© The Authors 2025

1 Introduction

An extrasolar origin has been proposed as a possible source of complex organic molecules and water detected in primitive Solar System bodies (e.g., Hanni et al. 2022). Many of the same molecules have been detected in the interstellar medium (ISM), in particular in protostellar cores and protoplanetary disks. However, the chemical pathways for their production via gas phase and grain catalysis from atomic and molecular species are not well established and neither are the mechanisms for their incorporation into solar bodies such as comets and planets. To answer these questions, searches have been conducted to detect molecules in both the gas phase and in ices in a variety of interstellar sources. To date, astronomical observations have detected over 320 individual molecular species in the gas phase in interstellar, protostellar, and circumstellar environments (McGuire 2022)¹. Furthermore, the majority of these detections are of organic molecules, which are made of carbon bonded with other elements or other carbon atoms. Most of these discoveries have been detections enabled by radio astronomy of spectral signatures of molecules ranging in size from two (e.g., OH and CO) to thirteen atoms in the long chain HC₁₁N and the recently detected ring molecule benzonitrile, C₆H₅CN (McGuire et al. 2018), and even larger polycyclic aromatic hydrocarbons, such as C₁₀H₇CN (McGuire et al. 2021), C₁₂H₈ (Cernicharo et al. 2024), or C₁₆H₉CN (Wenzel et al. 2024).

Systematic searches for species in a broad range of sources are needed so that a complete picture of their abundances can be developed and the chemical pathways forming complex organic molecules can be determined. In the northern hemisphere, the Astrochemical Surveys at IRAM (ASAI) carried out an unbiased spectral survey between 80 and 272 GHz of ten sources spanning a range of evolutionary states, including the starless core TMC- 1 in the Taurus Molecular Cloud (Lefloch et al. 2018). TMC-1 is one of the richest interstellar sources of organic molecules, and it has also been the target of dedicated radio observations of emission spectra at longer wavelengths. The two current systematic studies of TMC-1 at longer wavelengths are the Q-Band Ultrasensitive Inspection Journey to the Obscure TMC-1 Environment (QUIJOTE) using the Yebes telescope (Cernicharo et al. 2021), and GBT Observations of TMC-1: Hunting Aromatic Molecules (GOTHAM) at the Green Bank Telescope (McGuire et al. 2018). Most of the other sources that have been systematically surveyed are highly evolved regions of massive star formation, such as Sgr B2 and Orion, or evolved stars such as IRC+10216. However, broad spectral surveys of cold protostellar cores, even those residing in regions of massive star formation, which is thought to be the environment in which the Solar System formed, are much less explored. The excitation conditions at low temperatures and high densities in these cold cores favor observational surveys at wavelengths ≳3 mm, and for the heavier organics, at wavelengths ≳7 mm.

In the southern hemisphere, far fewer sources have been studied systematically at long wavelengths. The Australia Telescope Compact Array has conducted a Q-band survey of Sagittarius B2 and detected over 53 molecular species (Corby et al. 2015). However, Sgr B2 is an atypical Galactic source, so its chemistry may be more characteristic of energetic massive star-forming cores rather than the pathway to solar-type planet forming disks. In the southern hemisphere, the best-studied low-mass protostar at long wavelengths is Chameleon MMS1, which was observed by Kontinen et al. (2000) at 3 mm wavelength with the Swedish- ESO Submillimeter Telescope (SEST) and by Cordiner et al. (2012) at 7 mm wavelength with the ATNF Mopra 22-m telescope. The Chameleon complex is also a target of the JWST Ice Age Early Release Science program (McClure et al. 2023), which will ultimately make it possible to compare gas phase and ice phase chemistry.

Located at a distance of 190 pc (Galli et al. 2021), Chamaeleon I is one of the closest low-mass star-forming regions in the southern hemisphere (Belloche et al. 2011 and references therein). Embedded within the Chamaeleon I ridge traced by millimeter dust continuum emission are a known prestellar core, a Class 0 protostar; a binary protostellar system, Ced 110 IRS4 (Rocha et al. 2025); and an edge-on Class II protoplanetary disk HH 48 NE (Sturm et al. 2024). To cover a range of physical environments most relevant for the complex organic molecule chemistry, we selected the Class 0 protostar (Cha-MMS1) and the prestellar core (Cha-C2) as targets of our observations.

In this paper, we extend the survey of Chameleon to lower frequencies at the radio K -band using NASA’s Deep Space Network (DSN) 70-m antenna in Canberra, Australia, Deep Space Station-43 (DSS-43), covering the 18–25 GHz frequency range. The two Chamaeleon sources studied here have different ages, with the class 0 protostar Cha-MMS1 being more evolved than the prestellar core Cha-C2. Moreover, being surrounded by a group of YSO’s, the two cores are located in a different environment than the best-studied starless core, TMC-1. The observations thus provide interesting insights into the effects of environment and age on the dense core chemistry. We use multi-wavelength Herschel observations of the dust continuum emission and multiple transitions of ammonia, HC₃N, and HC₅N to derive the kinetic temperatures and densities of the two cores.

In Sect. 2, we discuss the DSS-43 observations, and in Sect. 3 we derive the properties of the cores and molecular column densities. In Sect. 4, we use a Markov chain Monte Carlo (MCMC) analysis to identify features in the spectral scans that are too weak to be identified from individual spectral lines. In Sect. 5, we discuss the differences between the Chamaeleon sources and TMC-1. In Sect. 6, we summarize our results and describe the next steps needed to trace the evolution of prebiotic compounds in protostellar cores.

2 Deep Space Network observations

The DSN radio telescopes had a nearly decade-long history of contributions to the search for organic molecules in the northern hemisphere and study of their formation environment beginning in the early 1990s (Langer et al. 1995; Velusamy et al. 1995; Kuiper et al. 1996; Langer et al. 1997, 1998; Peng et al. 1998; Dickens et al. 2001). The DSN technical capabilities for astrochemistry research have improved significantly in recent years with the installation of a new cryogenic dual-horn dualpolarization 17–27 GHz receiver at the DSS-43 in Canberra, Australia (Kuiper et al. 2019), with its broadband digital spectrometer (Virkler et al. 2020). The wide 8 GHz instantaneous bandwidth of the spectrometer allows for observations of multiple transitions of heavy species that can be used to characterize the density and temperature of the gas with radiative transfer models and dynamical information from the line width and velocity at the line peak. The spectrometer provides sufficiently high spectral resolution to resolve molecular line shapes even in the coldest regions (~10 K) of protostellar cores.

Several observing runs of the Chamaeleon I cloud using the DSN 70-m antenna at Canberra were carried out in February– September 2022. Figure 1 shows the overall morphology of the region as traced by 350 µm dust emission observed with the SPIRE instrument on Herschel. The black circles show the two DSS-43 pointings centered on the class 0 protostar MMS1 and the prestellar core C2. The J2000 source coordinates for Cha- MMS1 are α = 11^h06^m33.13^s, δ = −77º23′35.1″ and for Cha-C2 α = 11^h06^m15.51^s, δ = −77º24′04.9″. All observations were carried out in a position-switching mode using a reference position 5′ N-W of Cha-MMS1 in a direction perpendicular to the extent of the dust ridge (at 11^h05^m30.00^s, −77º21′00.0″; white circle in the upper-right corner of Fig. 1). Circle sizes correspond to the full width half maximum (FWHM) beam size of the Canberra telescope at 22 GHz, 45″. The filled black circle in the lower-left corner shows the SPIRE 350 µm FWHM beam size of 25.2″.

