ATOMS: ALMA Three-millimeter Observations of Massive Star-forming regions - VIII. A search for hot cores by using C₂H₅CN, CH₃OCHO and CH₃OH lines

As stated before, 453 compact dense cores were found in 146 massive clumps (Liu et al. 2021). We have inspected the line emission of the 453 cores one by one and extracted spectra at the line-rich positions. In total 60 line-rich cores are considered as hot core candidates. The results are in good agreement with those by Liu et al. (2021) where 54 cores were found in a statistic manner to have significant detection of at least 20 COM transitions with line intensities larger than 3 $\sigma$ level. Then we use the eXtended CASA Line Analysis Software Suite (XCLASS,²²2https://xclass.astro.uni-koeln.de/ Möller et al. 2017) for further line identification and parameter calculation. The XCLASS accesses the Cologne Database for Molecular Spectroscopy (CDMS; Müller et al. 2001, 2005) and Jet Propulsion Laboratory molecular databases (JPL; Pickett et al. 1998). Assuming the molecular gas satisfies local thermodynamical equilibrium condition (LTE), the XCLASS solves radiative transfer equation and produces synthetic spectra for specific molecular transitions by taking source size, beam filling factor, line profile, line blending, excitation, and opacity into account. In the XCLASS modelling, the input parameters are source size, beam size, line velocity width, velocity offset, rotation temperature and column density (Möller et al. 2017). In our case, we take deconvolved angular sizes of the continuum sources as source sizes which are listed in Table 1. The velocity offsets are determined when referred to commonly detected CH₃OH line at 100.6389 GHz. Therefore we set rotation temperatures, column densities and velocity widths as free parameters to simulate the observed spectra. To obtain optimized rotation temperature and column density parameters, we employ Modeling and Analysis Generic Interface for eXternal numerical codes (MAGIX; Möller et al. 2013) for further calculation. Note that we only consider line-rich species of C₂H₅CN, CH₃OCHO and CH₃OH molecule in this work.

The frequency setup covers many of C₂H₅CN and CH₃OCHO transitions in SPWs 7 and 8. The observed C₂H₅CN and CH₃OCHO transitions span upper level energies of 30–139 K and 22–58 K, respectively. Therefore we do not expect very hot gas components to be detected. Only a few CH₃OH lines are tuned in our observations with upper level energies ranging from 17 to 724 K. More than 3 transitions of C₂H₅CN and CH₃OCHO are detected in the 59 line-rich cores. The line transitions of C₂H₅CN and CH₃OCHO in IRAS 09018-4816 are too weak to derive rotation temperature, but 3 CH₃OH transitions are detected. Rotation temperatures and column densities can be derived simultaneously. The XCLASS calculations suggested that all the 59 line-rich cores have gas temperature derived from C₂H₅CN and CH₃OCHO higher than 100 K. IRAS 09018-4816 has gas temperature above 100 K based on CH3OH line data. Considering the distances (Liu et al. 2021a) and source sizes, the 60 cores have source size smaller than 0.1 pc. We then confirm the 60 cores as hot cores, and give their coordinates, rotation temperatures and column densities in Table 1. We have checked the literature and found that 15 of these hot cores have been reported before (I18089-1732, Beuther et al. 2004; I18032-2032/G9.62+0.19, Liu et al. 2015; Dall’Olio et al. 2019; I18507+0110/G34.43+0.24, Beltrán et al. 2009; Fu & Li 2017; Calcutt et al. 2014; I17441-2822/Sgr B2(M), Sánchez-Monge et al. 2017; IRAS18056-1952/G10.47+0.03, Rolffs et al. 2011; I18507+0121/G34.26+0.15 , Sakai et al. 2013; Mookerjea et al. 2007; I16060-5146, I16065-5158, I12326-6245, I14498-5856, I15254-5621, Araya et al. 2005; Dedes et al. 2011; I16547-4247, I17233-3606, I18182-1433, Hernández-Hernández et al. 2014; I17175-3544/NGC6334I, Brogan et al. 2016). Thus, 45 hot cores are newly detected in our work. This is currently the largest hot core sample observed with similar angular resolution and spectral coverage. Forty-one cores have more than 3 CH₃OH transitions detected. Note that this is a lower limit since the frequency setup of C43-2 and C43-3 configurations has a 48 MHz difference which leads to two CH₃OH transitions not covered in C43-3 observations. For CH₃OH with less than 3 transitions detected, column densities are estimated by assuming gas temperature equal to the rotation temperatures of oxygen-bearing molecule CH₃OCHO, or equal to the rotation temperatures of C₂H₅CN in case that CH₃OCHO is not detected.

