© The Authors 2023

1 Introduction

The formation of a high-mass star begins with the fragmentation of a massive clump into smaller structures known as molecular cores. However, what is not clear, is whether this fragmentation gives rise to prestellar cores that are sufficiently massive (a few tens of solar masses) to directly form these stars or leads to cores of low and intermediate masses that generate highmass stars, acquiring material from their environment (Palau et al. 2018; Moscadelli et al. 2021). In the first scenario, highmass stars form through an individual monolithic core collapse, whereas in the second one, they form from a global hierarchical collapse of a massive clump, where many low- and intermediate-mass cores competitively accrete material from the surrounding through converging gas filaments that feed the cores (Motte et al. 2018; Schwörer et al. 2019). These outlines of the high-mass star formation scenarios overlook several aspects of their chemical and physical complexity, both of which are treated in detail in the reviews by Krumholz & Bonnell (2009), Tan et al. (2014), and Vázquez-Semadeni et al. (2019).

Nowadays, an important line of research in the field of highmass star formation focuses on the fragmentation of massive clumps in their earliest stages. The main goal is to detect the presence of massive prestellar cores, but whether or not they exist is still a matter of debate; and if they do, whether or not they are stable enough against further fragmentation to give rise to the formation of a high-mass star through monolithic collapse is still unclear.

Several studies based on data from the Atacama Large Millimeter Array (ALMA) for infrared-quiet massive clumps have revealed different fragmentation properties: in some cases revealing limited fragmentation (very few cores and with superJeans masses well above the solar mass) with a large fraction of prestellar cores with masses in the range 8–120 M_⊙ (e.g. Wang et al. 2014; Csengeri et al. 2017a; Neupane et al. 2020), and in other cases a large population of low-mass (≤1 M_⊙) prestellar cores are seen with a maximum core mass of 11 M_⊙ (Sanhueza et al. 2019). In this regard, Kainulainen et al. (2013) pointed out that a possible explanation for the different fragmentation characteristics could be the size-scale-dependent collapse time scale that results from the finite size of real molecular clouds, which is indeed predicted by analytical models (Pon et al. 2011).

Csengeri et al. (2017a) carried out a fragmentation study towards a sample of ATLASGAL sources identified as infrared-quiet massive clumps using continuum ALMA data at a spatial resolution of about 0.06 pc, finding limited fragmentation towards most of the sources. According to the authors, a possible explanation could be that early fragmentation of massive clumps does not follow thermal processes, which leads to fragment masses largely exceeding the local Jeans mass. Thus, a combination of turbulence, magnetic field, and radiative feedback would be increasing the necessary mass for fragmentation. Another explanation could be that these early stages could correspond to a phase of compactness where the large level of fragmentation to form a cluster has not yet developed.

Among the sources characterised by Csengeri et al. (2017a) is AGAL G035.1330–00.745 (hereafter AGAL35), towards which the authors identified two molecular cores. Assuming an average clump temperature of 25 K, they derived masses of about 36 and 8 M_⊙.

However, Ortega et al. (2022), based on continuum and line data provided by ALMA, with a spatial resolution of about 0.007 pc, identified four molecular cores towards AGAL35. In addition, the authors estimated masses below 2 M_⊙ for the cores using core temperatures above 100 K derived from the CH₃CN J = 13−12 transition. The authors also found molecular outflow activity towards two molecular cores. They concluded that considering an average clump temperature for the estimation of the masses of the cores could be inadequate, even in the case of infrared-quiet massive clumps, which could lead to an overestimation of the masses. This latter study confirms that a prestellar clump candidate can present star formation activity manifesting as hot cores and/or molecular outflows when studied at the core scale. Paraphrasing Pillai et al. (2019), we wonder whether high extinction can hide very young and low-luminosity protostars within such seemingly starless clumps, which would mean that the existing cases of high-mass starless cores may not be starless in reality.

Finding massive prestellar cores stable against further fragmentation would support the monolithic collapse scenario, but this is not an easy task. On the other hand, it is equally important to carry out detailed characterisations of massive clumps with recent star formation activity, such as the hot molecular cores (HMCs), in order to obtain a more complete picture of how the fragmentation occurs. Additionally, HMCs are the most chemically rich regions in the interstellar medium (ISM; e.g. Herbst & van Dishoeck 2009; Bonfand et al. 2019), and the star forming processes strongly influence the chemistry of their surrounding environment (Jørgensen et al. 2020). Therefore, observing molecular lines and studying their emission and chemistry is important for characterising the physical and chemical conditions of the gas in view of understanding the evolutive stage of the fragmentation.

Nowadays, there are few works in the literature that connect high-mass star formation signatures at the clump scale with evidence of high-mass star formation at the core scale. A good candidate for use in this kind of study would be a massive molecular clump harbouring an extended green object (EGO).

Cyganowski et al. (2008) catalogued more than 300 EGOs based on their extended 4.5 µm emission in GLIMPSE images. EGOs are defined as massive young stellar objects (MYSOs) and are candidates to harbour molecular outflows. Therefore, the presence of a bright EGO embedded in a massive clump suggests that high-mass star formation is taking place. Keeping this in mind, we searched for ALMA observations¹ towards the brightest EGOs in the catalogue with the additional requirement that these EGOs be associated with a high-mass ATLASGAL source. Furthermore, it is essential for this study that the ALMA data include molecules from which accurate temperature values can be obtained for the cores. Based on these criteria, we selected EGO 338.92+0.55(b) (hereafter EGOG338) embedded in the massive clump AGAL G338.9188+0.5494, which is presented in detail in the following section.

2 Presentation of the region

The submillimetre source AGAL G338.926+00.554 (Contreras et al. 2013) is located towards the eastern border of the H II region G338.90+00.60 (see Fig. 1 left-panel). Wienen et al. (2015), based on ¹³CO J = 1−0 emission, estimated a systemic velocity of about −64.1 km s⁻¹ for this ATLASGAL source, which corresponds to a near kinematic distance of about 4.4 kpc.

Csengeri et al. (2014) extended the Contreras et al. (2013) ATLASGAL catalog of compact sources using an optimised source-extraction method. The authors identified a total of 10861 compact submillimetre clumps, significantly increasing the number of previously detected sources. In particular, we found out that AGAL G338.926+00.554 is composed of four minor dust condensations. Among them, AGAL G338.9188+0.5494 (hereafter AGAL 338; see yellow star in Fig. 1, left panel) is the dust condensation associated with EGO G338.

The right panel of Fig. 1 shows a close-up view of the location of EGO G338 at the same mid-infrared bands. The black contours represent the ALMA submillimetre continuum emission at 340 GHz (in the 7 m array) with an angular resolution of about 4″ (see Sect. 3). There is a conspicuous dust condensation in positional coincidence with the peak of emission at the mid-infrared bands associated with EGOG338.

3 Data

The data cubes from the projects 2015.1.01312 (PI: Fuller, G.; Band 6) and 2017.1.00914 (PI: Csengeri, T.; Band 7) were obtained from the ALMA Science Archive². The single pointing observations for the target were carried out using the following telescope configurations with L5BL/L80BL(m): 42.6/221.3 for project 2015.1.01312 and 34.5/226.8 for project 2017.1.00914, in the 12 m array in both cases. Table 1 shows the main ALMA data parameters.

