The warm-up phase in massive star-forming cores around RCW 120

The initial stage of high-mass star formation is a subject of debates. While starless dense clumps of molecular clouds, precursors of protostellar cores for low-mass stars, were observed by, for example, Tafalla et al. (1998), similar high-mass objects are rare and more difficult to find and confirm (see, e. g. modelling by Pavlyuchenkov et al., 2011). Several candidates were proposed by Sridharan et al. (2005); Olmi et al. (2010); Fontani et al. (2012); Svoboda et al. (2019); Zhang et al. (2020). One of the most massive and well-known dense gas condensations at the border of the H ii region RCW 120 has been considered as a high-mass analogue of a Class 0 object for about ten years since the studies by Deharveng et al. (2009); Zavagno et al. (2010). In particular, Deharveng et al. (2009) found a compact 870 $\mu$m core of 250 M_⊙ adjacent to the ionization front of RCW 120, which is embedded into a 800-M_⊙ gas condensation at the south-western border of RCW 120 (Clump 1 hereafter). Zavagno et al. (2010), using Herschel infrared data, confirmed Class 0 properties of the object and found that the emission towards the dense core is dominated by its 10³ M_⊙ envelope. The formation of the gas condensation with the massive dense core at the border of the H ii region was proposed as an outcome of a triggering collect-and-collapse (C&C) process related to expansion of RCW 120 in the studies by Zavagno et al. (2007); Deharveng et al. (2009); Minier et al. (2013). Tremblin et al. (2013) suggested that compression by an ionization front played a major role, along with gravity, for the dense core formation in Clump 1. Recently, Figueira et al. (2020) re-examined the triggered star-formation model with new observations of CO lines made by the APEX telescope, and confirmed that it might be at work at the edges of RCW 120. Zavagno et al. (2020) found a filamentary structure of a photodissociation region around RCW 120 using ArTéMiS on APEX. They concluded that the compression of pre-existing molecular filaments in Clump 1 may influence star-forming regions there.

Figueira et al. (2017) re-examined the Herschel data and identified 35 compact young stellar objects (YSOs) with reliable SEDs associated with RCW 120. They found that four out of five massive YSOs near RCW 120 are located in Clump 1. Analysing the ALMA data, Figueira et al. (2018) found that the most massive YSO (Core 2 hereafter, an object with ID 2 in Table 5 in Figueira et al., 2017) probably associated with an ultra-compact H ii region (ucH ii). They reported the detection of CH₃CN and SO₂ emission at 3 mm but did not analyse this emission in detail and did not determine the physical properties of the gas in the cores.

In this paper, we analyse data on molecular emission in Clump 1 at 1 mm in order to determine physical parameters of the gas in Core 2, and also in Core 1, which is the second most massive YSO after Core 2 in Clump 1. We focus mainly on methanol and CH₃CN emission, as these molecules are reliable tracers of physical conditions in molecular gas. Methanol is an abundant, well-known interstellar molecule, especially in regions of star formation (see, e. g. Kalenskii et al., 2002; Leurini et al., 2007). Being a slightly asymmetric rotor, methanol has a complex spectrum, sensitive to the physical conditions in dense interstellar gas (Kalenskii & Sobolev, 1994; Leurini et al., 2004; Salii, 2006). The CH₃CN transitions are used to obtain independent measurements of gas kinetic temperature, as has been shown, for example by Kalenskii et al. (2000); Gratier et al. (2013); Beltrán et al. (2018). We describe our observations in Sec. 2; we present our results (physical properties of the gas and molecular abundances) and analysis in Sec. 3. Our theoretical modelling of the chemical evolution of Core 2 is presented in Sec. 4; those results are discussed in Sec. 5. The summary of our study is presented in Sec. 6.