Figure 1 presents sample spectra toward a few hot cores with modeled molecular spectra (C₂H₅CN, CH₃OCHO and CH₃OH) overlaid on the observed ones. Clearly lots of C₂H₅CN and CH₃OCHO lines are observed and large number of lines are emitted from C₂H₅CN and CH₃OCHO species. The spectra emitted from the hot cores also show different emission features. From Figure 1 and the figures in the supplementary material, one can see that line peak intensities of C₂H₅CN are larger than those of CH₃OCHO in 19 cores while the opposite is seen in 5 cores. It is worth mentioning that C₂H₅CN line emission is absent in 4 cores while CH₃OCHO is not detected in 9 cores. The differences in line emissions from these cores may imply differences in physical and chemical environments among these sources.

Line images of various molecules can provide valuable information on their spatial distributions and possible chemical routes. We choose three molecular line transitions of C₂H₅CN, CH₃OCHO and CH₃OH at 98523, 98792 and 100639 MHz for the images, respectively. The sample images are shown in Figure 2, and other images are presented in the supplementary material. Overall, most of the C₂H₅CN, CH₃OCHO and CH₃OH emissions are associated with the continuum images, though there are position offsets between peaks of the line and continuum images. All the line images show compact source structure and the three line emissions mainly distribute over small regions, indicating hot core properties. It can be seen that the emission peaks of C₂H₅CN are coincident with those of CH₃OCHO in 28 cores. Considering position accuracy of 0.3 arcsec in line images, 29 cores clearly show nitrogen and oxygen separation, i.e., the emissions of C₂H₅CN and CH₃OCHO peak at different positions. So far only a few hot cores exhibit nitrogen and oxygen differentiation in space. Our observations provide the largest hot core sample with positional separation between nitrogen- and oxygen-bearing molecules.

From Table 1, one can see that the rotation temperatures of C₂H₅CN, CH₃OCHO and CH₃OH range from 100–285 K, 100–230 K and 100–290 K respectively. Most of the cores have gas temperatures of 100–200 K for the three species. When compared with C₂H₅CN and CH₃OCHO, CH₃OH has the largest column densities ranging from 8.6$\times$10¹⁶ to 1.5 $\times$10¹⁹ cm^-2 which is 2–3 orders of magnitude higher than those of C₂H₅CN and 1–2 orders of magnitude higher than those of CH₃OCHO. Figure 3 shows the relationships between rotation temperatures and column densities for the three molecules in our sample. Column densities of CH₃OH increase as their rotation temperatures rise. A linear fitting to the data gives log(N(CH₃OH))=0.022T_rot+14 with correlation coefficient of 0.86. A similar trend is seen for CH₃OCHO with the relation log(N(CH₃OCHO))=0.017T_rot+14 and correlation coefficient of 0.68. The linear fitting to C₂H₅CN data gives a correlation coefficent of 0.29, suggesting that column densities of C₂H₅CN are not sensitive to its rotation temperatures, though a weak increasing trend is seen. The results appear to indicate that the two oxygen-bearing molecules are chemically-related species or the two species reside in the same astrophysical environments.