Project 2017.1.00914 also includes observations of Band 7 in the 7 m array with angular resolution and continuum sensitivity of about 3″.7 and 1.2 mJy beam⁻¹, respectively, and a telescope configuration with L5BL/L80BL(m):8.7/27.5. We only used the continuum at 340 GHz in the 7 m array to introduce the region (see Fig. 1, right panel). When we refer to 340 GHz continuum in the present paper, we are referring to the 12m array.

We extracted all the molecular lines from Band 6, except the ¹²CO J = 3–2 transition, which, together with the continuum at 340 GHz, was obtained from Band 7. It is important to note that, even though the data of both projects passed the QA2 quality level, which assures a reliable calibration for ‘science ready’ data, the automatic pipeline imaging process may give rise to a clean image with some artefacts. For example, an inappropriate setting of the parameters of the clean task in CASA could generate artificial dips in the spectra. Thus, we reprocessed the raw data using CASA 4.5.1 and 4.7.2 versions and the calibration pipeline scripts. Particular care was taken with the different parameters of the task clean. The images and spectra obtained from our data reprocessing – after several runs of the clean task, varying some of its parameters – were very similar to those obtained from the archival data.

The task imcontsub in CASA was used to subtract the continuum from the spectral lines using a first-order polynomial. The frequency ranges without molecular line emission were carefully selected in each spectral window. The continuum map at 340 GHz in the 12 m array was obtained by averaging the continuum emission of each of the four spectral windows and was corrected for primary beam. Several continuum subtraction tests were performed to ensure a reliable 340 GHz continuum map, which involved the selection of different free line regions of the spectra. This map has an rms noise level of about 0.2 mJy beam⁻¹.

Given that high-spatial resolution is required to properly characterise the clump fragmentation and star formation activity at core scales, it is important to note that the beam size of the 340 GHz continuum data in the 12 m array provides a spatial resolution of about 0.01 pc (~2000 au) at the distance of 4.4 kpc, which is appropriate to spatially resolve the substructure of the clump AGAL 338.

Fig. 1 Large-scale surroundings of EGO G338.92+0.55(b). Left-panel: Overview of the H II region G338.90+00.60 at Spitzer 3.6 (blue), 4.5 (green), and 8.0 (red) µm bands. The white contours represent the continuum emission at 870 µm extracted from the ATLASGAL survey. Levels are: 1, 3, 6, 9, and 12 Jy beam−1. The green square highlights the region studied in this work, which includes the position of EGO G338.92+0.55(b) represented by the yellow star. Right-panel: A close-up view of the EGO at the same Spitzer bands. The black contours represent the ALMA continuum emission at 340 GHz (7 m array). Levels are: 0.2, 0.4, 0.7, and 1.4 Jy beam−1. The beam of the ALMA observation is indicated in the bottom-right corner. The blue, green, and red colour scales go from 10 to 500 MJy sr−1.

4 Results

In the following subsections, we present studies of fragmentation and star formation at core scales towards AGAL 338 using the ALMA data at bands 6 and 7 in the 12 m array.

4.1 Continuum emission: tracing the fragmentation

We begin the study of the fragmentation of the dust clump AGAL 338 by analysing the high-resolution and high-sensitivity sub-millimetre continuum emission map at Band 7 (array 12 m).

Figure 2 shows the ALMA continuum emission at 340 GHz in greyscale and blue contours. The green contours represent the ALMA continuum emission at 340 GHz (7 m array) presented in Fig. 1, in which we see a conspicuous dust condensation, labelled MM1 in Fig. 2, a faint tail-like feature towards the southwest direction, and a lobe-like feature aligned with the extended emission at 4.5 µm in the southeast–northwest direction but oriented in the opposite direction.

The better angular resolution of the 12 m array observations allows us to identify five dust cores towards AGAL 338, which are labelled C1 to C5. In particular, we can resolve the MM1 condensation in four cores (C1 to C4), while core C5 lies over the faint tail-like structure as seen in the continuum emission of the 7 m array. Faint emission can also be seen in positional coincidence with the lobe-like structure extending towards the northwest.

Table 2 presents the main parameters of the dust continuum cores observed at 340 GHz. Columns 2 and 3 give the absolute position, Col. 4 the angular size, and Cols. 5 and 6 show the peak intensity I_peak and the integrated intensity S, respectively. The core sizes are at least a factor of six smaller than the maximum recoverable scales of the observations, which ensures that all of the flux of the cores is recovered.

Fig. 2 ALMA continuum emission at 340 GHz (12 m array). The greyscale goes from 1 to 180 mJy beam−1. The blue contour levels are: 1 (about 5 σ), 10, 30, 60, 90, and 140 mJy beam−1. The dashed red contours correspond to −1 mJy beam−1. The beam of 340 GHz continuum emission (12 m array) is indicated in the bottom left corner. The green contours correspond to the ALMA continuum emission at 340 GHz (7 m array) shown in Fig. 1, with levels at 0.2, 0.4, 0.7, and 1.4 Jy beam−1.

Fig. 3 12CO J = 3−2 emission distribution integrated between −90 and −70km s−1 (blue), and between −55 and +5 km s−1 (red). The systemic velocity of the complex is about −64 km s−1. The contour levels are at 5, 8, 12, and 18 mJy beam−1. The beam of the 340 GHz continuum and the 12CO J = 3−2 emission is the same and is indicated in the bottom left corner.

4.2 ¹²CO and C¹⁷0: Tracing the outflow activity and the ambient gas

We used the ¹²CO emission to search for molecular outflow activity related to the clump AGAL 338 and, in particular, associated with EGO 338. After careful inspection of the channels of the spectral window containing the ¹²CO J = 3−2 transition, we found several extended structures that suggest the presence of molecular outflows related to some of the dust cores.

Figure 3 shows the ¹²CO J = 3−2 emission distribution integrated between −90 and −70 km s⁻¹ (blue), and between −55 and +5 km s⁻¹ (red). The systemic velocity of the complex is about −64 km s⁻¹ (Wienen et al. 2015).

At first glance, we see intense molecular outflow activity arising from the central region of the core cluster. The brightest core, C1, exhibits the most conspicuous molecular outflow, which is oriented in the southeast-northwest direction. The position of the red-OC1 outflow coincides with the lobe-like structure extending towards the northwest mentioned in Sect. 4.1, while the blue-OC1 outflow, which is less collimated, spatially coincides with the 4.5 µm extended emission of EGO G338 (see Fig. 1, right panel). The core C2 shows a more collimated outflow, referred to here as red-OC2, which can be seen in the northeast-southwest direction. From the contours displayed in Fig. 3, we can see that the outflow red-OC1 is more clumpy than red-OC2, which is the most collimated outflow in the region. While both red outflows (red-OC1 and red-OC2) exhibit a clear spatial separation, the blue ones, blue-OC1 and blue-OC2, are blended and appear as a single cone-like shape structure that opens towards the south.

Towards the core C4, a faint red outflow appears extending southwards, which is likely associated with a weak and/or incipient outflow activity, while its blue counterpart seems to be contaminated by the blue-OC2 outflow. A complete characterisation of the cores embedded in AGAL 338 must include a study of the associated molecular outflow activity. Therefore, following Li et al. (2018), we estimated the main parameters of the outflows red-OC1 and red-OC2. The blue counterparts, blue-OC1 and blue-OC2, are part of a single structure.

