The second map had a size of 231″ $\times$ 93″and was centered at an offset of -9″,-54″. For this map we used a chop amplitude of 135″ at a p.a. of 84°. This setup provided us a fully sampled map in [O i] over most of the globule and an oversampled map of [C ii] with excellent quality.

The atmospheric conditions were quite dry. The zenith optical depth for [C ii] varied from $\sim$0.19 at the beginning of the leg to $\sim$0.15 at the end of the leg. The system temperatures for most pixels were about 2400 K, some around 2900 K, and three outliers had higher system temperatures, 3300 - 4000 K. One noisy pixel was removed in the post processing. Beam efficiencies were applied individually to each pixel, see Table 1. For [O i] the zenith optical depth varied between 0.24 to 0.43 and system temperatures ranged from 2900 to 4650 K excluding one noisy pixel, which was not used.

All the GREAT data were reduced and calibrated by the GREAT team, including correction for atmospheric extinction (Guan et al., 2012) and calibrated in antenna temperature scale,T${}_{\rm A}^{*}$, corrected for a forward scattering efficiency of 0.97, and in main beam antenna temperature, Tmb.

The final data cubes were created using CLASS222CLASS is part of the Grenoble Image and Line Data Analysis Software (GILDAS), which is provided and actively developed by IRAM, and is available at http://www.iram.fr/IRAMFR/GILDAS. Maps were created with a velocity resolution of 0.5 km s*-1*. The rms noise per map resolution element with 0.5 km s*-1* velocity resolution is $\sim$ 1.3 K for the CO(11–10) map, 0.70 K for [C ii], and 2.4 K for [O i].

2.2 APEX

G287.84-0.82 was observed on March 28, 2017 using the PI230 receiver on the 12 m Atacama Pathfinder EXperiment (APEX333APEX, the Atacama Pathfinder Experiment is a collaboration between the Max-Planck-Institut für Radioastronomie at 55%, Onsala Space Observatory (OSO) at 13%, and the European Southern Observatory (ESO) at 32% to construct and operate a modified ALMA prototype antenna as a single dish on the high altitude site of Llano Chajnantor. The telescope was manufactured by VERTEX Antennentechnik in Duisburg, Germany) telescope, located at Llano de Chajnantor in the Atacama desert of Chile (Güsten et al., 2006). PI230 is a dual sideband separating dual polarization receiver. Each band has a bandwidth of 8 GHz and therefore each polarization covers 16 GHz. Each band is connected to two Fast Fourier Transform fourth Generation spectrometers (FFTS4G) (Klein et al., 2012) with 4 GHz bandwidth and 65536 channels, providing a frequency resolution of 61 kHz, i.e., a velocity resolution of $\sim$ 0.079 km s*-1* at 230 GHz. The PI230 receiver can be set up to simultaneously observe CO(2–1), 13CO(2–1), and C18O(2–1). This frequency setting also includes three H2CO lines and one SO line in the lower sideband, see Table 1, all of which were detected in G287.84-0.82.

G287.84-0.82 was mapped in on-the-fly total power mode and sampled every 10″ with an integration time of 0.5 seconds. The map size was $\pm$100″ in RA and from -150″ to +95″ in Dec. The off position used in these observations was at RA =10h 49m 14s, Dec = -59° 35′40″ (J2000). The weather conditions were quite good for 1.3 mm observations. The zenith optical depth varied from 0.146 - 0.156 at 230 GHz resulting in a system temperature of $\sim$ 155 K.

The final data cubes were created using CLASS by resampling the spectra to 0.5 km s*-1* velocity resolution.The rms noise per map resolution element (half a beam width) is $\sim$ 0.18 K.