Figure 4 presents the relationships of column densities among the three species. Overall, column densities of both C₂H₅CN and CH₃OCHO have positive correlation with that of CH₃OH. The linear fittings give log(N(CH₃OCHO))=1.1log(N(CH₃OH))–2.5 and log(N(C₂H₅CN))=1.2log(N(CH₃OH))–5.1 with correlation coefficients of 0.95 and 0.82 for CH₃OCHO and C₂H₅CN, respectively. The models suggested that oxygen-bearing molecules CH₃OCHO and CH₃OH and nitrogen-bearing molecule C₂H₅CN have different forming pathways (Charnley et al. 1992; Rodgers & Charnley 2001). Probably the column densities of the three molecules depend on the initial cloud environments. As shown in Figure 3, the column densities of the oxygen-bearing molecules of CH₃OCHO and CH₃OH correlate well with the rotation temperatures while the nitrogen-bearing molecule C₂H₅CN does not follow the same trend. A possible explanation is that excitation of CH₃OCHO and CH₃OH depends on both temperature and density while excitation of C₂H₅CN is mainly related to its column density.

Among the detected COMs so far, CH₃OH has the highest gas phase abundance relative to H₂. The infrared observations have shown that CH₃OH is the most abundant molecule relevant to water ice (e.g., Dartois et al. 1999; Gibb et al. 2000b; Ehrenfreund & Charnley 2000). Grain surface chemical models also suggested that formation of many of other COMs (especially the oxygen-bearing molecules) is related to CH₃OH. From the abundance criteria, the CH₃OH can be taken as one of the reference molecules to compare with other sources. The relative abundances defined by f = N(x)/N(CH₃OH) are listed in Table 1, where x is the specific molecule CH₃OCHO or C₂H₅CN. The relative abundances range from 8.4$\times$10^-4 to 4$\times$10^-2 for C₂H₅CN and from 2.2$\times$10^-2 to 2.8 $\times$10^-1 for CH₃OCHO, respectively, which are in agreement with previous results in hot cores (Bøelund et al. 2019; Mookerjea et al. 2007; Allen et al. 2017; Guzmán et al. 2018; Law et al. 2021; Feng et al. 2016; Bonfand et al. 2017, 2019; Molet et al. 2019; Qin et al. 2010).

Observationally the evolutionary sequence of high-mass star formation starts from massive pre-stellar cores, to high-mass protostellar objects (HMPOs), hot cores, HC Hii and UC Hii regions where massive stars are already formed (e.g. Beuther et al. 2007; Menten et al. 2005). During the processes of massive star formation, the gravitational energy is converted into thermal energy to form the HMPOs. The radiation from HMPOs, HC Hii and UC Hii regions can heat up their immediate environments to form hot cores. Therefore hot cores are observed to be associated with dense and hot regions near the young high-mass protostellar objects and the UC Hii regions.

In this section, we want to differentiate internal and external heating mechanisms by comparing the relative position of line images of COMs and UC Hii regions. Radio continuum at centimeter wavelength and radio recombination lines are characteristics of the UC Hii regions associated with newly formed high-mass young stars. ATOMS covers the H40$\alpha$ line transition which can be used for identifying UC Hii regions. Figure 2 presents sample images of the continuum, organic molecular, H40$\alpha$ and SiO lines. From the left panels of Figure 2 and the figures in the supplementary material, 28 UC Hii regions are identified based on H40$\alpha$ line images. When comparing the continuum and H40$\alpha$ images with the CH₃OH line images, we find that the morphologies of CH₃OH line images are similar to those of continuum images, and that the CH₃OH emissions peak at the peak positions of the continuum, for most sources without associated UC Hii regions. We note that although IRAS 08303–4303, IRAS 13484–6100, IRAS 16484–4603, IRAS 17158–3901, IRAS 17233-3606, and IRAS 18182–1433 have no UC Hii region counterparts, their CH₃OH emission peaks are located offset from those of continuum emissions. Probably shocks traced by the 4.5 $\micron$ and SiO push the molecular materials away. For the sources having H40$\alpha$ line emission, peaks of CH₃OH are offset from the continuum images except for IRAS 18056–1952. Previous molecular line images were observed to be not always consistent with the continuum emission in cases where continuum emission includes contributions from dust and free-free emission of UC Hii regions. Also no internal energy sources were identified, suggesting that these hot cores are externally heated (e.g., Wyrowski et al. 1999; De Buizer et al. 2003; Mookerjea et al. 2007). Then we simply classify the observed hot cores in our work into two categories: externally heated hot cores and internally heated hot cores. Twenty four hot cores are thought to be heated by adjacent UC Hii regions and the other sources are classified as internally heated hot cores, suggesting that some hot cores are not simply linked to HMPOs and precursors to UC Hii regions in evolutionary sequence during massive star formation (Law et al. 2021), but they only trace physical and chemical environments of hot, dense and line-rich regions. Note that we consider IRAS 18056-1952 and I18032-2032c1 as internally heated since both line and continuum images of the two sources peak at UC Hii region positions.