In July 2009, using APEX¹¹1This publication is based on the data acquired with the Atacama Pathfinder Experiment (APEX). APEX is a collaboration between the Max-Planck-Institut fur Radioastronomie, the European Southern Observatory, and the Onsala Space Observatory., we observed several selected positions in Clump 1 around the maximum of 870 $\mu$m emission found by Deharveng et al. (2009): $\alpha_{2000}=17^{\rm h}12^{\rm m}08^{\rm s}$ and $\delta_{2000}=-38^{\circ}$30′ 45″  (project number O-083.F-9311A-2009). Several selected positions and methanol spectra at 241791.431 MHz are shown in Fig. 1. Relative coordinates of the all observed positions are given in Table 4. The observations were carried out in service mode in good weather conditions, with a Precipitable Water Vapor (PWV) amount between 0.4 and 0.7 mm. Two spectral windows²²2The observations in one more window, centered at $330$ GHz, were presented by Kirsanova et al. (2019) and will not be discussed here., at $\sim 220$ and $\sim 241$ GHz, were observed with the SHeFI receiver APEX-1 (Belitsky et al., 2006; Vassilev et al., 2008). The spectrometer consisted of two FFTS units configured to provide a spectral resolution of 122 kHz (0.17 km s^-1) and to cover frequency bands of 1400 MHz for the 220 GHz spectral window and 1200 MHz for the 241 GHz window (0.16 km s^-1). The main instrumental parameters are summarised in Table 1. The data reduction was performed using the GILDAS software³³3http://www.iram.fr/IRAMFR/GILDAS. The identification of the detected lines was performed using the CDMS database (Endres et al., 2016).

The observed methanol line series allow us to estimate excitation conditions and line opacities. We used large velocity gradient (LVG) analysis to determine the physical parameters of gas where the methanol emission appears, namely: gas kinetic temperature ($T_{\rm gas}$, K), hydrogen number density ($n_{\rm H_{2}}$, cm^-3), methanol specific column density ($N_{\rm CH_{3}OH}/\Delta V$, cm^-3 s) and methanol relative abundance ($N_{\rm CH_{3}OH}/N_{\rm H_{2}}$). Since the beam filling factor for the methanol emission ($f$) is not known, we include this additional parameter in the methanol line intensity analysis. The details of the LVG calculation and a database of population numbers for the methanol energy levels, calculated for several values of line width (1, 3 and 5 km s^-1) can be found in, for example, the work by Kirsanova et al. (2017); Salii et al. (2018). In the present calculations, we use a line width of 5 km s^-1 for the detected methanol lines since this is the closest value to the observed widths given in Table 2. Dust emission and absorption within the emission region are taken into account, as described by Sutton et al. (2004), where dust and gas temperatures were set to be the same. Using this approach, we can obtain a fit with the best coincidence between the modelled and observed line intensities (with $\chi^{2}$ minimum). The upper limits of the undetected methanol lines were included in the consideration: it was controlled that their model intensities not exceed the rms level from Table 1.

To extend the analysis of the methanol emission and estimate the parameters in the regions with the weaker lines, we applied a Bayesian analysis (e.g. Ward et al., 2003) to obtain the confidence intervals. We calculated methanol model intensities ($T_{{\rm m}\,i}(p)$) for a regular network of parameters:

using the database of the population numbers for methanol quantum energy levels.

The probability to observe $N$ methanol lines with intensities $T_{{\rm o}\,i}$ and uncertainties $\sigma_{i}$ using a set of parameters $p$ can be calculated as:

Integrating over each parameter, we obtain the Bayesian probability function and estimate the parameters and confidence intervals.

The analysis of the CH₃CN line emission was done with the population diagram method (see, e.g. Goldsmith & Langer, 1999; Kalenskii et al., 2000). The column densities of CS and SO₂ molecules were determined under LTE assumption using a standard approach described by Mangum & Shirley (2015) and using the gas kinetic temperature derived from the methanol emission analysis as the excitation temperature in the LTE approach. We recognise all the restrictions of the LTE approach and consider the LTE analysis only as a way to obtain estimates of molecular column densities since we have only one line for each molecule. To convert column density of C³⁴S to CS, we used an elemental abundance ratio [S]/[³⁴S] = 22.5 (Wilson, 1999).

In order to calculate the molecular abundances relative to molecular hydrogen from their column densities, we use the column density of H₂ molecules $N_{\rm H_{2}}=3.2_{1.6}^{4.8}\times 10^{23}$ cm^-2 obtained in the ViaLactea project (Marsh et al., 2015, 2017), which is also in agreement with the column density values obtained by Anderson et al. (2012); Tremblin et al. (2013) and Figueira et al. (2017).