We calculated the column density and the total mass for each red lobe using the following equations (Buckle et al. 2010):

$$\text{N}\left({}^{12}\text{CO}\right)=7.96\times {10}^{13}\left(\frac{T_{\text{ex}}+0.92}{1-\exp \left(\frac{-16.6}{T_{\text{ex}}}\right)}\right)\exp \left(\frac{16.6}{T_{\text{ex}}}\right){\displaystyle \int {\tau }_{32}\text{d}\upsilon \text{,}}$$(1)

and assuming that the ¹²CO J = 3−2 emission is optically thin towards the outflows (e.g. Lebrón et al. 2006; Shimoikura et al. 2015),

$${\displaystyle \int {\tau }_{32}\text{d}\upsilon \text{=}\frac{1}{J\left(T_{\text{ex}}\right)-J\left(2.7\text{K}\right)}}{\displaystyle \int T_{\text{mb}}\text{d}v,}$$(2)

with

$$J\left(T\right)=\left(\frac{hv/k}{\exp \left(\frac{hv}{kT}\right)-1}\right),$$(3)

$$M_{\text{out}}=\text{N}\left({}^{12}\text{CO}\right)\left[{\text{H}}_2/\text{CO}\right]{\mu }_{{\text{H}}_2}{m}_{\text{H}}{A}_{\text{pixel}}{N}_{\text{pixel}},$$(4)

where N(¹²CO) is the average column density of each lobe, [CO/H₂] = 10⁻⁴ is the abundance ratio between the molecules, ${\mu }_{\text{H2}}=2.72$ is the mean molecular weight, m_H = 1.67 × 10⁻²⁴ g is the mass of the hydrogen atom, A_pixel is the pixel area, and N_pixel is the number of pixels filling each lobe. The outflows parameters were estimated using typical excitation temperatures ranging from 10 to 50 K (e.g. Li et al. 2020).

Table 3 shows the main parameters derived for outflows red-OC1 and red-OC2: mass, momentum ($P=M\overline{\upsilon }$), energy, outflow mechanical force (F_out = P/t_dyn), length, and dynamical age (t_dyn = Length/υ_max), where $\overline{\upsilon }$ and υ_max are the median and maximum velocity of each velocity interval with respect to the systemic velocity of the gas associated with AGAL 338.

Additionally, we find that AGAL 338 shows near-IR emission at K_s band associated with EGO G338. Figure 4 presents the K_s-band emission obtained from the VISTA Hemisphere Survey (VHS; McMahon et al. 2013). The green contours represent the continuum emission at 340 GHz. The bulk of emission coincides with the location of cores C1, C2, and C3. Conspicuous near-IR emission extends towards the southeast direction, which overlaps with the position of the outflow blue-OC1 (see Fig. 3). However, there is no evidence of extended emission at near-IR related to the red-OC1 outflow.

The top panel of Fig. 5 shows the C¹⁷O J = 2−1 moment 0 map integrated between −66 and −61 km s⁻¹. The contours represent the continuum emission at 340 GHz. The bulk of emission coincides with the location of core C1, and is extended northwards, where another C¹⁷O condensation appears in coincidence with a protrusion of the continuum emission (probably an incipient, or another unresolved core). Also, a curved filament (indicated in Fig. 5) emerges further northwards. Additionally, two elongated structures, labelled S1 and S2, can be seen connected with the bulk of the emission, which extend from east to west.

The bottom panel of Fig. 5 shows the C¹⁷O J = 2−1 moment 1 map integrated in the same velocity interval as the moment 0 map. The C¹⁷O J = 2−1 associated with the core C1 is at the systemic velocity, while the nearby gas towards the north and the south shows red- and blue-shifted velocities, respectively, which could be tracing the birth of the molecular outflows. The curved filament does not show any velocity gradient, and therefore we discard the possibility that it could be a typical converging filament feeding the cores (Schwörer et al. 2019; Pineda et al. 2023).

Fig. 4 Ks band emission from VISTA hemisphere survey with VIRCAM. The green contours represent the continuum emission at 340 GHz (12 m array). Levels are at 1, 10, 30, 60, 90, and 140 mJy beam−1. The beam of the continuum emission at 340 GHz is indicated in the bottom left corner.

Fig. 5 C17O J = 2−1 moment maps. Top panel: C17O J = 2−1 moment 0 map integrated between −66 and −61 km s−1. The colour scale goes from 0.01 to 0.50 Jy beam−1. The black contours represent the continuum emission at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The beams of the C17O J = 2−1 line and the 340 GHz continuum emission are indicated in the bottom right corner. Bottom panel: C17O J = 2−1 moment 1 map. The units of the colour bar are km s−1. The black contours represent the radio continuum at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The systemic velocity is about −64 km s−1.

4.3 Analysis of molecular species

We analysed several molecular lines that could be useful in characterising the physical conditions and the chemistry of the hot cores. The selected molecular lines are presented in Table 4. Figure 6 presents the moment 0 maps of these molecular lines, showing the spatial distribution of the emission of each molecular species towards the region of EGO G338 in comparison with the dust millimetre continuum emission (displayed in green contours).

4.3.1 CH₃CN and CH₃CCH: Temperatures and column densities of the core C1

Methyl cyanide (CH₃CN) and methyl acetylene (CH₃CCH) have been shown to be reliable tracers of physical conditions, such as temperature and density, and have been extensively studied towards several hot molecular cores (e.g. Remijan et al. 2004; Calcutt et al. 2019; Brouillet et al. 2022; Ortega et al. 2022, and reference therein). Their rotational transitions are characterised by two quantum numbers, namely the total angular momentum (J) and its projection on the principal symmetry axis (K). As these molecules are top-symmetric rotors, they present many K projections that are closely spaced in frequency, which favours their observation. Moreover, in both molecules, transitions with ΔK # 0 are forbidden. Therefore, the relative populations of different K-ladders are dictated only by collisions, and as a result, CH₃CN and CH₃CCH act as excellent temperature probes.

In particular, given the small electric dipole moment of the CH₃CCH molecule (µ = 0.78 D), line thermalisation occurs at densities as low as about 10⁴ cm⁻³ (e.g. Molinari et al. 2016). Figures 6a,b shows the moment 0 maps of the CH₃CN J = 13−12 and CH₃CCH J = 14−13 transitions, respectively, both at K = 2 projection. The green contours represent the ALMA continuum emission at 340 GHz. The spatial distribution of both molecules overlaps with the positions of cores C1 to C4, but while the CH₃CN peak positionally coincides with core C1, the emission peak of CH₃CCH appears shifted by about one beam size relative to this core. In particular, the methylacetylene emission exhibits an arc-like structure towards the northeast in positional coincidence with the faint extended emission of the continuum at 340 GHz.

Figure 7 shows the CH₃CN J = 13−12 (top panel) and CH₃CCH J = 14−13 (bottom panel) spectra towards the core C1. It is important to mention that for the other cores, the CH₃CN and CH₃CCH spectra have no K projections above the 5σ rms level from K = 3, and so the temperature estimate is restricted to core C1.

Table 5 shows the tabulated parameters for all K projections of CH₃CN J = 13−12 and CH₃CCH J = 14−13 transitions detected towards core C1. Columns 1 and 2 show the K projection and the rest frequency, respectively, obtained from the NIST catalogue³. Column 3 presents the upper energy level (E_u/k) extracted from the LAMDA database⁴, and Col. 4 shows the line strength of the projection multiplied by the dipole moment of the molecule (S_ulµ²).