The 350 µm continuum emission in Chamaeleon I is extended on scales comparable to the DSS-43 beam size (see 50% white dotted contour in Fig. 1). Belloche et al. (2011) derived FWHM source sizes of 55″ × 49″ and 98″ × 46″ for Cha-MMS1 and Cha-C2, respectively, based on observations of 870 µm dust continuum emission using the Atacama Pathfinder Experiment (APEX) telescope. We thus assume average source sizes of 52 and 67″ in the column density calculations.

The DSN Canberra K-band digital spectrometer processes sixteen 1-GHz wide bands, which are split into eight separate bands from 18 to 26 GHz for each polarization (Virkler et al. 2020). Each band consists of 32 768 channels with a 30.5 kHz resolution, corresponding to a velocity resolution of 0.35–0.49 km s⁻¹, depending on the frequency. A typical FWHM line width in Cha-MMS1 is 0.6–0.9 km s⁻¹, depending on the species. All lines are thus spectrally resolved. A typical system temperature at the elevation of the source was 77 K. The total on-source integration time was about 14 hours per source. For the seven lowest-frequency bands, two instrumental polarizations were observed, doubling the effective observing time. At frequencies above 25 GHz, only one instrumental polarization was available. The resulting spectra thus have higher noise and are not included in the analysis.

The raw data from the spectrometer were processed into calibrated ON-OFF spectra using the standard DSS-43 data reduction pipeline. The system temperature, continuously monitored using a power meter and scaling with a factor derived using a noise diode and an ambient load before the observation, was used in the standard ON-OFF calibration to obtain spectra in the antenna temperature units, $T_A^{*}$. We refer the reader to Kuiper et al. (2019) for details on the absolute system and receiver temperature calibration. A relative gain correction was applied to the data to account for antenna deformations as a function of elevation. The gain dependence on elevation was determined using measurements of a flux calibrator at different elevations, showing a peak at an about 45° elevation. Kuiper et al. (2019) fit a third-order polynomial to the data, but we used a second-order fit, as discussed in the ATNF Tidbinbilla 70-m Radio Telescope Guide², which provides more accurate values for observations taken at high elevations.

The beam efficiency was not measured directly during our observations. To convert the observed spectra to the main beam brightness temperature units, we used the main beam efficiency of η_mb = 50% (Pineda et al. 2019) rather than the measured DSN aperture efficiency of η_A = 35.5%. This choice is justified given the expected extent of the molecular emission. Moreover, the absolute intensity calibration is not critical, as opacity effects are small or moderate for all detected lines except for ammonia, and we used only the relative abundance ratios among the molecules rather than absolute abundances with respect to H₂ in the comparison with TMC-1.

Subsequent data reduction was carried out using the IRAM CLASS data reduction software³. The data reduction included blanking of noisy channels near the band edges and removing third-order polynomial baselines from individual scans fit across the full frequency range of each band. The resulting baseline- removed spectra were then averaged with 1/σ² weighting to produce the final spectra used in the analysis.

The resulting full-band spectra of Cha-MMS1 and C2 (in antenna temperature $T_{mb}^{*}$ units) are shown in Fig. 2, with the strongest molecular lines labeled. The noise level is not uniform and often increases significantly toward the band edges. In addition, some noise spikes (“spurs”) are present in the spectra. Channels with excess noise near band edges and narrow spurious signals have been blanked, resulting in some gaps in the frequency coverage. The origin of these artifacts was not investigated, as they do not affect any of the spectral lines discussed below.

Since in some cases the rms can vary significantly across the sub-band, spectra of all lines identified in the broadband survey were subsequently re-reduced by computing the local noise in the immediate vicinity of each line and using this value in the weighted average.

Fig. 1 Herschel/SPIRE image of 350 µm dust continuum emission toward the central part of the Chamaeleon I cloud. The dotted contour outlines the extent of the dust emission at 50% of the peak. Black circles mark the Cha-MMS1 and Cha-C2 pointings, and the white circle indicates the reference position used for the DSN observations. The size of the two circles corresponds to the FWHM beam size of the Canberra telescope (45″ ). The black circle in the lower-left corner shows the FWHM size of the SPIRE beam (25.2″). White stars mark locations of NIR38 and J110621228, the two background stars with ice spectra studied by McClure et al. (2023), and the cyan triangle marks the location of the binary protostar Ced 110 IRS4 (Rocha et al. 2025). The lower panels show the dust continuum SED in the Canberra beam, based on PACS and SPIRE observations. The black curves are modified blackbody fits to the SPIRE and PACS 160 µm surface brightness, as described in the text. Typical flux calibration uncertainties are 5%.

3 Results

In this section we discuss the physical conditions in the Chamaeleon cores based on prior observations. We then derived local thermodynamic equilibrium (LTE), and for some species non-LTE, molecular column densities and molecular abundance ratios and compare them with those toward the cyanopolyyne peak in TMC-1 (Gratier et al. 2016). TMC-1, a target of the QUIJOTE and GOTHAM surveys (Cernicharo et al. 2021; McGuire et al. 2018), is the best-studied dense core and a reference source for astrochemical studies. Observations of other cores, such as those studied here, will help determine to what extent the chemistry of TMC-1 is representative of typical dense cores.

Figure 2 shows the radio K-band spectra of the Chamaeleon MMS1 and C2 cores taken with the Canberra DSN telescope. The main individual lines detected, including those of HC₃N, HC₅N, c-C₃H₂, C₄H, CCS, C₃S, NH₃, and the rare isotopologues ¹⁵NH₃ and c-C₃HD, are shown in more detail in Figs. 3 and 4. The line parameters derived from Gaussian and hyperfine structure (HFS) fits are listed in Tables 1 and 2. Below, we derive the molecular column densities and relative fractional abundances of the detected molecules as well as upper limits for selected molecular species.

Fig. 2 Deep Space Network spectra of Cha MMS1 and Cha C2 (upper and lower panels, respectively) corrected for the main beam efficiency. Channels with excess noise near sub-band edges and spurious signals have been blanked, resulting in some gaps in the frequency coverage. Detected spectral lines are identified (see Tables 1 and 2).

3.1 Densities and temperatures of the Chamaeleon cores

Calculations of molecular column densities require prior knowledge of the gas temperature (LTE calculations) or temperature and density (radiative transfer models, such as the large velocity gradient approach). These physical parameters can be estimated from the existing dust continuum as well as molecular data, given that multiple transitions covering a range of excitation conditions are detected.