3 Results

Figure 3 shows a comparison of the integrated intensity image (in K km*-1*) of [C ii] at 158 $\mu$m (contours) overlaid on fluorescent H2, 8 $\mu$m PAH emission, 160 $\mu$m dust continuum as well as on integrated intensity images of most spectral lines that we have mapped. The CO, 13CO and 160 $\mu$m dust continuum images show that the G287.84-0.82 cloud has the shape of a cometary globule with three spatially separated extensions/tails in the south, which all point away from $\eta$ Car and Trumpler 16. The $\eta$ Car is to the northwest, approximately at a projected distance of 12 pc from G287.84-0.82. There is a secondary cloud surface $\sim$60″ south of CPD -59∘2661, which is also roughly perpendicular to $\eta$ Car. On the western side there is a narrow second pillar with a head that looks like the head of a duck, see Fig. 3 in Smith et al. (2005) and Fig. 2. The structure of the cloud is consistent with it being exposed to and shaped by the UV radiation and stellar winds from $\eta$ Car and from the O3 stars in Trumpler 16. The [C ii] emission matches extremely well with the other bona fide PDR tracers like the fluorescent H2, 8 $\mu$m PAH and [O i] 63 $\mu$m emission, suggesting that most of the [C ii] emission originates in the irradiated cloud surfaces. The PDR gas inside the pillar forms a shell around the Treasure Chest cluster. Here the emission appears to be dominated by the stellar cluster with only a minor contribution from the externally illuminated cloud surfaces. High density tracers like CO(11–10), C18O(2–1) and H2CO(303–202) and SO (the latter only shown as a velocity-channel map in Fig. 17) all peak to the north-east of the positions of both the PDR peak and the embedded star cluster. The distribution of intensities and the relative location of the high density peak suggest that the head of the cloud is more compressed towards the north-eastern edge and the PDR shell curving around the Treasure Chest cluster shows that the expanding H ii region both heats up and compresses the surrounding cloud, especially to the north east. Clear detection of H2CO and CO(11–10) at the position of the column density peak suggest gas densities in excess of 105 cm*-3*.

Figure 4 shows the channel maps for the [C ii] and [O i] 63 $\mu$m emission. Both maps show that the bulk of the gas in the head of the region is at a velocity of $\sim$-14.5 km s*-1*, while the tails are generally red-shifted. This suggests that the globule is somewhat behind $\eta$ Car, which has a radial velocity of -20 km s*-1* (Smith, 2004), so that the stellar wind and the radiation pressure from the star push them in the radial direction from $\eta$ Car, making them appear red-shifted. The higher angular resolution of the [O i] 63 $\mu$m observations (particularly around $\upsilon=$-13 km s*-1in Fig. 4) helps identify the dense PDR shell around CPD -59∘2661. The [C ii] channel maps also show a tail-like structure to the east at a velocity of $\sim$ -15 km s-1*, which is not seen in [O i] 63 $\mu$m or CO(11–10) emission (Fig. 12). The comparison with channel maps of all observed CO, H2CO and SO lines (Figs 12, 13, 14, 15, 16, and 17), show that the ”eastern tail-like” -14.8 km s*-1* component is clearly detected in the CO(2–1) channel maps and only partially in the 13CO(2–1) channel maps. This suggests that the -14.8 km s*-1* emission component in the tail has lower density, since it is not detected in the transitions, which either (a) have high critical density like [O i], CO(11–10) and H2CO or (b) trace regions of high column density e.g., C18O(2–1). This gas is probably part of the halo around the globule, which has not yet been affected by the wind from $\eta$ Car.

We next compare the spectra of the different tracers at eight selected positions which are marked in Fig. 2. These positions are selected to sample positions in the head of the globule (#3 to #5), as well as some specific features, such as the CO peak (#1), the [C ii] peak (#2), the neck (#6), the duck head (#7), and the eastern tail (#8). Since the line profiles of the $J$=2–1 transitions of CO (and isotopologues) are similar, we only show the 13CO spectra along with other spectra in Fig. 5. At the position of the column density peak (#1), we find that all spectra have similar line widths, with [O i] being slightly broader than the others. At position

2 the PDR tracers [C ii] and [O i] are significantly broader than the

molecular gas tracers. Table 2 gives the result of fitting the spectra using a single Gaussian velocity component for all positions except position #8 where we have fitted two Gaussian velocity components. We note that for positions #7 and #8 only the tracers that are detected are presented in Table 2.

We find that except for position #1, the [C ii] and [O i] lines are broader than the CO lines by 1–1.5 km s*-1*. The line widths and spectral line profiles of [C ii] and [O i] match very well, suggesting that both [O i] and [C ii] emission originate in PDR gas with no significant contribution from (a) shocks to [O i] and (b) the H ii region to the [C ii] emission. A closer inspection of the [O i] and [C ii] data in the head of the globule suggests that although the outer profiles of the spectra from the two species match well, the [O i] spectra show a flattened or double-peaked profile close to the column density peak and further south-east (e.g. positions 1, 3 and 5 in Fig.5), whereas the [C ii] spectra always show a smooth single-peaked Gaussian profile. The two velocity components detected in CO, 13CO, [C ii], and [O i] at position #8, correspond to the main globule and the red-shifted eastern tail.

Figure 6 shows para-H2CO spectra at three positions where all the three transitions $J$=(303–202), (322–221) and (321–220) are detected with good SNR. The strongest transition, $J$=(303–202), is detected over a much larger area around the column density peak (as shown in Fig. 3 and 17). At the positions (0,26) and (15,26), the para-H2CO  303–202 transition shows a single peaked profile, while at (0,40), where the line is the strongest, there is a hint of a red-shifted component. The position of the peaks of the two fainter transitions match with the brighter peak. The widths of the H2CO lines match well with the line widths of the other molecular lines at these positions. The emission from the (6${}_{5}\rightarrow 5_{4}$) transition of SO spans an even narrower range of velocities and is mostly confined to the higher density head region of the globule (Fig. 17).