Table 6 shows the main parameters derived from the Gaussian fittings to the CH₃CN and CH₃CCH spectra towards core C1. Columns 2, 3, 4, and 5 show the peak intensity, the central velocity (υ_c), the FWHM (Δυ), and the integrated intensity (W), respectively. The integrated intensities were used to construct the rotational diagram (RD) presented in Fig. 8. Therefore, using the RD analysis (Goldsmith & Langer 1999, and references therein) and assuming LTE conditions, optically thin lines, and a beam filling factor equal to unity, we can estimate the rotational temperatures (T_rot) and the column densities for core C1 using both molecules. This analysis is based on a derivation of the Boltzmann equation,

$$\ln \left(\frac{N_u}{g_u}\right)=\ln \left(\frac{N_{\text{tot}}}{Q_{\text{rot}}}\right)-\frac{E_u}{k{T}_{\text{rot}}},$$(5)

where N_u represents the molecular column density of the upper level of the transition, g_u is the total degeneracy of the upper level, N_tot the total column density of the molecule, Q_rot the rotational partition function, and k is the Boltzmann constant.

Following Miao et al. (1995), for interferometric observations, the left-hand side of Eq. (5) can be estimated from:

$$\ln \left(\frac{N_{\text{u}}^{\text{obs}}}{g_{\text{u}}}\right)=\ln \left(\frac{2.04\times {10}^{20}}{{\theta }_{\text{a}}{\theta }_{\text{b}}}\frac{W}{g_{\text{k}}{g}_1{v}_0{}^3{S}_{\text{ul}}{\mu }_0{}^2}\right),$$(6)

where $N_{\text{u}}^{\text{obs}}$ (in cm⁻²) is the observed column density of the molecule under the above-mentioned conditions, θ_a and θ_b (in arcsec) are the major and minor axes of the clean beam, respectively, W (in Jy beam⁻¹ km s⁻¹) is the integrated intensity of each K projection, g_k is the K-ladder degeneracy, g_l is the degeneracy due to the nuclear spin, v₀ (in GHz) is the rest frequency of the transition, S_ul is the line strength of the transition, and µ₀ (in Debye) is the permanent dipole moment of the molecule. The free parameters, N_tot/Q_rot and T_rot were determined by a linear fitting to Eq. (5) (see Fig. 8). Finally, using the tabulated value for Q_rot at the corresponding temperature extracted from the CDMS database⁵, we obtain the CH₃CN and CH₃CCH column densities for core C1 (see Table 7). The CH₃CN K = 5 to K = 7 projections are blended with some CH₃¹³CN isotopologue projections (see Fig. 7, top panel). Therefore, in such cases, two Gaussian components were fitted.

The CH₃CN K = 7 and K = 8 components show a central velocity shift of about 1 km s⁻¹ with respect to the systemic velocity, which suggests possible contamination of other lines. In fact, the rotational diagram with the measured W for these K projections yields a temperature above 500 K, which seems to too high for the gas traced by this molecular species. We therefore looked for potential contamination lines on the Splatalogue platform (JPL and CDMS databases). We did not find any obvious contamination line beyond several rare complex molecules with very weak intensities. However, it is likely that, in such line-rich spectra (see Appendix A), there could still be unidentified lines, probably from vibrational or torsional states of known molecules. Therefore, assuming that 50% of the component area for K = 7 and K = 8 comes from contamination of unidentified lines, we use half of the integrated intensity values for these projections in order to build the rotational diagram.

The optical depth effect, which could result in an underestimation of the intensity of a line, tends to be more noticeable in lower projections. This would produce a flattening of the slope in an RD graphic, leading to anomalously large values for T_rot. The method proposed by Goldsmith & Langer (1999) iteratively corrects individual N_u/g_u values by multiplying by the optical depth correction factor, C_τ = τ/(1 – e^−τ). However, we find that the τ corresponding to K = 0 projection is lower than 0.06 for both molecules, which leads to a correction factor of less than 3%, and therefore the rotation temperature would not be overestimated.

Fig. 6 Integrated emission maps (moment 0) of the selected molecular lines (indicated in each panel and in Table 4). Dashed magenta contours represent the molecular line emission at the following levels (in Jy beam−1 km s−1): 0.1, 0.3, 1, and 4 (CH3CN), 0.05, 0.10, 0.20, 0.50, and 0.65 (CH3CCH), 0.04, 0.1, 0.25, 0.6, 1.3, and 3 (HC3N), 0.05, 0.2, 0.5, 1.2, and 3 (H2CS), 0.1, 0.3, 0.6, 1.5, and 5 (CH3OH), 0.05, 0.2, 1, and 3 (HNCO), 0.02, 0.04, 0.08, and 0.16 (CN), 0.07, 0.12, 0.2, 0.4, 0.6, 1.2, 2, and 3 (C34S), and 0.5, 1, and 2 (HDO). The integration velocity interval ranges from −67 to −61 km s−1 for all molecular lines, except for H2CS and CH3OH (−67 to −62 km s−1) and for HNCO (−67 to −63 km s−1). The green contours represent the continuum emission at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The beams of 340 GHz continuum and line emissions are indicated in the bottom left corner.

Fig. 7 Average spectra of the temperature tracer molecules taken over the full size of core C1. Top panel: CH3CN J = 13−12. Bottom panel: CH3CCH J = 14−13. The K projections are indicated.

Fig. 8 Rotational diagrams for the core C1. Top panel: CH3CN J = 13−12. Bottom panel: CH3CCH J = 14−13. The red lines show the best linear fitting of the data. The magenta squares in the top panel correspond to CH3CN J = 13−12 K = 7 and K = 8 projections, for which half of the integrated intensity values were considered.

4.3.2 Mass and kinematics of the core C1

We estimate the mass of the core C1 from the continuum emission at 340 GHz using each rotational temperature derived in the previous section.

Considering that, at the early evolutionary stage of AGAL 338, the contribution of free-free continuum emission at 340 GHz is negligible (e.g. Isequilla et al. 2021), it is reasonable to assume that, at this frequency, the submillimetre continuum is mainly tracing the dust emission. The mass of the gas of the core C1 was then estimated from the dust continuum emission at 340 GHz (λ ~ 0.9 mm) following Kauffmann et al. (2008),

$$\begin{array}{l}{M}_{\text{gas}}=0.12M_{\odot }\left[\exp \left(\frac{1.439}{\left(\lambda /\text{mm}\right)\left(T_{\text{dust}}/10\text{K}\right)}\right)-1\right]\hfill \\ \times {\left(\frac{{\kappa }_v}{0.01{\text{cm}}^2{\text{g}}^{-1}}\right)}^{-1}\left(\frac{{\text{S}}_v}{\text{Jy}}\right){\left(\frac{\text{d}}{100\text{\hspace{0.17em}pc}}\right)}^2{\left(\frac{\lambda }{\text{mm}}\right)}^3,\hfill \end{array}$$(7)

where T_dust is the dust temperature and κ_v is the dust opacity per gram of matter at 870 µm, for which we adopt the value of 0.0185 cm² g⁻¹ (Csengeri et al. 2017b, and references therein). We assume thermal coupling between dust and gas (T_dust=T_kin), where T_kin=T_rot.

In any case, we use a T_kin ranging from a typical desorption temperature in hot cores of about 120 K (e.g. Busch et al. 2022) to the temperature estimated from the rotational diagram of the CH₃CN (see Sect. 4.3) in order to obtain a core mass ranging from 3 to 10 M_⊙. Despite the fact that estimating mass for cores based on dust emission is the most reliable method, there are some sources of uncertainty. Considering an absolute flux uncertainty of ≤10% for ALMA observations in band 7, a dust temperature uncertainty of about 20% and a distance uncertainty of ~10%, the mass uncertainty would be about 50%.