The full list of the detected lines in Cores 1 and 2 and the parameters of Gaussian fits are shown in Table 2. The spectra of the methanol and CS line emission are shown in Fig. 1, 2 and 3, while the spectra of CH₃CN are shown in Fig. 5. We note that the lines with an upper level energy $E_{\rm u}>50$ K are detected only towards Core 2. Three lines with $T_{\rm mb}\approx 0.5-1$ K around 220050, 241895 and 240903 MHz remain unidentified in Fig 2. The brightest among of them at 220050 MHz is detected in both Core 1 and 2 with $T_{\rm mb}\approx 1$ K.

In order to check whether our abundance estimates are viable for any particular stage of the star-formation process, we performed a simple astrochemical analysis relying upon the physical parameters found above for Core 2. The astrochemical model PRESTA (Kochina et al., 2013) is used for that purpose. This model solves the chemical kinetic equations accounting for both gas-phase and solid-phase processes, which are taken from the ALCHEMIC database (Semenov & Wiebe, 2011) with additions described by Wiebe et al. (2019). Abundances and physical constraints are summarised in Table 3. We note that chemical modelling procedure normally uses number density of hydrogen atoms $n_{\rm H}=2n_{\rm H_{2}}$.

First, we run a small set of models for number densities $n_{\rm H}$ from $10^{4}$ cm^-3 to $10^{6}$ cm^-3 and for temperatures from 10 K to 30 K (gas and dust temperatures are assumed to be equal), neglecting any UV irradiation, except for cosmic-ray-induced photons. The results, shown in Fig. 6, indicate that the observed methanol abundance can only be reproduced in models with $T=20$ K, $n_{\rm H}=10^{4}-10^{5}$ cm^-3, and a chemical age of about 1 Myr or greater. This age should be treated with caution, as its zero moment probably predates the formation of RCW 120. The abundances of CH₃CN and CS give more weight to $n_{\rm H}=10^{4}$ cm^-3 (dotted lines in Fig. 6), while the abundance of SO₂ apparently indicates $n_{\rm H}=10^{5}$ cm^-3 (dashed lines in Fig. 6). These results should not be considered as mutually exclusive, as it is quite possible that the observed lines of various molecules do not arise in exactly the same conditions.

Given the high H₂ column density ($3.2\times 10^{23}$ cm^-2) and the small physical size of the cores ($\sim 0.1$ pc), the average density is, indeed, of the order of $10^{5}$ cm^-3. However, our modelling seemingly indicates that higher densities, $\geq 10^{6}$ cm^-3, in combination with the low temperature, are excluded for methanol, as in that dense medium it mostly resides on dust grain surfaces (e.g. Garrod & Herbst, 2006; Wirström et al., 2011; Punanova et al., 2018). It is possible to obtain both high density and high gas-phase methanol abundance by adding a warm-up phase. We treat this in a simple manner, by considering a second stage with the same density of $10^{6}$ cm^-3, but with elevated temperatures. To reproduce the observed abundance, a growth of the gas/dust temperature up to $\sim 90$ K or higher is needed to evaporate a sufficient amount of methanol from icy mantles for the adopted desorption energy of 4235 K (Cuppen et al., 2017). It should be kept in mind that even in this case, the gas-phase methanol abundance rapidly drops due to enhanced ion-neutral chemistry. Also, after the temperature increase, the abundances of other species grow well above the observed values. This again hints that the emission regions for the various molecules are not spatially coincident. For example, there is a possibility that the methanol emission is associated with outflows (see Sec. 5), and we find line wings on the observed spectra of methanol in contrast with the CS spectra. Note that the abundances of CH₃CN, CS and SO₂ can be underestimated because of the unknown beam filling factor for these molecules.

In summary, we conclude that our estimates for the abundances of CH₃OH, CH₃CN (low-excitation value), CS and SO₂ are reasonable. The specific values of the abundances hint at the warm-up phase in Core 2. The detection of the high-excitation CH₃CN lines suggest that the central source, responsible for the warm up, might develop a compact hot core, but does not provide enough warming to enhance the abundances up to values typical for hot gas. Therefore, we may be observing the beginning of hot-core chemistry in Core 2. To extract more clues from the molecular composition of the cores, more thorough modelling is needed to take into account the physical and evolutionary complexity of the region.