The two bottom panels in Fig. 1 show dust spectral energy distributions (SEDs) in the DSS-43 beam toward the MMS1 and C2 cores. Modified blackbody fits to the SPIRE and PACS 160 µm surface brightness, I_v = B_v(Td) × [1 − exp(−τ₃₅₀ × (350 µm/λ)^β)], (shown as black curves in the bottom panels of Fig. 1) give average dust temperatures of ~13 K and a grain emissivity exponent β = 1.9 in both cores. We excluded the PACS 70 µm data from the fit, as the emission may be contaminated by a warmer dust component distributed across the outer layers of the cloud.

The H₂ column densities in the DSS-43 beam can be computed using the formula N (H₂) = 2 aρ R_gd/3m_H × τ₃₅₀/Q₃₅₀ (Lis & Goldsmith 1990), where a = 0.1 µm is the grain radius, ρ = 3 g cm⁻³ is the mean density, R_gd = 100 is the gas to dust ratio, m_H is the hydrogen mass, and Q₃₅₀ is the 350 µm grain emissivity coefficient. Extrapolating the 125 µm grain emissivity of Hildebrand (1983, 7.5 × 10⁻⁴) with a v² frequency dependence gives Q₃₅₀ = 1 × 10⁻⁴, corresponding to the grain mass opacity coefficient κ₃₅₀ = 3 Q/4/a/ρ/ R_gd = 0.025 cm² g⁻¹. The values of Q₃₅₀ and κ₃₅₀ are highly uncertain, and values four times higher have been suggested for the Orion Molecular Cloud (see the discussion in Lis et al. 1998 and Goldsmith et al. 1997).

The resulting H₂ column densities toward MMS1 and C2 are thus in the range $N_{{\text{H}}_2}=4-17\times {10}^{22}$ and 2.5–10 × 10²² cm⁻², respectively. Assuming a line-of-site depth equal to ~55″ (as implied by the 50% contour of the 350 µm emission in Fig. 1) and a distance of 150 pc, the mean H₂ densities in the DSS- 43 beam toward MMS1 and C2 are $n_{{\text{H}}_2}=3-14\times {10}^5$ and 2−8 × 10⁵ cm⁻³, respectively.

A wide range of column and volume densities in Cha MMS1 has been reported in the literature. Kontinen et al. (2000) derived a low H₂ column density of 1.2[−]1.6 × 10²² cm⁻² and an average H₂ density of ~7 × 10⁴ cm⁻³ based on C¹⁸O observations using the SEST telescope (55″ beam). However, this value depends on the assumed fractional abundance of C¹⁸O, which may be lower than the canonical Taurus value (Frerking et al. 1979) if partial freeze-out occurs in cold high density gas.

Tennekes et al. (2006) reported a slightly higher H₂ column density range of 1.9−4.0 × 10²² cm⁻² based on 1.3 mm dust continuum observations. Their spherical Monte Carlo model that fits SEST observations of HCN isotopologues in Cha-MMS1 has densities ranging from 2.3 × 10⁴ cm⁻³ at the outer 60″ radius and 1.4 × 10⁶ cm⁻³ at the center. Cordiner et al. (2012) assumed an H₂ density of 10⁶ cm⁻³ in the analysis of their ATNF Mopra observations at 32–50 GHz (96–77″ beam), while Belloche et al. (2011) derived peak column densities of 9.2 and 2.7 × 10²² cm⁻² in a 21″ beam toward Cha-MMS1 and C2, respectively, and average gas densities in a 50″ diameter aperture of 9.8 × 10⁵ and 3.9 × 10⁵ cm⁻³, respectively, based on 870 µm dust continuum observations using the APEX telescope.

Observations of HC₃N and HC₅N for which theoretically derived collisional rate coefficients are available can be used to provide an independent estimate of the gas temperature and density. One advantage to this approach is that their emission most likely arises from the same region as other organic molecules. In Appendix A, we show that our HC3N and HC₅N data in conjunction with SEST data of Kontinen et al. (2000) in a similar beam are well reproduced by large velocity gradient (LVG) models with a kinetic temperature of ~8.5 K and a density ≳3 × 10⁵ cm⁻³. This kinetic temperature is close to the range 7.1–7.2 K derived by Cordiner et al. (2012) based on Mopra observations of HC₃N and HC₅N in a larger beam.

The average kinetic temperature of the gas within the DSS- 43 beam can also be constrained by observations of the inversion lines of ammonia (Appendix B). The rotational temperature of 10.9 K for Cha-MMS1 derived in Appendix B agrees with previous ammonia observations, 12.1 K (Tennekes et al. 2006). These values are higher than the temperature implied by LVG models ofHC₃N andHC₅N (Appendix A).

Fig. 3 Spectra of molecular lines other than ammonia detected in Chamaeleon MMS1 (black and magenta histograms) with fits shown in green.

Fig. 4 Spectra of molecular lines other than ammonia detected in Chamaeleon C2 (black and magenta histograms) with fits shown in green.

3.2 Local thermodynamic equilibrium molecular column densities

There are two possible approaches to converting the observed line intensities to molecular column densities, depending on the availability of collisional rate coefficients. For molecules with calculated rate coefficients, these can be used along with LVG radiative transfer models to derive physical properties such as density and temperature of the molecular hydrogen and the column density of the trace molecule given sufficient transitions.

Since collisional rates are not available for some of the molecules studied here, we used the Weeds software package⁴, which can perform a simple modeling of the observed spectra under the assumption of LTE. We used molecular spectroscopy data from the Cologne Database for Molecular Spectroscopy (CDMS) catalog (Müller et al. 2001, 2005) in the calculations.

For ammonia, we used the excitation temperatures derived from the hyperfine structure fits (Sect. 3.2.5 below), while for other molecules we used a value of 8.5 K, as derived from the LVG analysis of the HC3N and HC5N data in Cha-MMS1 (Appendix A). We fixed the source sizes for the MMS1 and C2 cores to the mean values derived above. We then adjusted the molecular column density to be the only parameter to match the observed line intensities reported in Tables 1 and 2. The resulting column densities are listed in Table 4 along with selected column density ratios in Table 5. For molecules with multiple lines detected, the column densities reported in Table 3 correspond to averages of values derived from individual lines.

In the following subsections, we discuss observations of individual molecules detected in the DSS-43 spectra of Chamaeleon MMS1 and C2 and compare their column densities and abundance ratios with reference values for the cyanopolyyne peak in TMC-1 from Gratier et al. (2016). In general, column densities of molecular species in the Chamaeleon cores are an order of magnitude lower than in TMC-1, despite the higher H₂ column densities (Table 3) compared to 10²² cm⁻² in TMC-1 (Gratier et al. 2016). The high molecular column densities make TMC-1 a preferred target for searches for new molecular species. Notable exceptions are c-C₃H₂ and NH₃ , as discussed below.

3.2.1 Cyanopolyynes

A single rotational transition of HC₃N, J = 2 − 1, is within the DSS-43 frequency range. The hyperfine splitting is clearly detected in both cores, indicating optically thin emission. The LVG models of the DSS-43 and SEST observations are consistent with a kinetic temperature of ~8.5 K and a density ≳3 × 10⁵ cm⁻³. Three rotational transitions of HC₅N, from J = 7 − 6 to 9 − 8, are detected in both cores. Also, HC₇N is detected via stacking analysis (see Sect. 4).