Figure 9 shows the CH₃CN J = 13−12 moment 1 map for the K = 4 projection integrated between −66 and −62 km s⁻¹. It can be appreciated that, at this K-projection, the emission is concentrated towards the core C1. The gas related to this core exhibits a clear velocity gradient perpendicular to the molecular outflow direction. This velocity gradient has been interpreted in several works as evidence of a rotating disc (e.g. Louvet et al. 2016; Ortega et al. 2022). It is important to mention that this signature of disc rotation has only been found in this molecular species.

Fig. 9 CH3CN J = 13–12 moment 1 map (from K = 4 projection). The black contours represent the continuum emission at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The dashed black line indicates the direction of the molecular outflow related to core C1. The beams are indicated in the bottom left corner.

5 Discussion

In this section, we discuss the implications of our findings and their potential impact on our understanding of high-mass star formation. We also compare our results with previous studies and highlight the most significant differences and similarities. Finally, we discuss the limitations of our study and suggest avenues for future research.

5.1 A massive clump fragmented into low-intermediate mass cores?

Massive clumps usually have a relatively low thermal Jeans mass, which predicts a high level of fragmentation. Csengeri et al. (2014) estimated an integrated flux of about 7.36 Jy at 870 µm for AGAL 338. Assuming a typical temperature of 20 K, and considering a radius of the clump of 0.3 pc, we derive a clump mass of ~1260 M_⊙ and a Jeans mass of ~4 M_⊙, which suggest that AGAL 338 would be unstable to fragmentation. Specifically, based on the dust continuum emission at 340 GHz, we find that the fragmentation of AGAL 338 gave rise to at least five molecular cores, labelled C1 to C5.

Although the CH₃CCH and CH₃CN molecules are detected towards most of the cores, it is only possible to estimate temper- atures for core C1. Using the CH₃CCH and CH₃CN molecules, we derived temperatures of about 72 and 340 K, respectively. In particular, a temperature value of 340 K is among the highest temperature values found by Hernández-Hernández et al. (2014) towards the compact component of several hot cores. The detec- tion above 5cr noise level of the CH₃CN J = 13–12 K=8 projection (E_u = 537 K), the presence of several CH₃CN υ₈ = 1 lines (see Fig. A.l), and the richness of the spectra in the four spectral windows suggests that a high-temperature gas component is present in core C1. However, because of the large number of lines in the spectra, and the contamination in the lines that this entails, it is very difficult to accurately estimate this high temperature value.

The discrepancy in the temperatures derived from CH₃CCH and CH₃CN may be an indication that each molecule is tracing different gas layers associated with the hot core. Andron et al. (2018) predicted that the desorption of the CH₃CN molecule from the dust grains occurs at radii closer to the protostar (and therefore at higher temperatures) than the desorption of the CH₃CCH molecule. This would explain the higher temperature value found for the core C1 from the methyl cyanide molecule.

Using a range of temperatures going from 120 K (about the typical molecular desorption temperature in hot cores) to 340 K obtained from the rotational diagram of the CH₃CN, the mass of core C1 ranges from 3 to 10 M_⊙. The mass of such a core, the brightest and most active core embedded in AGAL 338, is significantly below the limit for a massive core (a few tens of solar masses) candidate to form high-mass stars in a scenario of monolithic collapse. In such a scenario, following Duarte-Cabral et al. (2013), who indicate that the efficiency with which the core mass is converted to stellar mass is about 50%, core C1 would give rise to a low-mass star.

It is not unreasonable to assume that fragmentation of the molecular clump AGAL 338 seems to have produced low- and/or intermediate-mass cores. Therefore, the only path for the forma- tion of massive stars in this region should be cores acquiring mass through gas infalling from their parent structures. In other words, a competitive accretion scenario. However, although we have searched for signatures of converging gas filaments through a kinematic analysis of the gas in all molecules, we did not find any evidence for streams of gas feeding the cores.

5.2 EGO G338 and the core molecular outflow activity

We discuss the molecular outflow activity associated with the cores C1 and C2 in relation to the presence of EGO G338, one of the brightest in the Cyganowski et al. (2008) catalogue, tak- ing into account that an EGO is a MYSO candidate to produce molecular outflows.

In Sect. 4.2, we characterise the outflow activity related to the cores C1 and C2. As shown in Fig. 3, while the red lobes associated with the cores C1 and C2 are spatially separated and relatively well collimated, the inner region between the blue lobes shows extended emission that connects them with a cone- like shape structure that opens towards the south. It is likely that this morphology is due to the presence of core C3 and the core C2 itself, which could be scattering the gas of both lobes. This could be a case of an interaction between molecular outflows and dense cores, as was found in the OMC-2 region (Shimajiri et al. 2008; Sato et al. 2023).

The near-IR counterpart of the molecular outflow activ- ity manifests as extended emission arising from the core C1 and pointing towards the southeast direction, which perfectly matches the position of the outflow blue-OC1 (see Fig. 4). It is well known that the origin of the continuum emission at K_s-band around protostars can be explained as a scattered light nebulos- ity, where the light scattering process occurs in the walls of a cavity that was cleared out in the circumstellar material by a jet (Bik et al. 2006) and/or emission of H2 likely associated with shocked gas (McCoey et al. 2004, and refences therein). Inter- estingly, the cavity/jet nebula, as observed towards other similar sources (see Weigelt et al. 2006; Paron et al. 2016, and references therein), extends only to one side. To justify this unidirectional asymmetry, it was proposed that the observed near-IR features might be related to a blueshifted jet with the redshifted coun- terpart not detected at the near-IR bands because they are more highly extinct. This phenomenon is clearly seen in this source.

Regarding the mid-IR emission, a comparison between Figs. 1 and 3 unambiguously shows that the 4.5 µm extended emission of the EGOG338 and the outflow blue-OC1 positionally coincide and exhibit the same inclination in the plane of the sky. This suggests that the main contribution to the EGO emission comes from the outflow activity of the core C1, and in particular from the blue lobe. As is the case for the near-IR emission, the mid-IR counterpart associated with the outflow red-OC1 is not detected.

Below, we present a comparison between some of our main results regarding molecular outflow activity and the work of Li et al. (2020), which is one of the most comprehensive and most recent studies carried out with similar observations to those used in the present work. The authors present a statistical study with ALMA data (beam ~ 1.″2) towards more than 40 dense cores with associated molecular outflow activity.

We estimated molecular outflow masses ranging from 0.08 to 0.77 M_⊙ (see Table 3) and dynamical ages on the order of 10³ yr. Li et al. (2020) found outflow masses ranging from 0.001 to 0.32 M_⊙ and dynamical ages going from about 10³ to some 10⁴ yr, which indicates that the outflows found towards AGAL 338 are among the youngest and most massive.

Li et al. (2020) found a median ratio of outflow mass to core mass of about 8 × 10⁻³. Therefore, considering that the range of masses estimated for the outflow red-OC1 goes from 0.25 to 0.77 M_⊙, we conclude that a mass of about 10 M_⊙ for the core C1 would be more likely than the lower value of about 3 M_⊙.

We estimated energies for the molecular outflow red-OC1 going from 3.9 × 10⁴⁵ to 1.2 × 10⁴⁶ erg, while Li et al. (2020) found energies for the molecular outflows of the sample ranging from 4.0 × 10⁴¹ to 1.2 × 10⁴⁵ erg. Therefore, the molecular out- flow red-OC1 is at least three times more energetic than the most energetic outflow found by Li et al. (2020).