The high S/N of the observed [C ii] spectra allows us to simultaneously detect the brightest hyperfine component, F = 2–1, of the 13C+ fine structure line at 50 positions in the area we have mapped. If we average these spectra together, we also see the two fainter hyperfine transitions F = 1–0 and F = 1–1 of [13C ii]. However, they are too faint in individual positions for a meaningful analysis, and we therefore only use the F = 2–1 transition for this purpose. Fig. 7 shows examples of [13C ii] F = 2–1 line, which is offset by +11.2 km s*-1* from the [C ii] line, at two selected positions. The line-integrated intensity of the [13C ii] line was obtained by fitting the [C ii] and [13C ii] lines simultaneously. In this fitting procedure the brighter [C ii] line is fitted first and the fainter [13C ii] line is constrained to be located at a velocity relative to the brighter component given by the velocity separation of the two transitions.

4 Analysis

4.1 LTE & Non-LTE analysis of molecular emission

We derive a first order estimate of the total gas column density using the observed integrated line intensities of CO and 13CO $J$=2–1. Since the CO(2–1) emission is mostly optically thick, we use the maximum temperatures for CO(2–1) as the kinetic temperature ($T_{\rm kin}$). We further assume that $T_{\rm ex}$ for all the isotopes of CO are identical and equal to $T_{\rm kin}$. For all pixels where both CO and 13CO is detected we estimate the $N$(13CO) to lie between (1.3–76)$\times 10^{15}$ cm*-2*. Using a 12CO/13CO ratio of 65 (Rathborne et al., 2004), and CO/H2 = 10*-4*, we find $N$(H2) to range between 8.5$\times 10^{20}$ – 4.9$\times 10^{22}$ cm*-2*. This estimate of $N$(H2) agrees well with what Breen et al. (2018) found from observations of NH3 (1,1) and (2,2) with the Australia Compact Array (ATCA), 2.2$\times$1022 $\pm$ 0.9$\times$1022 cm*-2*, as well as with the estimate by Rebolledo et al. (2016) based on observations of $J$=1–0 transitions of 12CO and 13CO. It also matches reasonably well with the column density estimated by both Roccatagliata et al. (2013) and Schneider et al. (2015) based on dust continuum observations with the Herschel Space Observatory. Our estimate of the CO column density, 8.5$\times 10^{16}$ – 4.9$\times 10^{18}$ cm*-2*, is also consistent with the results by Rathborne et al. (2004), who used CO(2–1) & CO(1–0) observed with SEST to derive $T_{\rm ex}$ = 40 K and $N$(CO) = 1.4$\times 10^{18}$ cm*-2*.

The intensity ratios of the 303–202/322–221 and the 303–202/321–220 transitions of para-H2CO are good thermometers for determining kinetic temperature (Mangum & Wootten, 1993). The 321–220 has a slightly better SNR than 322–221 in the three positions. Therefore we use the observed 321–220/303–202 ratios to estimate the kinetic temperature ($T_{\rm kin})$ in these positions. To constrain the kinetic temperature, density and column density of the gas, a grid of models based on the non-Local Thermodynamic Equilibrium (non-LTE) radiative transfer program RADEX (van der Tak et al., 2007) was generated and compared with the intensity ratios of para-H2CO. Model inputs are molecular data from the LAMDA database (Schöier et al., 2005) and para-H2CO collisional rate coefficients from LAMDA (Wiesenfeld & Faure, 2013). RADEX predicts line intensities of a given molecule in a chosen spectral range for a given set of parameters: kinetic temperature, column density, H2 density, background temperature and line width. A value of 2.73 K is assumed as the background temperature for all calculations presented here. The synthetic line ratios are calculated for a line width of 2 km s*-1*, close to the line widths we deduce for the three positions. The observed line ratios of 303–202/321–220 at the three positions (14,26), (0,26), and (0,40) are 3.0, 3.4, and 4.4 respectively. We generated a grid of models with $T_{\rm kin}$= 20–120 K, $n$(H2) = 104–107 cm*-3* and $N$(H2CO) = 1012–1015 cm*-2*. Figure 8 shows a plot of the predicted 303–202/321–220 ratios as a function of $T_{\rm kin}$ and $N$(H2CO) for $n$(H2) = 105 and 106 cm*-3*. The observed values of the H2CO ratios are shown as contours, which constrain the temperature very well up to $N$(H2CO) of 2$\times 10^{13}$ cm*-2*. Based on observations of high mass star forming clumps, the value of $N$(H2CO)/$N$(H2) ranges between 10*-10* to 10*-9*, which correspond to $N$(H2CO) of 1012–1013 cm*-2* for the derived $N$(H2) of $\sim 10^{22}$ cm*-2* for the globule.