Column densities of cyanopolyynes in the Chameleon cores are more than an order of magnitude lower than those in TMC-1 (Gratier et al. 2016). However, the relative abundance ratios (i.e., HC₃N:HC₅N:HC₇N) are similar in both cores (~five; Table 5). As a reference, the HC₃N:HC₅N abundance ratio in TMC-1 is ~4.0, which is comparable to the Chamaeleon values, but the HC₅N:HC₇N ratio is lower, ~1.3 (Gratier et al. 2016). This difference may indicate that the production of long carbon chains is less efficient in Chamaeleon compared to TMC-1.

3.2.2 Polyynes

Three lines of C₄H are detected in both cores. Our column density computations used the latest CDMS value of the dipole moment, 2.1 D, from a quantum chemical calculation by Oyama et al. (2020). For comparison, we used the Gratier et al. (2016) C₄H column density in TMC-1 scaled to a dipole moment of 2.1 D in Tables 4 and 5. The resulting C₄H:HC₅N ratios in the Chamaeleon cores are comparable to that in TMC-1.

3.2.3 Sulfur species

Lines of two sulfur-bearing species, CCS and C₃S, are detected in both cores. Compared to TMC-1, the CCS abundance appears slightly enhanced compared to cyanopolyynes, while the CCS/C₃S ratio is similar to that measured in TMC-1.

3.2.4 Cyclopropenylidene isotopologues

Two lines of c-C₃H₂ are detected within the DSS-43 frequency range. The 2₂₀ − 2₁₁ line at 21.6 GHz is seen in absorption against the CMB, as previously observed in TMC-1 (Madden et al. 1989) and consistent with the LVG models. The 1₁₀ − 1₀₁ line at 18.3 GHz is seen in emission and was used to derive column density estimates. In contrast to cyanopolyynes, column densities of c- C₃H₂ in the two Chamaeleon cores are comparable to TMC-1. The resulting c-C₃H₂/HC₅N ratio is thus a factor of ~25 higher compared to TMC-1 (Table 5).

A line of the deuterated isotopologue c-C₃HD is also detected with a high S/N ratio. The resulting D/H ratio is the same in the two cores, ~0.23, and consistent with the TMC-1 value. Majumdar et al. (2017) carried out a detailed study of c-C₃HD in the solar-type protostar IRAS16293-2422 and also derived a high deuterium fraction of 0.14, which is an order of magnitude higher than the value predicted by their chemical model. A possible explanation is that current astrochemical models appear to overpredict the abundance of c-C₃H₂ (Agúndez & Wakelam 2013; Sipilä et al. 2016).

3.2.5 Ammonia isotopologues

The HFS is clearly detected in the (1,1) and (2,2) lines in both sources. The HFS fits to the ammonia inversion line spectra, carried out using the IRAM CLASS software (Fig. 5), indicate that the NH₃ (1,1) line is optically thick in both Cha-MMS1 and C2 (optical depths of the main hyperfine component equal to 7.6 and 2.9, respectively, Tables 1 and 2). The (2,2) line is optically thin, within the fit uncertainties, and the (3,3) line is only detected in Cha-MMS1. The excitation temperatures within the K = 1 rotational ladder can be derived directly from the HFS fits⁵. From the optically thick spectra of the (1,1) inversion transitions, corrected for the source coupling, we derived excitation temperatures of 7.6 and 5.5 K for Cha-MMS1 and Cha-C2, respectively.

In addition to the excitation temperature, the two K-ladders of ammonia can be used to derive the kinetic temperature. In Appendix B, we discuss the derivation of the ammonia rotational temperature, taken as a measure of the kinetic temperature of the gas, and the corresponding correction to the molecular column density derived from LTE analysis. We derived kinetic temperatures of 10.9 and 11.1 K in Cha-MMS1 and Cha-C2, respectively.

The ammonia column densities in the two Chamaeleon cores are an order of magnitude higher than in TMC-1. We note that our Cha-MMS1 value of 3.6 × 10¹⁵ cm⁻² is in good agreement with the value of 1.4 × 10¹⁵ cm⁻² derived by Tennekes et al. (2006), after correcting for the difference in beam filling factors between DSS-43 and a much larger 80″ beam of the Parkes telescope. The resulting NH₃ /HC₅ N abundance ratio is a factor of ~125 higher in the Chamaeleon cores compared to TMC-1. The NH₃ abundance enhancement with respect to cyanopolyynes in Chamaeleon compared to TMC-1 is thus a factor of -five higher compared to that in c-C₃H₂ .

Because of the very high ammonia column density, the (1,1) line of ¹⁵NH₃ is also detected toward Cha-MMS1 at an S/N ratio of eight. The LTE models using Weeds imply a high ¹⁴N/¹⁵N isotopic ratio of ~690 in ammonia. We note that we assumed the same source size of 52″ in the computations of the ¹⁴NH₃ and ¹⁵NH₃ column densities. Given the high optical depth of the ¹⁴NH₃ (1,1) line, it is possible that the effective source size is larger compared to ¹⁵NH₃. A 50% increase in the ¹⁴NH₃ source size would reduce the column density by a factor of 1.3, bringing the ¹⁴N/¹⁵N isotopic ratio in ammonia down to ~525.

As discussed by Lis et al. (2010), nitrogen displays the largest isotopic variations in the Solar System after hydrogen, typically explained by mixing of various proto-solar or pre-solar reservoirs. Earth, Mars, Venus, and most primitive meteorites have nitrogen isotopic ratios within 5% of the terrestrial atmospheric value, ¹⁴N/¹⁵N = 272 (see Marty et al. 2009). However, the protosolar nebula was poorer in ¹⁵N, with ¹⁴N/¹⁵N = 450, as evidenced by infrared and in situ measurements in the Jupiter atmosphere $({530}_{-170}^{+380}$ , ISO, Fouchet et al. 2000; 435 ± 57, Galileo, Owen et al. 2001; 448 ± 62, Cassini, Abbas et al. 2004; 450 ± 106, Cassini; Fouchet et al. 2004) and solar wind measurements (442 ± 131, 2σ, Genesis; Marty et al. 2009). The proto-solar ¹⁴N/¹⁵N ratio is in agreement with the local ISM value (450 ± 22, Wilson & Rodd 1994; or 414 ± 32 at the birth place of the Sun, Wielen & Wilson 1997).

As discussed by Redaelli et al. (2023) and references therein, nitriles are often enriched in ¹⁵N with typical ¹⁴N/¹⁵N abundance ratios of 140–460, while N₂H⁺ instead appears deficient in ¹⁵N with isotopic ratios 580–1000. Using GBT, Lis et al. (2010) measured ¹⁴N/¹⁵N ratios of 334 ± 50 and 344 ± 173 (3σ) in ammonia in Barnard 1 and NGC 1333, respectively. In a recent study, Redaelli et al. (2023) derived a low isotopic ratio of 210 ± 50 toward NGC 1333 IRAS4A and 390 ± 40 toward the center of L1544.