Core 1 and Core 2 consist of two and five fragments, respectively, on a 0.01 pc scale, according to the ALMA 3-mm continuum observations made by Figueira et al. (2018). The most massive fragment is found in Core 2, with an estimated mass of 27 M_⊙ with a factor of 2 for the absolute mass uncertainty. Figueira et al. (2018) detected CH₃CN and SO₂ lines there and proposed the existence of an ucH ii region on the basis of the detection of both of these lines. However, no radio continuum emission point sources was found toward Clump 1 in the NVSS image (Kirsanova et al., 2019).

Voronkov et al. (2014) detected six methanol maser spots at 36 GHz and seven spots at 44 GHz towards Clump 1. Four of these are observed in the same directions (their source G348.18+0.48). These Class I methanol masers arise via collision pumping by molecular hydrogen (e.g. Cragg et al., 1992; Leurini et al., 2016) and are often considered as tracers of interface regions of outflows (see, e.g. Voronkov et al., 2012). Indeed, the maser spectra in the spots observed by Voronkov et al. (2014) have profiles with significant wings, indicating complex kinematics and probably outflows in Clump 1. The wings of the thermal methanol lines reported by us suggest that both types of the methanol emission (maser and partly thermal) arise in the outflows. Water vapour masers were detected here by Braz & Scalise (1982), which are also associated with collision pumping and molecular outflows (e.g. Genzel et al., 1981; Gwinn et al., 1992; Reid et al., 1995), but sometimes also with ucH ii regions (Hunter et al., 1994). However, no methanol maser at 6.7 GHz was found here during the Methanol Multibeam Survey (Caswell et al., 2010), while the 6.7 GHz masers are considered as excellent signposts of the location of massive YSOs (e.g. Breen et al., 2013) and often observed near ucH ii regions (e.g. van der Walt et al., 2003).

We show that warming-up gas-phase chemistry, related with the appearance of a heating source, is observed in Core 2. The methanol abundance in Core 2 found here is higher than observed in massive and low-mass dark clouds (e.g. Friberg et al., 1988; Kalenskii & Sobolev, 1994; Dickens et al., 2000; Sanhueza et al., 2013; Vasyunina et al., 2014; Punanova et al., 2018), but the upper limit of the methanol abundance is 10–100 times lower, or, in rare cases, consistent with the lower limits found in hot cores or ucH ii regions (e.g. Schöier et al., 2002; Sutton et al., 2004; Garrod et al., 2008; Purcell et al., 2009; Sánchez-Monge et al., 2014; Mendoza et al., 2018; Gorai et al., 2020). As found by Remijan et al. (2004); Kalenskii et al. (2000); Sánchez-Monge et al. (2014); Hernández-Hernández et al. (2014); Mendoza et al. (2018); Taniguchi et al. (2020), the typical relative abundance of CH₃CN in hot cores is $10^{-9}-10^{-7}$, which is also much higher than the abundance obtained with the detected $k\leq 3$ transitions in Sec. 3.2. The study of Kalenskii et al. (2000) also contains a list of high-mass YSOs with low CH₃CN abundances. These sources might contain hot regions accompanied by colder gas, exactly as we see in Core 2. Fayolle et al. (2015) studied poorly explored organic-poor massive young stellar objects with weak hot organic molecular lines and found that complex organic molecules might be released from dust grains not only at the hot core stage, but earlier, depending on ice composition and temperature.

The CH₃OH and CH₃CN molecules are considered to be parent and daughter species in chemical models of gas-grain chemistry. This is why the observations of these lines show the ratio of relative abundances $[\rm CH_{3}CN/CH_{3}OH]\leq 1$. Ospina-Zamudio et al. (2018) combined measurements of the $[\rm CH_{3}CN/CH_{3}OH]$ ratio in protostars of various masses and found values between 0.002–0.006 in low- and intermediate-mass regions. However, in high-mass regions, this ratio shows a range of values of up to two orders of magnitude: from 0.25 in G29.96 and 0.1 in Orion-KL (Beuther et al., 2009) and Sgr B2 (Belloche et al., 2013), to $0.01-0.04$ in several hot cores (Bisschop et al., 2007; Mendoza et al., 2018), and 0.0025–0.01 in the massive young protostellar object G345.4938+01.4677 (Cores C and B, respectively, Taniguchi et al., 2020). Our value is close to the last value detected in G331 Core B (hot core), which is more evolved than Core C with a lower $[\rm CH_{3}CN/CH_{3}OH]$ ratio. Therefore, the observed $[\rm CH_{3}CN/CH_{3}OH]$ ratio agrees with our conclusion about the warm up and beginning of the hot-core stage in the Core 2.