The critical density of the H2CO transitions that we observe is 1–6 105 for 50 K gas (Shirley, 2015) and the critical densities for low-$J$ CO and NH3 are much lower ($\sim 10^{3}$ cm*-3* for NH3). Thus, while a density of 104 cm*-3* is too low, $n$(H2) of 106 cm*-3* appears to be too high. We thus assume a gas density, $n$(H2) = 105 cm*-3*, which we consider to be a reasonable compromise. The outcome of the analysis stays about the same even for slightly higher densities. With the assumption of a gas density, $n$(H2) = 105 cm*-3*, we obtain kinetic temperatures of 55, 75, and 90 K respectively. However, we note that since the H2CO lines are weak, the uncertainties in $T_{\rm kin}$ determined from spectra at individual positions are also high. In order to derive a more reliable estimate of $T_{\rm kin}$ we averaged the H2CO lines over an area where all three lines are detected. The observed line ratios then lie in the range of 3.2–3.4 for 322–221 and 321–220, respectively. The average kinetic temperatures is therefore in the range 60–70 K. Breen et al. (2018) used NH3 (1,1) and (2,2), another good molecular gas thermometer (Wamsley & Ungerechts, 1983), to derive a rotational temperature, Trot, of 23 K averaged over the area where both (1,1) and (2,2) was detected. This correspond to a kinetic temperature of 39 K, which is similar to what we get from CO(2–1). The kinetic temperatures estimated from far infrared dust emission (Roccatagliata et al., 2013), are somewhat lower. Roccatagliata et al. (2013) find $\sim$ 32 K for the head region and $\sim$ 25 K for the column density peak. Thus, the $T_{\rm kin}$ derived from H2CO is higher than other estimates, although the reason for this discrepancy is not clear.

4.2 C+ column density

While [C ii] is mostly known to be optically thin, recent velocity-resolved observations of Galactic massive star forming regions have shown that this is not always the case, because the detection of strong hyperfine transitions of 13C+, indicate that [C ii] can be significantly optically thick (Mookerjea et al., 2018; Ossenkopf et al., 2013, among others). If [13C ii] is not detected, the determination of C+ column density, $N$(C+) requires assumptions about the excitation temperature or using the same excitation temperature as molecular emission. Simultaneous detection of the [C ii] and [13C ii] lines is independent of calibration errors and allow us to determine the excitation temperature and optical depth, if [C ii] is optically thick. The F = 2–1 hyperfine line of [13C ii] was detected in 50 positions. As described in Section 3 we derived the integrated line intensities of the [13C ii] emission at all these positions and below we estimate the $T_{\rm ex}^{\rm CII}$ and $N$(C+) following the formulation by Ossenkopf et al. (2013).

The optical depth $\tau_{12}$ of the [C ii] line is determined using

[TABLE]

where $I_{12}$ and $I_{13}$ represent $\int T^{CII}_{\rm mb}d\upsilon$ and $\int T^{{}^{13}CII}_{\rm mb}d\upsilon$ respectively, and $\frac{\rm{}^{12}C}{\rm{}^{13}C}$ = 65 (Rathborne et al., 2004).

The optical depth derived for [C ii] range between 0.9 to 5.

The optical depth-corrected integrated intensity of [C ii] is derived from

[TABLE]

Subsequently, the excitation temperature of [C ii] ($T_{\rm ex}^{\rm CII}$) is derived by dividing Eq (1) by Eq (3) from Ossenkopf et al. (2013)

[TABLE]

The values of $T_{\rm ex}^{\rm CII}$ vary between 80 to 255 K. Most positions show values between 110–130 K, which is consistent with the $T_{\rm kin}$ derived (Sec. 4.1) from H2CO and CO(11–10).

Because we know the optical depth and the excitation temperature, we can now estimate the column density of C+ by inverting the following Equation (3) of Ossenkopf et al. (2013):

[TABLE]

The column density of C+ ranges between 3–7$\times 10^{18}$ cm*-2*.