Redaelli et al. (2023) argue that ammonia ices are enriched in ¹⁵N, leading to a decrease in the ¹⁴N/¹⁵N ratio when the ices are sublimated back into the gas phase, for instance, due to the temperature rise in protostellar envelopes. The high ¹⁴N/¹⁵N in Cha-MMS1 may suggest that ice sublimation is not a dominant process within the DSS-43 beam.

Fig. 5 Spectra of the detected ammonia inversion lines in Chamaeleon MMS1 and C2. The transition and source are labeled in each panel. The observational data are shown as black histograms, and spectral fits are shown in green.

4 Detections and limits obtained using line stacking analysis

Spectral line stacking of different transitions arising from the same species has been proposed to overcome the low signal- to-noise ratio of these faint signals across wavelength regimes (Chen et al. 2013; Lindroos et al. 2016) and has already shown success in detecting new species at radio frequencies (Loomis et al. 2018; Walsh et al. 2016). By aligning signals from multiple weaker transitions and considering their collective effect, the signal-to-noise ratio can surpass a detection threshold that individual lines might fall below. Loomis et al. (2021) introduced a detection method specialized for sparse data, combining MCMC inference with spectral stacking. Their version of the MCMC code, available on GitHub, is specific to GBT observations of TMC-1. We generalized this code for use with other sources as described in Appendix C and applied this revised MCMC code to our DSS-43 spectra of HC₇N in the Chamaeleon molecular cloud.

4.1 Longer cyanopolyynes

We detected HC₇N with a high S/N ratio by stacking its rotational transitions from J = 16 − 15 to 21 − 20. Individual spectra are normalized to the brightest line, as predicted by the MCMC model, and weighted by 1/σ² in the final average. This approach leads to a robust 9.1σ detection of HC₇N in Cha-MMS1 and a 7.5σ detection in Cha-C2. The resulting spectra are shown in Figs. 3 and 4, lower-right panels, and the column densities and abundance ratios are reported in Tables 4 and 5, respectively.

4.2 Upper limit for benzonitrile

The DSS-43 frequency range covers 350 rotational transitions of benzonitrile covering a wide range of line intensities and upper level energies. Figure 6 (upper panel) shows a template LTE model of the benzonitrile emission for a temperature of 8 K, normalized to the intensity of the 18.41 GHz line. Transitions above the green horizontal line, which corresponds to 32% of the intensity of the brightest line, are included in the stacking analysis. This cutoff is arbitrary, but it corresponds to a factor of ten increase in the integration time to reach the same S/N ratio compared to the brightest line. Including weaker line lines in the analysis does not improve the S/N ratio in the stacked spectrum.

We extracted spectra of all lines above the cutoff defined above and scaled them to the intensity of the 18.41 GHz line using the LTE template. We then averaged all spectra using 1/σ² weighting. The resulting spectra toward Cha-MMS1 and Cha-C2 are shown in Fig. 6 (lower panel). No benzonitrile emission is detected in either source. The corresponding 3σ upper limits for the column density are listed in Table 4, and the resulting model spectra are shown as green curves in Fig. 6. The 3σ upper limits for the benzonitrile column density achieved in the Chamaeleon sources are a factor of two higher than the value derived for TMC-1 (McGuire et al. 2018), and the corresponding upper limits for the relative abundance of benzonitrile with respect to HC₅N are a factor of three higher than the TMC-1 value.

Fig. 6 Top: local thermodynamic equilibrium model spectrum of benzonitrile in the DSS-43 frequency range for a temperature of 8.5 K. Line intensities are normalized to the 18.41 GHz transitions. Transitions above the green horizontal lines are included in the stacking analysis. Bottom: stacked spectra of benzonitrile toward Cha-MMS1 and Cha-C2 normalized to the 18.41 GHz line. The green curves are 3σ upper limit LTE models corresponding to column densities reported in Table 4.

4.3 Unidentified features

A single ~7.5σ unidentified feature is seen in the Cha-C2 spectrum at 22.902 GHz. The feature is not present in the MMS1 spectrum.

5 Discussion

One general result from our study is that column densities of most molecules detected in the Chamaeleon sources are comparatively lower than those in TMC-1. Interestingly, the average temperatures of both Chamaeleon sources are comparable to that of TMC-1, approximately 10 K (an average of the gas and dust temperatures), while the average density of our sources is ten times or more higher than that of TMC-1.

While the cynaopolyyne peak in TMC-1 is a particularly well studied starless core, the presence of molecular abundance gradients across TMC-1 is well established in the literature. Pratap et al. (1997) compared relative abundances of several molecules toward the cynaopolyyne, ammonia, and sulufur monoxide peaks in TMC-1. Of particular interest for the results presented here is the NH₃/HC₃N ratio, determined to be 5.8, 16.7, and 25.4 at the three positions. Hirahara et al. (1992) attributed such variations to differences in the chemical evolutionary stage, with carbon-chain molecules produced in early stages and ammonia in late stages. While the abundance ratios at the ammonia and sulfur monoxide peaks in TMC-1 are higher than that at the cynaopolyyne peak, they are still significantly lower than the values derived here for the Chamaeleon cores (Table 5). In the Hirahara et al. (1992) scenario, this would imply that the Chamaeleon cores are characterized by late stage chemistry.

Law et al. (2018) studied carbon chain molecules toward 16 embedded low-mass protostars. The median molecular column densities of HC₃N, HC₅N, CCS, and C₃S in their sample are a factor of five to 12 lower that those in the Cha-MMS1 core, while the C₄H median column density is a factor of 1.4 higher. The resulting median HC₃N/HC₅N and CCS/HC₅N abundance ratios (~ten) are about a factor of two higher than the Chamaeleon values, while the CCS/C₃S median abundance ratio, ~25, is a factor of five higher. The median C₄H/HC₅N ratio, ~110, is a factor of 30 higher than the Chamaelon values (see Table 5 of Law et al. 2018).

Several additional factors may explain the differences in the observed column densities between these sources. First, in higher-density environments, such as those in the Chameleon sources, molecules likely freeze-out onto dust grains more efficiently, as the timescale for freeze-out is inversely proportional to the gas density (Seo et al. 2019). Second, lower-density regions such as TMC-1 may have longer chemical timescales, allowing molecular species to persist for extended periods and resulting in higher gas-phase column densities (Majumdar et al. 2015). Third, lower-density cores may allow for greater penetration of ambient UV photons, facilitating photodesorption of molecules from grain surfaces and leading to higher abundances (Öberg et al. 2007). Fourth, variations in cosmic-ray ionization rates between these sources could significantly influence gas-phase chemistry, contributing to the differences in the observed column densities (Taniguchi et al. 2024; Seo et al. 2019). It is also possible that these cores have different ages as well as different initial elemental abundances, thus leading to different C/O ratios compared to TMC-1 (Taniguchi et al. 2024). For example, Loison et al. (2014) reported that several carbon-chain groups (such as C_n, C_nH, C_nH₂, C_2n+1O, C_nN, HC_2n+1N, C_2nH⁻, and C₃N⁻) show a strong dependence on the assumed C/O ratios and evolutionary stage. In particular, for carbon chains, gas-phase chemistry dominates in the early stages ( 10⁵ years), whereas depletion becomes significant in the later stages.