Due to low $E_{\rm u}$ values (see Table 2), C³⁴S($5-4$) and SO₂($5_{2,4}-4_{1,3}$) lines are not straightforward signposts of a hot core stage. The detection of lines of such molecules as CS and SO₂ suggests that the cold phase of protostellar evolution has completed. During the warm up phase, H₂S molecules evaporate from grain mantles and give rise to gas-phase sulfur chemistry, producing SO, SO₂, CS, etc. (see, e.g. chemical modelling by Charnley, 1997). However, the abundance of SO₂ found in this work is up to several orders lower than the typical value found at the hot core stage as shown in the review of several hot cores by Schöier et al. (2002) and Wakelam et al. (2004). The value of $x_{\rm CS}\approx 10^{-10}-10^{-9}$ obtained in this study is typical for the stage of freeze-out of molecules onto dust grains (e.g. Schöier et al., 2002). Beuther et al. (2002) studied several dozen high-mass protostellar candidates and found an average abundance of $x_{\rm CS}\approx 8\times 10^{-9}$, which is about an order of magnitude higher than our value in Core 2. More-evolved stages of ucH ii regions show similar abundances (Zinchenko et al., 2009, e. g.). Therefore, the cold stage seems to have finished quite recently, and the abundances are still quite low. We note again that our CS and SO₂ abundances are obtained using the same filling factor as we found from the analysis of the methanol emission. If the emission of SO₂ and CS arise in less extended regions as compared to methanol, the SO₂ and CS filling factors should be smaller; therefore, the abundances will be higher.

Bearing in mind the studies mentioned above, and our own results, we state that the Core 2 cannot be considered as an example of a high-mass analogue of a pre-stellar Class 0 object. We find warm-up chemistry and signatures of the beginning of the hot core stage there.

We report the detection of methanol lines towards massive Core 1 and Core 2, belonging to the neutral envelope of the H ii region RCW 120, and also the detection of CH₃CN, CS and SO₂ lines towards Core 2. We estimate the gas physical parameters using the methanol lines and obtain gas temperature less than 100 K in both regions. The value of $n_{\rm H_{2}}$ is in the range of $10^{5}-10^{7}$ cm^-3 in Core 2, and has a higher uncertainty in Core 1. However, the detection of high-excitation CH₃CN lines in Core 2 indicates that the region might contain hot gas. The relative abundances of CH₃OH, CS, SO₂ and CH₃CN found in the present study are quite low for a hot-core stage, but agree with a model of warm chemistry in dense gas. We suggest that Core 2 is in the beginning of the hot core stage.

We are thankful to the staff of the Onsala Space Observatory for their care for technical details of the O-083.F-9311A-2009 project, to Maxim Voronkov for discussions of the methanol maser activity and the referee for comments and suggestions which helped us to clarify our statements and improve the presentation. Astrochemical modelling performed by D. S. Wiebe was supported by the RFBR grant 20-02-00643. M. S. Kirsanova and P. A. Boley were supported by RSF, research project 18-72-10132. S. V. Salii and A. M. Sobolev work was supported by the Ministry of Education and Science of the Russian Federation within the framework of the research activities (project no. FEUZ-2020-0030), as well as Decree 211 of the Government of Russian Federation (contract no. 02.A03.21.0006). This research has made use of NASA’s Astrophysics Data System Bibliographic Services; SIMBAD database, operated at CDS, Strasbourg, France (Wenger et al., 2000), Astropy,⁴⁴4http://www.astropy.org a community-developed core Python package for Astronomy (Astropy Collaboration et al., 2013; Price-Whelan et al., 2018), Matplotlib (Hunter, 2007) and APLpy, an open-source plotting package for Python Robitaille & Bressert (2012).

The data underlying this article will be shared on reasonable request to the corresponding author.