4.3 Mass of the cometary globule

Since we have mapped most of the globule in CO(2–1) and isotopologues we can use our maps to estimate the total mass of the globule. For this we use 13CO(2–1) and compute a column density for each pixel in the map, where 13CO(2–1) is detected. For the column density estimates we use the LTE approximation and estimate $T_{\rm ex}$ from the peak of 12CO at the same position and then derive the opacity $\tau$ for 13CO, i.e. following the method outlined in White & Sandell (1995). With the assumptions that $\frac{\rm{}^{12}C}{\rm{}^{13}C}$ = 65 and $\frac{\rm CO}{\rm H_{2}}$ = 10*-4* we get a total mass of $\sim$ 600 $\mathrm{M}_{\odot}$.

We derive similar mass estimates from the observed [C ii] intensities by assuming the line to be optically thin, a gas density of $n=n_{\rm cr}$=3000 cm*-3* and an excitation temperature, $T_{\rm ex}$ = 100 K. If we assume that half of the available carbon is in C+ (0.65 $\times$ 10*-4*), then the total mass of the globule seen in [C ii] is $\sim$ 440 $\mathrm{M}_{\odot}$, which is comparable to what we derived from 13CO(2–1). This is an underestimate, since we know that there are positions where [C ii] is optically thick. The fraction of pixels where [C ii] is optically thick, however, is only $\sim$ 8% of all pixels in which [C ii] was detected. Therefore we at most underestimate the mass by 15%. The uncertainty in our mass estimate is completely dominated by how much of the carbon is in [C ii].

The masses we derive from CO and C+ are also in reasonably good agreement with the value of 1200 $\mathrm{M}_{\odot}$ estimated from the 1.2 mm dust continuum emission by Rathborne et al. (2004). The mass estimates from different tracers are therefore comparable and suggest a total mass of the globule of $\sim$ 1,000 $\mathrm{M}_{\odot}$.

4.4 Position-Velocity diagrams along the cuts

In order to understand how the PDR emission relates to the molecular gas in the globule, we have made position-velocity (PV) plots in [C ii], [O i] and 13CO(2–1) along the two cuts shown in Fig. 2. The PV plots were made using the task velplot in Miriad (Sault et al., 1995). Velplot uses a 3 $\times$ 3 pixel interpolation kernel resulting in beam sizes of 21${}^{\prime\prime}\mskip-7.6mu.\,$2, 11${}^{\prime\prime}\mskip-7.6mu.\,$1, and 28${}^{\prime\prime}\mskip-7.6mu.\,$5 for [C ii], [O i], and 13CO(2–1)444For 13CO(2–1) we created a map with a pixel size of 9${}^{\prime\prime}\mskip-7.6mu.\,$5 $\times$ 9${}^{\prime\prime}\mskip-7.6mu.\,$5 , i.e., one third of a HPBW., respectively. Fig. 9 shows the PV plots along the cut going from: (top) north-west to south-east (NW-SE) and (bottom) north-east to south-west (NE-SW).

Along the NW-SE cut, the PDR tracers rise sharply in the northwest and then fall off and become narrower after passing through the H ii region powered by CPD -59∘2661. The tail region is distinctly red-shifted in the eastern tail of the globule, although both [C ii] and 13CO(2–1) show a velocity component at -14.5 km s*-1*, probably arising from unperturbed gas from the head of the globule. As we already saw from the channel maps, this velocity component is not seen in C18O or [O i], suggesting that it is more diffuse, low density gas. The [C ii] and [O i] PV plots show clear peaks where the cut intersects the PDR shell illuminated by CPD -59∘2661, while 13CO(2–1) peaks a bit further south. The 13CO(2–1) line width remains unchanged across the PDR shell, which indicates that the molecular gas in the globule is not yet affected by the expanding H ii region.

In the NE-SW position-velocity plot the PDR shell stands out clearly in [C ii] and [O i] being more blue-shifted close to the exciting star (marked by O in Fig. 9) while 13CO(2–1) peaks at the molecular column density peak. [O i]  and [C ii] also have peaks at the south western PDR rim of the main globule. [O i] is not present in the low density region between the main globule and the Duck neck, and 13CO(2–1) is very faint or absent as well. [C ii], however, is relatively strong and distinctly redshifted, presumably low density gas expanding from the PDR region illuminated by CPD -59∘2661. Both [C ii] and [O i] show extended emission on the side of the neck facing towards the Treasure Chest, indicating that the neck is associated with the PDR region illuminated by CPD -59∘2661. At the angular resolution of our observations the northern PDR illuminated by $\eta$ Car and the PDR shell illuminated by CPD -59∘2661 blend together, whereas they are seen as two separate PDRs in the 8 $\mu$m image, see Fig. 2. The northern PDR, however, is close to the systemic velocity of the globule head, while the internal PDR shell is more blue-shifted north of the star. In the NE-SW cut we see a similar jump in velocity between the external PDR rim and the PDR shell.