Determining which of these processes dominates in our case will require further studies employing advanced chemical models, including isotope chemistry observed in our samples. This will be the focus of a future investigation.

6 Summary

With this work, we have extended the survey for organics in the southern hemisphere to 1.3 cm by observing two cores in the Chamaeleon complex using NASA’s DSN antenna in Canberra, Australia, over the frequency range of 18–25 GHz. We surveyed the class 0 protostar Cha-MMS1 and the prestellar core Cha-C2, which represent two stages in the evolution of dense cores. We used the detections of ammonia, cyanopolyynes, and far-infrared dust continuum to characterize the density and temperature in the Chamaeleon cores and calculate the molecular column densities and their relative ratios. The main results can be summarized as follows:

• Several molecules are detected in both cores, including HC₃N, HC₅N, C₄H, CCS, C₃S, NH₃, and c-C₃H₂. A longer cyanopolyyne, HC₇N, is detected with high confidence via spectral stacking analysis.

• While molecular column densities in the two Chamaeleon cores are typically an order of magnitude lower compared to the cynaopolyyne peak in TMC-1, the molecular abundance ratios are in general agreement with the TMC-1 values. The two exceptions are c-C₃H₂, which is enhanced by a factor of ∼25 with respect to cyanopolyynes in the Chamaeleon cores, and ammonia, which is enhanced by a factor of ∼125.

• A deuterated isotopologue c-C₃HD is detected in both cores, with a high D/H ratio of ∼0.23 in c-C₃H₂, in general agreement with observations of other sources, such as TMC-1 or IRAS1629-2422.

• A rare isotopologue of ammonia, ¹⁵NH₃, is also detected in Cha-MMS1, suggesting a high ¹⁴N/¹⁵N ratio of ∼690 in ammonia. However, this ratio may be artificially enhanced due to the high optical depth of the ¹⁴NH₃ (1,1) line, which increases the effective source size.

• The ring molecule benzonitrile, a tracer for the non-polar molecule benzene, is not detected in either Chamaeleon core. The 3σ upper limits for the benzonitrile column density achieved are about a factor of two higher than the TMC-1 value (McGuire et al. 2018), and the resulting upper limits for the relative abundance of benzonitrile with respect to HC₅N are a factor of three higher than that measured in TMC-1.

Our results suggest that the chemical composition of the two Chameleon cores is different in several aspects from that of TMC-1. However, the chemical composition of the observed organics in the two Chameleon cores are similar to each other despite representing different stages of core evolution. This result suggests there may be a steady state solution during this period, but comparison to earlier and later stages of cores will be required to reveal the relationship between chemical composition and age. Ongoing DSS-43 observations of additional sources will determine whether these conclusions are generally applicable to other southern star-forming cores.

Acknowledgements

This research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (80NM0018D0004) and funded through the internal Research and Technology Development program. We thank Steve Lichten, Joe Lazio, and the DSN staff for their support and assistance with the DSS-43 observations, and an anonymous referee for helpful comments. L.M. acknowledges financial support provided by DAE and the DST-SERB research grant (MTR/2021/000864) from the Government of India.

Appendix A Large velocity gradient analysis of HC₃N and HC₅N

We use a large velocity gradient (LVG) model of HC₃N and HC₅N spectra to determine the best fit densities and temperatures for the cores where organic molecules are detected. For both species theoretical calculations exist for the collisional rate coefficients and there are sufficient number of spectral lines detected from different transitions to constrain the solutions. We use the offline version of the RADEX code (van der Tak et al. 2007) to calculate intensities I(K km s⁻¹), opacity τ, and level population as a function of density, kinetic temperature, and column density. We consider a range of T_k, n(H₂), and column density N(HC_nN) suggested by earlier observations of the Chamaeleon cores to find a good fit to the data. We use the line widths listed in Table 1 and a background temperature T_bɡ =2.73K. For HC₃N we use collisional rate coefficients from Faure et al. (2016) which are available from the Leiden Atomic and Molecular database⁶ while those for HC₅N are from Bop & Lique (2025). Given the low gas temperatures, in both cases we assume that hydrogen is para-H₂. The resulting intensities were compared to the DSN and SEST data for HC₃N. The DSN and SEST beam widths are similar, ~45″ and 55″, respectively and comparable to the size of the cores, as discussed in Sect. 2. We exclude the 7 mm Mopra observations of HC₃N and HC₅N (Cordiner et al. 2012) because its beam size, 96″ to 77″, is much larger than those of the DSN and SEST, and larger than the source size. These differences in angular resolution introduces uncertain filling factor corrections for beam dilution and beam coupling.

Figure A.1 shows results for HC₃N intensities for densities, n(H₂) = 3×10⁴, 3×10⁵, and 3×10⁶ cm⁻³ which cover the range of densities derived from dust emission and C¹⁸O, as discussed in Sect. 3.1. We also consider four values of kinetic temperatures, as follows: T_k = 7.1 K is the LTE value derived by Cordiner et al. (2012) for HC₃N and HC₅N, 8.5 K is our best solution for HC₃N and HC₅N, 10.9 K is derived from our ammonia data, and 16 K was chosen to study the impact of a higher temperature on the distribution of intensities. The solution for each T_k is shown in a separate panel and the DSN observed intensity is plotted as a circle and the SEST ones as squares. The best fit solution is for T_k = 8.5 K and column density N(HC₃N) =8.4×10¹³ cm⁻². There is little difference between the solutions at densities 3 × 10⁵ and 3 × 10⁶ cm⁻³. At densities above 3 × 10⁵ cm⁻³ the transitions considered here are approaching thermalization. In fact, low-frequency HC5N lines are fully thermalized at densities as low as 10⁵ cm⁻³, as suggested by their critical densities. In the LVG solution all the observed HC₃N lines have an opacity τ < 0.2 and the excitation temperatures range from 7.4 K to 8.6 K consistent with the lines approaching thermal equilibrium for the case T_k = 8.5 K. We note that a solution with a low density of 3 × 10⁴ cm⁻³ and a higher temperature of 16 K is also consistent with the observations. However, these temperature and density values are inconsistent with other the values suggested by other tracers (Sect. 3.1).

For HC₅N we ran an LVG analysis over the same range of T_k and n(H₂) and the results are shown in Fig. A.2. The column density N(HC₅N) = 1.75 × 10¹² cm⁻² for the best fit temperature T_k = 8.5 K, and is very similar for the other three values of T_k. The intensities of the three lines detected with the DSN are shown as red circles. The constraints on physical parameters from HC₅N is less stringent than for HC₃N as SEST did not detect any HC₅N lines. We have excluded the HC₅N data from Mopra (Cordiner et al. 2012) for the same reasons discussed above for HC₃N. Without enough contrast in excitation conditions we can only constrain T_k from 8.5 K to 16K, and n(H₂) to be greater than 3 × 10⁴ cm⁻³ . However, the column density is insensitive over the likely range of temperatures and densities considered here, as well as those indicated by the HC₃N analysis, and that is the key parameter needed for interpreting chemical abundances.