Finally, following Li et al. (2020), we estimate for the core C1 a mass accretion rate, M, that goes from 1.5 × 10^-5 to 4.2 × 10⁻⁵ M_⊙ yr⁻¹. Therefore, using the highest value for Ṁ, and considering the estimated dynamical age of about 4.2 × 10³ yr for the outflow red-OC1, the young protostar embedded in the core C1 could have at most 0.4 M_⊙. Therefore, even assuming the highest value of about 10 M_⊙ for the mass of the core C1, we consider that a high-mass star is unlikely to form in this core.

5.3 Chemistry

Star-forming regions, and in particular HMCs, are excel- lent astrochemical laboratories with which to study complex-molecule formation in space (e.g. Jørgensen et al. 2020; Coletta et al. 2020). In turn, this understanding helps us to better charac- terise these interesting condensations of gas and dust where the stars form.

In this section, we discuss the presence of the molecules pre- sented in Fig. 6 in light of the most up-to-date astrochemical knowledge. Such molecular species are discussed individually, addressing the morphology of the emission in the whole inves- tigated region, the chemical and physical conditions that they trace, and so on, in order to obtain a comprehensive chemi- cal interpretation of the analysed molecular cores in terms of the star formation processes. Regarding the main core, C1, in Appendix A we present the spectra of the four ALMA band6 spectral windows, which show the chemical richness and com- plexity of this core.

5.3.1 CH₃CN and CH₃CCH

Propyne (also called methyl acetylene, CH₃CCH) and methyl cyanide (CH₃CN) are symmetric top molecules used as good indicators of temperature (e.g. Brouillet et al. 2022, and see Sect. 4.3.1). These molecular species are usually detected in hot molecular cores (e.g. Brouillet et al. 2022) and Ch₃CN was also found in protoplanetary discs (Öberg et al. 2015). The CH₃CCH is likely produced in interstellar ices through combination of radicals (Kalenskii et al. 2022) and via successive hydrogena- tion of physisorbed C₃ (Hickson et al. 2016; Wong & An 2018). CH₃CN is also formed in interstellar grains through the recombi- nation of radicals such as CN and CH₃ (Hernández-Hernández et al. 2014). However, as different authors point out (Andron et al. 2018; Brouillet et al. 2022), CH₃CN would trace the inner regions of the cores because it needs a higher temperature to sublimate from the surface of dust grains, while CH₃CCH emis- sion is found to preferentially trace the colder envelopes, which would explain the discrepancy in the temperature obtained from both emissions. By inspecting Figs. 6a,b, the morphology of the emission of both molecules is quite similar. They are concen- trated mainly at the bulk of the emission that contains cores C1, C2, and C3. Core C4 also presents emission of both molecules. In the case of CH₃CCH, a feature also appears towards the north- west in correspondence with the position of the outflow red-OC1 (see Fig. 3). This may suggest that the outflow activity in this region could desorb molecular species frozen in the dust grains, enriching the gas phase chemistry in the diffuse gas, for instance releasing CH₄, which seems to be important for the CH₃CCH chemistry in the gas phase (Calcutt et al. 2019).

5.3.2 HC₃N

It is known that the shortest cyanopolyyne HC₃N, the cyanoacetylene, is helpful when exploring gas associated with hot molecular cores (Bergin et al. 1996; Taniguchi et al. 2016; Duronea et al. 2019). As shown in Fig. 6c, the HC₃N emission is mainly concentrated in the cores, and in general encompasses the continuum emission. Towards the northwest, another maxi- mum of the emission of this molecular species is observed that is not associated with any core traced in the continuum emis- sion. This maximum is in positional coincidence with the red lobe of the molecular outflow associated with core C1, suggest- ing that the HC₃N would trace not only the chemistry generated in the envelopes of the hot cores but also that related to the shocked gas (Hervías-Caimapo et al. 2019). The slightly elon- gated HC₃N feature extending towards the northeast from the bulk of the emission, which is coincidental with the direction of the red lobe related to core C2, suggests the same interpretation.

5.3.3 H₂CS

Astrochemical modelling shows that thioformaldehyde can originate from the organosulfur chemistry that can be initiated in star-forming regions via the elementary gas- phase reaction of methylidyne radicals with hydrogen sulfide (Doddipatla et al. 2020). H₂CS has been studied much less than its oxygen-substituted analogue, formaldehyde (H₂CO), but has been used to study cores and outflows (Minh et al. 2011; el Akel et al. 2022). For instance, Xu et al. (2023) observed a H₂CS line with multiple components, and used it to estimate the tem- perature of a core embedded in a massive hub-filament system. In our case, the H₂CS is mainly concentrated in the core C1 with some surrounding extended emission (see Fig. 6d). The core C4 presents weaker but well-defined and correlated emis- sion. A faint structure can be seen towards the northwest in correspondence with the position of the outflow red-OC1 (see Fig. 3).

5.3.4 CH₃OH

Methanol is a very important molecule that has been widely observed in both gas phase and solid state in the ISM (Qasim et al. 2018 and references therein). As indicated by these latter authors, it is generally accepted that CH₃OH formation is more efficient through solid state interactions on icy grain mantles, as the cold dense cores are highly suitable sites for its forma- tion chemistry. It is known that, in gas phase, this species is a precursor to several complex molecular species (Ceccarelli et al. 2017). Multiple lines in a wide range of frequencies, even in maser emission, are usually observed towards star-forming regions, and many of them are used to trace molecular outflows (e.g. Bachiller et al. 1995; Palau et al. 2007). Shocks gener- ated by jets and outflows are known to be efficient at boosting methanol to their gas phase (Rojas-García et al. 2022). Figure 6e shows that all the analysed cores exhibit CH₃OH emission. It is worth noting the methanol feature extending towards the north- east, which has a perfect morphological correspondence with the molecular outflow red-OC2 (see Fig. 3). This confirms the nature of this molecular feature studied with the ¹²CO emission in Sect. 4.2 and suggests that the outflow activity is releasing CH₃OH from the solid state to the gas phase in the region. Additionally, some extended methanol emission appears towards the northwest, which may be related to the molecular outflow red-OC1.

5.3.5 HNCO

Isocyanic acid is a simple molecule containing the four main atoms essential for life as we know it, and can therefore be con- sidered as a prebiotic molecule. Indeed the smallest molecule possessing the biologically important amide bond, formamide (NH₂CHO), appears to be a close relative of HNCO (Haupa et al. 2019) in terms of its chemistry. It was suggested that, while HNCO can be formed in the gas phase during the cold stages of star formation, NH₂CHO forms most efficiently on the dust mantles, remaining frozen until the temperature rises enough to sublimate such icy mantles. The hydrogenation of HNCO is a likely formation route leading to NH₂CHO (López-Sepulcre et al. 2015). As shown in Fig. 6f, the HNCO emission is strongly concentrated at core C1. A small protrusion of weaker emission extends southwards containing cores C2 and C3, and some weak emission also seems to be associated with core C4. We suggest that the emission of this molecular species is tracing warm gas associated with the external layers of the cores.