Figure 5 shows the cumulative distributions of the rotational temperatures and column densities of C₂H₅CN and CH₃OCHO of the two groups of hot cores, the internally heated ones and the externally heated ones, respectively. No statistical difference of the rotational temperatures of C₂H₅CN between the two groups can be found. The rotational temperatures of CH₃OCHO tend to be higher in the externally heated hot cores than in the internally heated hot cores, although this difference is not very much significant with a p-value of 0.14 given by the Kolmogorov-Smirnov test. The externally and internally heated cores show no obvious difference (with a p-value of 0.25) in the cumulative distributions of the column densities of CH₃OCHO. While significant difference (with a p-value of 0.027) can be seen between the cumulative distributions of the column densities of C₂H₅CN ($N_{\rm C_{2}H_{5}CN}$). The values of $N_{\rm C_{2}H_{5}CN}$ in the externally heated hot cores are a factor of two higher than those in the internally heated hot cores. Of course, one may expect that higher temperature component from highly excited lines are shifting towards the UC Hii regions for the externally heated hot cores while higher temperature components will be located at the dusty continuum peaks for internally heated hot cores. Future high angular resolution and high frequency observations covering high energy line transitions can verify the occurrence and characteristics of externally and internally heated mechanisms in detail.

Nitrogen and oxygen differentiation has been suggested in some hot cores (Blake et al. 1987; Wyrowski et al. 1999; Fayolle et al. 2015; Qin et al. 2015, 2010), showing different rotation temperatures and abundances, and nitrogen- and oxygen-bearing COMs peaking at different spatial positions. Observations have shown that many hot cores have excitation temperatures of nitrogen-bearing COMs higher than those of oxygen-bearing COMs (van’t Hoff et al. 2020, and references therein). Derived abundances of oxygen-bearing and nitrogen-bearing COMs have positive correlations with the other oxygen-bearing and nitrogen-bearing COMs respectively (e.g., Bisschop et al. 2007; Bergner et al. 2017; Suzuki et al. 2018). The observations have also shown that nitrogen-bearing COMs peak at protostar position while oxygen-bearing COMs are peaked offset from the central source (Qin et al., 2010; Fayolle et al. 2015).

From Table 1, rotation temperatures of C₂H₅CN in most hot cores are generally lower than those of CH₃OH, which is inconsistent with previous observations. As stated in Section 3.1, a possible explanation is that our 3 mm observations of C₂H₅CN lines have low upper level energies of 30–139 K, and then the hot components of the cores are not sampled. The LTE calculations find CH₃OCHO and CH₃OH molecules to have larger column density than that of the nitrogen-bearing molecule C₂H₅CN. Column density of CH₃OH correlates well with that of CH₃OCHO, but also has a positive correlation with C₂H₅CN which also conflicts with previous results as stated above. Probably our targeted lines of the C₂H₅CN and CH₃OCHO have low upper level energies which only sample the freshly evaporated species, and then can not probe the innermost and hottest structure of the hot cores. Nitrogen and oxygen separation is clearly seen in 29 hot cores in which C₂H₅CN and oxygen-bearing molecules peak at different spatial positions (see Figure 2 and the figures in the supplementary material).

COMs are suggested to be particularly important in both astrophysics and astrochemistry, but also to be linked to origin of life. However no firm conclusions about their formation are reached, even for commonly detected CH₃OCHO and C₂H₅CN. Various chemical models are proposed to interpret their origination (Charnley et al. 1992, 1995; Caselli et al. 1993; Millar et al. 1991; Garrod et al. 2006, 2008; Taquet et al. 2012, 2015, 2016; Rodgers & Charnley 2001, 2003; Pols et al. 2018; Choudhury et al. 2015). In summary, main chemical routes are gas-phase and grain-surface chemical reactions as well as interaction between gas-phase and grain-surface molecules. Large sample observations of their relative abundances and spatial distributions will be able to provide clues in investigating their forming pathways.