Both position-velocity plots show that there is [C ii] emission outside the thin PDR interface of the globule. This is the same photo evaporated gas seen for example in ionized Paschen $\beta$ (Hartigan et al., 2015). At the sharp northern PDR surface, one can see faint [C ii] emission tracing photo evaporated gas up to $\sim$ 20″ north of the head of the globule.

5 Discussion

We compare the high spatial resolution (7″) velocity-resolved observations of the [O i] at 63 $\mu$m lines in G287.84-0.82 with the [C ii], CO (and its isotopologues) and H2CO to understand the small-scale structure of the PDR created by the embedded Treasure Chest cluster. As shown in Fig. 4 the emission from the region has clearly blue- and red-shifted components. Figure 10 compares the emission in the blue ($\upsilon_{\rm LSR}$ = -19 to -14 km s*-1*) and red ($\upsilon_{\rm LSR}$ = -13 to -10 km s*-1*) parts of the spectrum. In the blue part we find [O i] tracing out mostly the higher density regions including the PDR surface to the southwest of the main pillar. The same is seen in [C ii], which additionally is seen in the tail regions of the globule. The overlay of red-shifted [C ii] and [O i] emission also match very well, suggesting the PDR shell surrounding CPD -59∘2661 is dominantly at this velocity. The eastern, red-shifted tail of the globule is strong in both [C ii] and [O i], while the middle tail is only seen in [C ii], suggesting that it has lower densities. The western pillar (the Duck Head) clearly stands out in the high spatial resolution [O i] data and is seen in [C ii] as well. It appears to have about the same velocity as the main pillar. The head is seen in CO(11–10) as well.

The higher spatial resolution [O i] data resolve the PDR shell around the Treasure Chest cluster, with the semi shell being easily seen in the red-shifted [O i] image. This is also consistent with our [C ii] data. There is little overlap between the peaks in the blue- and red-shifted emission, which is consistent with an expanding shell, in which the north-eastern part is somewhat blue-shifted, while the emission is more red-shifted to the west. The red- and the blue-shifted components of [C ii] and [O i] emission differ in velocity by $\sim$ 4 km s*-1*. This is much smaller than the velocity split seen in H$\alpha$, 25 km s*-1*, toward CPD -59∘2661 by Walsh (1984). Although Walsh (1984) interpreted it as a measure of the expansion velocity (12 km s*-1*) of the nebula, he noted that ”there is no evidence for the split components gradually joining up to the velocity of the adjacent material, as might be expected for a shell.” Our observations show that the expansion velocity is only $\sim$ 2 km s*-1*, hence the dynamical age of the H ii region, a few times 104 yr estimated by Smith et al. (2005), is severely underestimated. In order for the H ii region to expand into the surrounding dense molecular cloud, a more realistic estimate of the age is $\sim$ 1 Myr and this is in good agreement with the age of the cluster, 1.3 Myr, as estimated by Oliveira, Bica & Bonatto (2018).

5.1 The PDR gas

We can to some degree distinguish the emission from the diffuse and dense components of the gas by looking at the emission in different velocity intervals. Thus, if we compare the intensities due to the blue-shifted [C ii] and CO(2–1) emission from the eastern tail with PDR models, it will be possible to derive an estimate of the density of the diffuse gas. Additionally, we use the line intensities of [O i] and CO(11–10), both high density PDR tracers, at the positions of the intensity peaks of [O i] seen clearly in the shell around the embedded cluster (Fig. 10).

In order to estimate the physical conditions in the dense and diffuse PDR in the cloud surrounding the Treasure Chest nebula we compare the observed line intensity ratios with the results of the model for PDRs by Kaufman, Wolfire & Hollenbach (2006). These models consider a semi-infinite slab of constant density, which is illuminated by far-ultraviolet (FUV) photons from one side. The model includes the major heating and cooling processes and incorporates a detailed chemical network. Comparison of the observed intensities with the steady-state solutions of the model, helps to determine the gas density of H nuclei, $n_{\rm H}$, and the FUV flux (6 eV $\leq$ h$\nu$ $<$ 13.6 eV), $G_{0}$, measured in units of the Habing (1968) value for the average solar neighborhood FUV flux, 1.6$\times 10^{-3}$ ergs cm*-2* s*-1*.

Since the PDR inside the Treasure Chest cloud is created by the embedded cluster, the strength of the FUV radiation in the region can be estimated from the observed total far-infrared (FIR) intensity, by assuming that FUV energy absorbed by the grains is re-radiated in the FIR. Based on this method Roccatagliata et al. (2013) have derived the FUV map for the region using PACS and SPIRE dust continuum observations, with the values of FUV radiation field lying between 900 to 5000 times G0, where G0 is in units of the Habing field. From their FUV map (Fig. 7 in Roccatagliata et al. (2013)) we estimate the FUV intensity to be around 1000 G0 in the eastern tail and around 5000 G0 in the region where the PDR shell is detected.