5.3.6 CN

The cyano radical (CN), one of the first detected interstel- lar molecular species (McKellar 1940; Adams 1941), is a key molecule in many astrochemical chains. For instance, given that CN is very reactive, with molecules possessing double and triple carbon bonds (C=C and C≡C respectively), it is involved in the formation of cyanopolyynes (Gans et al. 2017), such as the HC₃N presented in Sect. 5.3.2. Figure 6g displays the distribution of the CN emission, showing that in general, CN maximums do not spatially coincide with the peaks of the continuum emission as found in other works (Beuther et al. 2004; Paron et al. 2021). In our case, this can be appreciated mainly in cores C2 and C3. As Beuther et al. (2004) point out, this issue may be due to the fact that the source embedded in the cores is at such an early evo- lutionary stage that it does not generate enough UV photons to produce CN emission. Another possibility is that the lack of CN is due to its depletion related to the production of HC₃N. In any case, the CN emission would trace diffuse and extended gas sur- rounding the cores, as it was found in several molecular cores by Paron et al. (2021). However, given the extended features in the CN emission, which seem to coincide with the positions of the outflows, we carried out a detailed kinematic analysis of this emission. Figure 10 shows the CN N = 2−1, J = 5/2−3/2 (F = 5/2−3/2) emission distribution integrated between −80 and −70 km s⁻¹ (in blue) and between −55 and −40 km s⁻¹ (in red). The systemic velocity of the complex is about −64 km s⁻¹. A similar morphology and kinematics can be seen to that shown in the ¹²CO J = 3−2 emission (see Fig. 3), suggesting that the CN is also tracing the molecular outflow activity related to the cores C1, C2, and C3. Moreover, the CN emission shows features likely related to red and blue lobes of a molecular outflow arising from the core C4.

It is worth noting that the outflow cavity walls, which are narrow zones in between the cold dense quiescent envelope material and the lower-density warm cone where outflows are propagating, are pronounced in UV irradiation tracers such as CN. Therefore, CN emission might highlight the border of these cavity walls (Tychoniec et al. 2021). We conclude that we are presenting very clear observational evidence that the CN traces the molecular gas related to the external part of the outflows, mainly to the cavities generated by them.

Fig. 10 CN N = 2–1, J = 5/2−3/2 (F = 5/2−3/2) emission distri- bution integrated between −80 and −70 kms−1 (blue) and between −55 and −40 km s−1 (red). The systemic velocity of the complex is about −64 kms−1. The white contours represent the radio continuum at 340 GHz (12 m array). Levels are at 1, 10, 30, 60, 90, and 140 mJy beam−1 . The beam of the line emission is indicated in the bottom left corner.

5.3.7 C³⁴S

CS (carbon monosulfide) is the most ubiquitous among the sulfur-bearing molecules in the ISM. This molecular species was widely used to trace dense gas in star-forming regions (e.g. Bronfman et al. 1996), and more recently in molecular fila- ments, and hot and prestellar cores (Kim et al. 2020; Zhou et al. 2021; el Akel et al. 2022). The less abundant isotopologue C³⁴S has been used to measure a possible ³²S/³⁴S Galactic gradient (Yu et al. 2020) and the CS depletion in prestellar cores (Kim et al. 2020). These latter authors found that the C³⁴S emis- sion is not centrally peaked; they found that the position where the intensity peaked is significantly shifted when compared with the dust continuum maps, suggesting that the CS species became significantly depleted in the central high-density region of prestellar cores. In our case, we find that cores C1 and C4 traced in the dust continuum coincide with the C³⁴S peaks, while cores C2 and C3 lie in the region of the bulk of the extended emission (Fig. 6h). Given that the investigated cores are active, we suggest that after the depletion of the CS molecules in the prestellar phase, the chemistry produced by the star-forming pro- cesses would contribute to increasing the abundance of such sulfur-bearing molecular species. Additionally, the elongation in the C³⁴S emission towards the northeast and the feature extending towards the northwest, which are coincident with the molecular outflows red-OC2 and red-OC1, respectively, allow us to suggest that this molecule may also be a tracer of molecular outflows.

5.3.8 HDO

Water is a fundamental requirement for life as we know it; under- standing its evolution, from its formation in molecular clouds to its presence in protoplanetary discs, is a challenge. In meeting this challenge, some major questions will likely be answered, such as whether or not life can arise in other planetary sys- tems. However, H₂O emission lines are rarely observed from the ground, which is not the case of partially deuterated (HDO) and fully deuterated (D₂O) water isotopes. Molecules tend to attach a D atom rather than an H atom because deuterated species have larger reduced masses and lower binding energies caused by the different zero-point vibrational energy (Phillips & Vastel 2003), and this favours the production of species such as HD. The degree of deuterium fractionation in water is par- ticularly related to the environmental conditions where it takes place (Jensen et al. 2019) and serves as a robust tracer of the chemical and physical water evolution in star-forming regions (Ceccarelli et al. 2014). The enrichment of species such as HDO is initiated by exothermic reactions, and therefore the deu- terium fractionation in water is expected to occur in the cold cloud molecular phase and later on the surface of dust grains (Kulczak-Jastrzebska 2017; Jensen et al. 2021). Figure 6i dis- plays the HDO emission map, centred and compacted only at core C1, for which high temperatures were derived (see Table 7). Given its spatial distribution, we suggest that HDO emission comes from evaporated molecules due to the heating suffered by the ice layers on the grain surface. The release of this molecule into the gas-phase caused by desorption enriches the environment in deuterated species, as was studied in classical hot cores (Kulczak-Jastrzebska 2017; Csengeri et al. 2019). Based on the lack of HDO in the other cores, we suggest that in these regions, such a molecular species would still be frozen at the dust grains.

6 Summary and conclusion

We present a study of the fragmentation and the star for- mation activity towards a massive molecular clump using high-resolution and high-sensitivity ALMA data. The main goal of this work is to find evidence of high-mass star formation at core scale towards the massive clump AGAL G338.9188+0.5494, which harbours the EGO 338.92+0.55(b).

The continuum emission at 340 GHz shows that the clump is fragmented into at least five cores, labelled C1 to C5. The¹² CO J = 3−2 emission reveals the presence of molecular out- flows arising from cores C1, C2, and C4. Core C1 exhibits the most intense outflow activity. The molecular outflow related to core C1 is among the most massive (from 0.25 to 0.77 M_⊙) and energetic (from 0.4 × 10⁴⁶ to 1.2 × 10⁴⁶ erg) outflows, consid- ering studies carried out with similar observations towards this type of source.

Interestingly, the cyanide radical extended emission exhibits the same morphology and kinematics as the ¹²CO J = 3–2 emis- sion, suggesting that the CN molecule is also tracing the same molecular outflow activity. Given that the CN is a UV irradia- tion tracer, we point out that its emission highlights the border of the cavity walls carved out by the outflows.

The CH₃ CN J = 13–12 (K = 2) moment 1 map shows a clear velocity gradient towards the core C1, which is attributable to a rotating disc whose direction is perpendicular to the molecular outflow direction.

The rotational diagrams for CH₃CN and CH₃CCH yield tem- peratures of about 340 and 72 K, respectively, for the core C1. This suggests that the methyl cyanide would be placed closer to the protostar than the methyl acetylene, which would be trac- ing outermost layers of gas. Using a range of temperatures going from 120 K (about the typical molecular desorption temperature in hot cores) to 340 K obtained from the rotational diagram of CH₃CN, the mass of core C1 ranges from 3 to 10 M_⊙. We point out that the use of typical desorption temperatures or temper- atures derived from molecular species such as methyl cyanide, tracing the gas at core scales, is more appropriate for charac- terising cores than using the typical dust temperatures obtained from the clump scales.