Higher gas-phase abundances of CH₃OH are derived from (sub)millimeter wavelength observations toward massive star formation regions, which is attributed to icy CH₃OH evaporated from grain mantles leading to large abundance by various chemical models. From Figure 3, column densities of CH₃OH increase as rotation temperatures rise, providing additional evidence that CH₃OH originates from grain surface and is then released into gas phase when gas temperature is higher than 100 K. Our large sample observations give positive correlation between CH₃OH and CH₃OCHO which appears to indicate that CH₃OCHO is formed in gas phase and is related to the evaporated CH₃OH as proposed by Charnley et al. (1992, 1995). However, gas phase mechanisms by molecule-ion reactions have very low efficiency in producing CH₃OCHO (Horn et al. 2004; Geppert et al. 2006). The relationship between the observed column densities and rotation temperatures of CH₃OCHO show the same trend as of CH₃OH that has also been repeorted by previous observations (Law et al. 2021), which may hint that CH₃OCHO also originates from grain surface and then is evaporated into phase as temperature increases. The warm up model developed by Garrod et al. (2006) suggested that CH₃OCHO is mainly synthesized on grain surface in lower temperatures and gas phase reactions play a dominant role in forming CH₃OCHO when the gas temperature is above 200 K. In this scheme, formation of CH₃OCHO is attributed to both grain surface and gas phase reactions. In our case, the observations of CH₃OCHO transitions at 3 mm band have low upper level energies and should sample freshly evaporated species. The rotation temperatures of CH₃OCHO in most cores are lower than 200 K. We then favor CH₃OCHO observed by us at lower frequency band is synthesized on grain surface. Future higher angular resolution observations of high energy level lines can detect inner and even hotter components of the hot cores, and can test if the gas phase reactions play an important role in forming CH₃OCHO.

C₂H₅CN can be formed by ion-molecule reactions in gas phase, but the processes are quite slow. Grain surface reactions through hydrogenation of HC₃N is a more efficient mechanism (Blake et al. 1987; Caselli et al. 1993; Charnley et al. 2004). Contrary to the two complex oxygen-bearing molecules, the column densities of C₂H₅CN did not show positive correlation with its rotation temperatures. A possible explanation is that C₂H₅CN tends to have higher gas temperature than the oxygen-bearing COMs while our lower energy level lines can only trace lower excitation temperatures in the outer envelopes of the hot cores (Öberg et al. 2013). Large amount of C₂H₅CN are not fully evaporated from grain surface. Surprisingly the column densities of C₂H₅CN correlate well with those of CH₃OH which is inconsistent with previous observations. Previous observations have shown that nitrogen-bearing molecules have strong correlation with other nitrogen-bearing molecules and no correlation with oxygen-bearing COMs (Bisschop et al. 2007; Bergner et al. 2017; Suzuki et al. 2018). As argued before, our observations may only sample freshly evaporated species which have not involved in subsequent gas phase reactions yet. The observed correlation between C₂H₅CN and CH₃OH may reflect initial chemical environments or physical differences (Van’t Hoff et al. 2020).

The spatial separation between C₂H₅CN and the two oxygen-bearing molecules can be seen from Figure 2 and the figures in the supplementary material. But no obvious temperature differentiation is observed as in Orion KL, W3(OH) complex and G34.26+0.15 (Blake et al. 1987; Crockett et al. 2014; Wyrowski et al. 1999; Qin et al. 2015; Mookerjea et al. 2007), due to our targeted lines having low upper level energies. Another reason is that in most cases our beam may cover part of both nitrogen and nitrogen cores. Nitrogen and oxygen differentiation is a long-standing problem. Different chemical models give different explanations (Caselli et al. 1993; Garrod et al. 2008; Rodgers & Charnley 2003; Van’t Hoff et al. 2020). Higher resolution observations of highly excited lines from various COMS toward a large sample of hot cores are needed to interpret the observed differentiation.