Fig. A.1 Large velocity gradient models of the HC3N emission in Cha- MMS1 for different densities (filled cyan, blue, and green squares, as labeled). The four panels correspond to kinetic temperatures of 7.1, 8.5, 10.9, and 16 K, left to right, respectively. The open red circle shows our DSN observation while the red squares are SEST observations of Kon-tinen et al. (2000). The error bars correspond to a typical 20% absolute calibration uncertainty.

Fig. A.2 Large velocity gradient models of the HC5N emission in Cha- MMS1 for different densities (filled cyan, blue, and green squares, as labeled). The four panels correspond to kinetic temperatures of 7.1, 8.5, 10.9, and 16 K, left to right, respectively. The open red circles show our DSN observations. The error bars correspond to a typical 10% relative calibration uncertainty.

Appendix B Ammonia rotational temperatures and column densities

In the case of ammonia, the column density calculation is complicated by the presence of two different temperatures describing the rotational level population: the excitation temperature T_ex within a given K–ladder, and the rotational temperature T_rot describing the relative populations of the metastable rotational levels at the bottom of each K–ladder. In the absence of allowed radiative transitions between different K –ladders, the latter can be taken as a measure of the gas kinetic temperature.

Despite this complication, we can use the excitation temperature derived from the HFS fit to determine the ammonia column densities in the lowest metastable level of each K–ladder. These values can then be used to determine the rotational (kinetic) temperature of the gas.

The upper level molecular column density of a rotational transitions within the telescope beam can be computed from the observed integrated line intensity of the transitions using the standard formula (see, e.g., Lis et al. 2002) $$N_u=\frac{8\pi k{v}^2}{h{c}^3{A}_{ul}}\frac{1}{1-\left[J_v\left(T_{bq}\right)/J_v\left(T_{ex}\right)\right]}{\displaystyle \int T_R}dv,$$(B.1)

where ν is the line frequency, T_ex is the excitation temperature, A_ul is the Einstein spontaneous emission coefficient, E_u is the upper level energy, J_ν(T) = hv/k/(e^hν/kT − 1) is the radiation temperature of a blackbody at a temperature T, T_bɡ is the cosmic background temperature (2.7 K), and ∫ T_Rd_υ is the opacity and beam efficiency corrected integrated line intensity.

The ratio of the populations of the upper and lower levels of the inversion transitions at the bottom of each K–ladder is given by N_l/N_u = ɡ_l/ɡ_u exp((E_u − E_l)/kT_ex. Therefore, the population in the lowest metastable level of each K–ladder can be computed as $$N_l=\frac{{𝑔}_l}{{𝑔}_u}{e}^{\left(E_u-E_l\right)/k{T}_{ex}}\frac{8\pi k{v}^2}{h{c}^3{A}_{ul}}\frac{1}{1-\left[J_v\left(T_{bg}\right)/J_v\left(T_{ex}\right)\right]}{\displaystyle \int T_R}dv.$$(B.2)

Here, the excitation temperature is that derived from the HFS fit and the resulting column densities for Cha-MMS1 and Cha-C2 are listed in Table B.1.

The rotational temperature can then be computed from the populations of two metastable levels at the bottom of different K–ladders, denoted i and j, using the formula $$N_i/N_j=\frac{{𝑔}_i}{{𝑔}_j}\exp \left(-\frac{E_i-E_j}{k{T}_{\mathit{rot}}}\right).$$(B.3)

From the lower-level column densities of the (1,1) and (2,2) transitions reported in Table B.1 we derive rotational temperatures of 10.9 K both in Cha-MMS1 and Cha-C2.

The total NH₃ column densities computed by the Weeds package use the partition functions computed assuming that all levels are populated at the assumed excitation temperature, T_ex. However, the rotational levels of ammonia are populated according to two temperatures: the excitation describing the population within a given K–ladder and the rotational temperature connecting the different K–ladders. To first approximation, in the low temperature limit applicable to the Chamaeleon sources, the correction factor to the partition function can be computed including only the K = 1 and 2 ladders, $$Q\left(T_{ex},T_{rot}\right)=Q\left(T_{ex}\right)C\left(T_{rot}\right)=Q\left(T_{ex}\right)\left(1+\frac{{𝑔}_2}{{𝑔}_1}\exp \left(-\frac{E_2-E_1}{k{T}_{rot}}\right)\right),$$(B.4)

where the Q(T_ex) is the standard partition function from the spectroscopic catalog and indices 1 and 2 refer to the lowest energy metastable levels in the K = 1 and 2 ladders, respectively. For T_rot = 11 K, the correction factor C = 1.038, suggesting that most of the population is within the K = 1 ladder. The population in the ground state of the K = 3 ladder is less than 0.1% of that in the ground state of the K = 1 ladder. For consistency with other molecules, the total NH₃ column densities reported in Table 4 are computed from opacity corrected line intensities of the (1,1) line using the Weeds package, and applying the correction factor C, derived above.

Appendix C Line stacking analysis

We adapted the TMC-1 Markov Chain Monte Carlo fitting and LTE spectral simulator scripts of Loomis et al. (2021), refactoring the codebase and improving user accessibility. Originally tailored for Green Bank Telescope (GBT) observations of TMC-1, the code has been restructured to support a range of customizable user inputs, making it adaptable to new sources. It has also been supplemented with documentation. A user-friendly logging system with real-time progress tracking and upgraded file management was introduced to handle various MCMC runs and automate all statistical preprocessing. Walkers have been reconfigured to consistently initialize within physical bounds, reducing the need for manual fine-tuning between molecular species. To address the degeneracy between source size and column density, users may fix the source size if it can be determined externally, thereby tightening constraints on column density, or leave it free if unknown. Column density may also be initialized via maximum likelihood estimation to further stabilize the inference process. The result is a robust open-source software tool for MCMC inference of spectra, successfully validated on the GOTHAM dataset, and publicly available on GitHub.⁷

To apply the Python-based MCMC tool to the DSN dataset, HC₅N was selected as a benchmark molecule due to the presence of multiple spectral lines within the DSS-43 frequency range. This choice was strategic for several reasons: 1) the MCMC algorithm fits all emission lines simultaneously, diverging from traditional single-line methods; 2) each line possesses a high signal-to-noise ratio; 3) as HC₅N is a member of the cyanopolyyne family, accurate parameter estimation for this molecule provides informed initial guesses for parameter space exploration of structurally related, larger molecules like HC₇N or HC₉N.

The MCMC workflow begins with data reduction and preparation, following the methodology of the GOTHAM survey analysis (Loomis et al. 2021). We performed an initial spectral simulation across the entire DSN bandwidth using the LTE spectral simulator configured with DSS-43 telescope properties. For each transition exceeding a threshold intensity, a spectral window of 4.1 ± 1.5 km s⁻¹ was defined and a local noise level was calculated in the vicinity of the line.