The mid-IR 4.5 µm and near-IR K_s band extended emis- sions coincide in position and inclination with the molecular outflow of core C1, in particular with the blueshifted lobe. We suggest that the molecular outflow activity related to core C1 is responsible for most of the brightness of EGO 338.92+0.55(b) at 4.5 µm. Therefore, the counterpart of the EGO, at core scale, should be a molecular outflow with average mass and energy of about 0.5 M_⊙ and 10⁴⁶ erg, respectively.

Based on the estimated accretion rate and the dynamical age of the outflow towards core C1, we estimate an upper mass limit on the protostar forming inside the core of 0.4 M_⊙. Therefore, considering that the mass of the core is at most 10 M_⊙, and that we do not find any evidence of accreting gas filaments, we con- clude that the formation of a high-mass star within this core is unlikely.

Acknowledgements

We thank the anonymous referee for her/his useful com- ments and corrections. M.O. and S.P. are members of the Carrera del Investigador Científico of CONICET, Argentina. N.I. is posdoctoral fellow and N.M. and A.M. are doctoral fellows of CONICET, Argentina. This work was partially supported by the Argentina grant PIP 2021 11220200100012 from CONICET. This work is based on the following ALMA data: ADS/JAO.ALMA # 2015.1.01312, and 2017.1.00914. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ.

Appendix A Band 6 spectra towards core C1

Figures A.1, A.2, A.3, and A.4 display the spectra of band 6 spec- tral windows 0, 1, 2, and 3, respectively, extracted from a beam centred at the position of core C1. A tentative molecular line identification was made using CASA software by cross checking with the JPL and CDMS databases using the Splatalogue.

Fig. A.1 Band 6 spw0 towards core 1.

Fig. A.2 Band 6 spw1 towards core 1.

Fig. A.3 Band 6 spw2 towards core 1. (*) The CN emission is very weak towards this core.

Fig. A.4 Band 6 spw3 towards core 1.

References

All Tables

All Figures

Fig. 1 Large-scale surroundings of EGO G338.92+0.55(b). Left-panel: Overview of the H II region G338.90+00.60 at Spitzer 3.6 (blue), 4.5 (green), and 8.0 (red) µm bands. The white contours represent the continuum emission at 870 µm extracted from the ATLASGAL survey. Levels are: 1, 3, 6, 9, and 12 Jy beam−1. The green square highlights the region studied in this work, which includes the position of EGO G338.92+0.55(b) represented by the yellow star. Right-panel: A close-up view of the EGO at the same Spitzer bands. The black contours represent the ALMA continuum emission at 340 GHz (7 m array). Levels are: 0.2, 0.4, 0.7, and 1.4 Jy beam−1. The beam of the ALMA observation is indicated in the bottom-right corner. The blue, green, and red colour scales go from 10 to 500 MJy sr−1.

In the text

Fig. 2 ALMA continuum emission at 340 GHz (12 m array). The greyscale goes from 1 to 180 mJy beam−1. The blue contour levels are: 1 (about 5 σ), 10, 30, 60, 90, and 140 mJy beam−1. The dashed red contours correspond to −1 mJy beam−1. The beam of 340 GHz continuum emission (12 m array) is indicated in the bottom left corner. The green contours correspond to the ALMA continuum emission at 340 GHz (7 m array) shown in Fig. 1, with levels at 0.2, 0.4, 0.7, and 1.4 Jy beam−1.

In the text

	Fig. 3 12CO J = 3−2 emission distribution integrated between −90 and −70km s−1 (blue), and between −55 and +5 km s−1 (red). The systemic velocity of the complex is about −64 km s−1. The contour levels are at 5, 8, 12, and 18 mJy beam−1. The beam of the 340 GHz continuum and the 12CO J = 3−2 emission is the same and is indicated in the bottom left corner.
In the text

	Fig. 4 Ks band emission from VISTA hemisphere survey with VIRCAM. The green contours represent the continuum emission at 340 GHz (12 m array). Levels are at 1, 10, 30, 60, 90, and 140 mJy beam−1. The beam of the continuum emission at 340 GHz is indicated in the bottom left corner.
In the text

Fig. 5 C17O J = 2−1 moment maps. Top panel: C17O J = 2−1 moment 0 map integrated between −66 and −61 km s−1. The colour scale goes from 0.01 to 0.50 Jy beam−1. The black contours represent the continuum emission at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The beams of the C17O J = 2−1 line and the 340 GHz continuum emission are indicated in the bottom right corner. Bottom panel: C17O J = 2−1 moment 1 map. The units of the colour bar are km s−1. The black contours represent the radio continuum at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The systemic velocity is about −64 km s−1.

In the text

Fig. 6 Integrated emission maps (moment 0) of the selected molecular lines (indicated in each panel and in Table 4). Dashed magenta contours represent the molecular line emission at the following levels (in Jy beam−1 km s−1): 0.1, 0.3, 1, and 4 (CH3CN), 0.05, 0.10, 0.20, 0.50, and 0.65 (CH3CCH), 0.04, 0.1, 0.25, 0.6, 1.3, and 3 (HC3N), 0.05, 0.2, 0.5, 1.2, and 3 (H2CS), 0.1, 0.3, 0.6, 1.5, and 5 (CH3OH), 0.05, 0.2, 1, and 3 (HNCO), 0.02, 0.04, 0.08, and 0.16 (CN), 0.07, 0.12, 0.2, 0.4, 0.6, 1.2, 2, and 3 (C34S), and 0.5, 1, and 2 (HDO). The integration velocity interval ranges from −67 to −61 km s−1 for all molecular lines, except for H2CS and CH3OH (−67 to −62 km s−1) and for HNCO (−67 to −63 km s−1). The green contours represent the continuum emission at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The beams of 340 GHz continuum and line emissions are indicated in the bottom left corner.

In the text

	Fig. 7 Average spectra of the temperature tracer molecules taken over the full size of core C1. Top panel: CH3CN J = 13−12. Bottom panel: CH3CCH J = 14−13. The K projections are indicated.
In the text

	Fig. 8 Rotational diagrams for the core C1. Top panel: CH3CN J = 13−12. Bottom panel: CH3CCH J = 14−13. The red lines show the best linear fitting of the data. The magenta squares in the top panel correspond to CH3CN J = 13−12 K = 7 and K = 8 projections, for which half of the integrated intensity values were considered.
In the text

	Fig. 9 CH3CN J = 13–12 moment 1 map (from K = 4 projection). The black contours represent the continuum emission at 340 GHz with levels at 1, 10, 30, 60, 90, and 140 mJy beam−1. The dashed black line indicates the direction of the molecular outflow related to core C1. The beams are indicated in the bottom left corner.
In the text

Fig. 10 CN N = 2–1, J = 5/2−3/2 (F = 5/2−3/2) emission distri- bution integrated between −80 and −70 kms−1 (blue) and between −55 and −40 km s−1 (red). The systemic velocity of the complex is about −64 kms−1. The white contours represent the radio continuum at 340 GHz (12 m array). Levels are at 1, 10, 30, 60, 90, and 140 mJy beam−1 . The beam of the line emission is indicated in the bottom left corner.

In the text

	Fig. A.1 Band 6 spw0 towards core 1.
In the text

	Fig. A.2 Band 6 spw1 towards core 1.
In the text

	Fig. A.3 Band 6 spw2 towards core 1. (*) The CN emission is very weak towards this core.
In the text

	Fig. A.4 Band 6 spw3 towards core 1.
In the text