In order compare the observed intensity ratios of [C ii]/CO(11–10), [O i]/CO(11–10), [O i]/[C ii] and [C ii]/CO(2–1) for the dense gas with PDR models, we compare emission arising in the same velocity range for the tracers involved in a particular ratio. Since among all the tracers [C ii] is the broadest at most positions, we first fit a single Gaussian to the spectrum of the other tracer in the ratio. Subsequently we fit a Gaussian with identical central velocity and FWHM to obtain the velocity integrated line intensity for [C ii]. In order to do this position-by-position analysis we spatially regridded the datacubes corresponding to a particular ratio to the grid of the cube with lower resolution. Thus the analysis is restricted to around 300 positions where such fits were possible. We find that the intensity ratios (in energy units) [C ii]/CO(11–10), [C ii]/CO(2–1) and [O i]/[C ii] range between 10–70, 50–450, and 3–7 respectively for these positions. For the [O i]/CO(11–10) we fit the CO(11–10) spectra first and then use the same parameters to extract the intensity of the [O i] spectrum. The [O i]/CO(11–10) intensity ratio thus estimated range between 30 to 210. For the diffuse gas traced only by [C ii] and CO(2–1), we use the ratio of intensities of [C ii] and CO(2–1) between -18 to -14 km s*-1* in the eastern tail region, where no [O i] emission is seen at these velocities. The CO(2–1) intensities are estimated from the observed intensities of the optically thin 13CO(2–1) spectra. The [C ii]/CO(2–1) ratio for the eastern tail ranges between 1000–2000 (in energy units).

Figure 11 shows results of comparison of the observed intensity ratios with the predictions of the PDR models. The two horizontal lines correspond to 1000 and 5000 G0 for the western tail and the PDR shell around the cluster. For each intensity ratio two contours are drawn to mark the range of observed values. We assume that the beam-filling factors of the different PDR tracers are identical so that the line intensity ratios considered here are independent of the beam-filling.

The observed [C ii]/CO(2–1) ratios for the diffuse gas in the eastern tail are consistent with $n$(H) between 600 to 2200 cm*-3*. For the denser gas the [C ii]/CO(2–1) suggest a somewhat wider range of values between 104–106 cm*-3*. The [C ii]/CO(11–10) primarily estimated in the head of the globule suggest a narrower range of densities between (0.8–1.6)$\times 10^{5}$ cm*-3*. The [O i]/CO(11–10) ratio with both high density tracers suggest densities between (2–8)$\times 10^{5}$ cm*-3*. The [O i]/[C ii]/ ratio estimated for positions spanning both the head and tails of the globule suggest densities between 2 103–1.8 104 cm*-3*. The much lower densities suggested by the [O i]/[C ii]/ ratio indicate that the [C ii] emission is dominated by lower density, more diffuse PDR gas.

5.2 Thermal Pressure in the region

Based on the analysis of the tracers of molecular and PDR gas of different densities we identify several temperature components in the G287.84-0.82 globule. The first component corresponds to the temperature of the molecular gas, which is estimated to be $\sim 40$ K based on the observed peak temperatures of CO(2–1). The second component corresponds to PDR gas at a temperature of 100 K, as derived from the analysis of [13C ii] intensities. Additionally, the high density ($\sim 10^{5}$ cm*-3*) molecular gas emitting in H2CO show average kinetic temperatures of 60–70 K within a limited region in the head of the globule. It is likely that a fourth component with densities of $\sim 600$–2200 cm*-3*, and not so well determined temperature exist in the tail regions.

From analysis of [O i] and CO(11–10) using PDR models we determine the densities of dense PDR gas to lie between (2–8)$\times 10^{5}$ cm*-3*. For typical PDR temperatures of 100-300 K, this corresponds to a thermal pressure of 2 107–2.4 108 K km s*-1*. Ratio of intensities involving [C ii] suggest for the diffuse PDR gas, densities around 104 cm*-3*, which for a temperature of 100 K correspond to a thermal pressure of 106 cm*-3*. The global pressure of the molecular gas traced by CO(2–1) with a density of $\sim 10^{4}$ cm*-3* and $T_{\rm kin}$= 40 K corresponds to 4$\times 10^{5}$ K cm*-3*. If we consider the diffuse PDR gas in the eastern tails with densities between 600–2200 cm*-3* to have thermal pressure ($P_{\rm th}$) of the same order as the cold molecular gas, then the temperature of the gas would correspond to temperatures exceeding 200 K.

The sharp contrast in estimated thermal pressure of the diffuse and dense phases of PDR gas in the region, is similar to the results of Stock et al. (2015), where two phases with pressures of 105 K cm*-3* and 108 K cm*-3were detected in multiple sources. It is also similar to the results of Wu et al. (2018), who estimate $P_{\rm th}\sim 10^{8}$ K cm-3* across dissociation front in Car I-E region. Wu et al. (2018) has also suggested an empirical relationship $P_{\rm th}=2.1\times 10^{4}\,G_{\rm UV}^{0.9}$ between the UV field and thermal pressure. For GUV $\sim$ 5000 G0 for the dense regions, $P_{\rm th}$ will be $\sim 4\times 10^{7}$ K cm*-3*, which is approximately consistent with the $P_{\rm th}$ we derive from the $n$(H2) and $T_{\rm kin}$.

5.3 Is the Treasure Chest a case of triggered star formation?

Although it is clear that star formation can be triggered in cometary globules (Lefloch et al., 1997; Ikeda et al., 2008; Mookerjea & Sandell, 2009), this is not the case for the Treasure Chest cluster. There are 19 OB clusters in the Carina Nebula (Tapia et al., 2003; Smith, 2006; Oliveira, Bica & Bonatto, 2018). Even though the Treasure Chest is one of the youngest, 1.3 Myr (Oliveira, Bica & Bonatto, 2018), it is still too old to have been triggered by the formation of the cometary globule. It is far more likely that the Treasure Chest had already formed before G287.84 became a cometary globule and then it was a dense, massive cloud core surrounded by lower density gas. When the Carina H ii region expanded, it blew away the lower density gas, first creating a giant pillar, which later became the cometary globule that we see today. Although the cometary globule continues to be eroded by the stellar winds from $\eta$ Car and Trumpler 16, the expanding H ii region from the Treasure Chest cluster appears to erode it even faster. It has already expanded through the western side of the globule and most of the molecular cloud is already gone on the northwestern side as well. There is, however enough dense gas, $\sim$ 1,000 $\mathrm{M}_{\odot}$, which could collapse and form stars, although it is likely that the time scale for star formation is longer than the time it takes for the globule to be evaporated by the FUV radiation from the Treasure Chest cluster and the FUV radiation from $\eta$ Car and and the O stars in Trumpler 16.

The heavily reddened young star, 2MASS J10453974-5957341 in the Duck head, could be a case of triggered star formation, although it is more likely that it formed about the same time the Treasure Chest cluster formed, and is therefore more likely a member of the Treasure Chest cluster.

6 Summary

We have used velocity-resolved observations of various tracers of (i) PDR ([O i], [C ii], CO(11–10)), (ii) ambient molecular material ($J$=2–1 transitions of CO, 13CO and C18O), and (iii) high density gas (H2CO and SO) to study detailed physical and velocity structure of the G287.84-0.82 cometary globule in Carina. The high quality of [C ii] data enabled simultaneous detection of the $F$=2–1 transition of the rarer isotope of C+ and subsequent determination of $N$(C+)$\sim 0.3$–1$\times 10^{19}$ cm*-2*. We conclude that the overall structure of the source including its velocities is consistent with the globule being sculpted by the radiation and winds of the nearby OB associations in Trumpler 16 and $\eta$ Carina. The details of the distribution of PDR gas inside the globule as revealed with enhanced clarity by the higher spatial resolution (7″) [O i] data, is consistent with illumination by the brightest member, CPD -59∘2661 of the Treasure chest cluster. The density and temperature of the PDR gas, estimated to consist of a diffuse and a dense part were determined using observed line intensity ratios in combination with non-LTE radiative transfer models as well as plane-parallel PDR models. We identify at least two components of PDR gas with densities of 104 and (2–8)$\times 10^{5}$ cm*-3*. Based on temperatures estimated, the thermal pressure in the diffuse and dense PDR components are 106 K cm*-3* and 107–108 K cm*-3* respectively. This contrast in thermal pressure of the diffuse and dense phases of PDR gas is similar to the findings which exist in the literature.

Acknowledgements.

Based in part on observations made with the NASA/DLR Stratospheric Observatory for Infrared Astronomy (SOFIA). SOFIA is jointly operated by the Universities Space Research Association, Inc. (USRA), under NASA contract NNA17BF53C, and the Deutsches SOFIA Institut (DSI) under DLR contract 50 OK 0901 to the University of Stuttgart.

Appendix A Additional channel maps

Here we show channel maps of CO(11–10), 2–1 transitions of CO (and isotopologues), H2CO(303–202) and SO(6${}_{5}\rightarrow 5_{4}$).