Fig. B.1 Observed spectra of the J=7−6 to 9−8 lines of HC5N in Cha-MMS1 (black histograms) compared to model spectra generated using best-fit values from the Python MCMC fit (red histograms).

Following spectral reduction, we applied a Python-based MCMC script to derive posterior distributions and covariances for the five free parameters: source size, column density (N_c), excitation temperature (T_ex), source velocity (V_LSR), and line width (ΔV). We implemented the affine-invariant ensemble sampler provided by the emcee toolkit (Foreman-Mackey et al. 2013), exploring the parameter space with 128 walkers for 10,000 steps. Non-informative priors were used to maintain physically plausible best-fit values. The log-likelihood function was defined as the negative half of the sum of squared residuals between observed and modeled spectra, weighted by the inverse variance.

For HC₅N in both MMS1 and C2 cores, we began with a “template run” — initializing walkers in a wider range using no prior information or assumptions to prioritize a thorough exploration of the parameter space. After assessing the quality of these fits visually through corner plots and ensuring that the parameter distributions align with our physical expectations, the posteriors from these template runs were used as priors for subsequent analyses of HC₅N and HC₇N, and they can also be extended to larger cyanopolyynes. The best-fit values align with previous surveys, successfully benchmarking the MCMC algorithm and providing confidence for future applications. The 50th percentile parameter values for Cha-MMS1 are overlaid with the reduced DSN spectra in Fig. B.1.

Having verified that the MCMC approach successfully derives parameter values in agreement with previous higher- frequency surveys of Chamaeleon (Cordiner et al. 2012), we applied it to the detection of HC₇N. This longer cyanopolyyne presents 8 emission lines in the DSS-43 frequency range, with 7 falling within the region relatively free of excess noise. The posteriors from the HC₅N template run serve as the priors for the HC₇N analysis. The resulting corner plot for HC₇N in Cha- MMS1 is shown in Fig. C.1, with similar posterior distributions for most parameters compared to HC₅N, and a column density an order of magnitude lower, consistent with those discussed in Sect. 5.

For this detection, we also employ spectral line stacking to further boost the signal-to-noise ratio. This technique aligns multiple transitions along the velocity axis, enhancing the overall signal by averaging across several lines. In our analysis, we stack the rotational transitions from J = 16 − 15 to 23 − 22, normalizing individual spectra to the brightest line and weighting by 1/σ² for the final average, as shown in Fig. 4. This approach leads to a robust 9.1σ detection of HC₇N in Cha- MMS1 and a 7.5σ detection in Cha-C2 (Fig. 3 and 4, lower-right panels). The MCMC method shows great promise for detecting other heavy molecules.

Fig. C.1 Corner plot for the MCMC fit ofHC7N parameter covariances and distributions in the Cha-MMS1. The diagonal shows the probability distributions of each parameter as histograms, with vertical lines marking the 16th, 50th, and 84th percentiles, corresponding to ±1σ for a Gaussian posterior. The off-diagonal scatter plots depict the correlations between pairs of parameters, with each axis representing one of the fit parameters. The beam-averaged best-fit column density is consistent with the Weeds pencil beam value listed in Table 4.

References

All Tables

All Figures

Fig. 1 Herschel/SPIRE image of 350 µm dust continuum emission toward the central part of the Chamaeleon I cloud. The dotted contour outlines the extent of the dust emission at 50% of the peak. Black circles mark the Cha-MMS1 and Cha-C2 pointings, and the white circle indicates the reference position used for the DSN observations. The size of the two circles corresponds to the FWHM beam size of the Canberra telescope (45″ ). The black circle in the lower-left corner shows the FWHM size of the SPIRE beam (25.2″). White stars mark locations of NIR38 and J110621228, the two background stars with ice spectra studied by McClure et al. (2023), and the cyan triangle marks the location of the binary protostar Ced 110 IRS4 (Rocha et al. 2025). The lower panels show the dust continuum SED in the Canberra beam, based on PACS and SPIRE observations. The black curves are modified blackbody fits to the SPIRE and PACS 160 µm surface brightness, as described in the text. Typical flux calibration uncertainties are 5%.

In the text

	Fig. 2 Deep Space Network spectra of Cha MMS1 and Cha C2 (upper and lower panels, respectively) corrected for the main beam efficiency. Channels with excess noise near sub-band edges and spurious signals have been blanked, resulting in some gaps in the frequency coverage. Detected spectral lines are identified (see Tables 1 and 2).
In the text

	Fig. 3 Spectra of molecular lines other than ammonia detected in Chamaeleon MMS1 (black and magenta histograms) with fits shown in green.
In the text

	Fig. 4 Spectra of molecular lines other than ammonia detected in Chamaeleon C2 (black and magenta histograms) with fits shown in green.
In the text

	Fig. 5 Spectra of the detected ammonia inversion lines in Chamaeleon MMS1 and C2. The transition and source are labeled in each panel. The observational data are shown as black histograms, and spectral fits are shown in green.
In the text

Fig. 6 Top: local thermodynamic equilibrium model spectrum of benzonitrile in the DSS-43 frequency range for a temperature of 8.5 K. Line intensities are normalized to the 18.41 GHz transitions. Transitions above the green horizontal lines are included in the stacking analysis. Bottom: stacked spectra of benzonitrile toward Cha-MMS1 and Cha-C2 normalized to the 18.41 GHz line. The green curves are 3σ upper limit LTE models corresponding to column densities reported in Table 4.

In the text

Fig. A.1 Large velocity gradient models of the HC3N emission in Cha- MMS1 for different densities (filled cyan, blue, and green squares, as labeled). The four panels correspond to kinetic temperatures of 7.1, 8.5, 10.9, and 16 K, left to right, respectively. The open red circle shows our DSN observation while the red squares are SEST observations of Kon-tinen et al. (2000). The error bars correspond to a typical 20% absolute calibration uncertainty.

In the text

	Fig. A.2 Large velocity gradient models of the HC5N emission in Cha- MMS1 for different densities (filled cyan, blue, and green squares, as labeled). The four panels correspond to kinetic temperatures of 7.1, 8.5, 10.9, and 16 K, left to right, respectively. The open red circles show our DSN observations. The error bars correspond to a typical 10% relative calibration uncertainty.
In the text

	Fig. B.1 Observed spectra of the J=7−6 to 9−8 lines of HC5N in Cha-MMS1 (black histograms) compared to model spectra generated using best-fit values from the Python MCMC fit (red histograms).
In the text

Fig. C.1 Corner plot for the MCMC fit ofHC7N parameter covariances and distributions in the Cha-MMS1. The diagonal shows the probability distributions of each parameter as histograms, with vertical lines marking the 16th, 50th, and 84th percentiles, corresponding to ±1σ for a Gaussian posterior. The off-diagonal scatter plots depict the correlations between pairs of parameters, with each axis representing one of the fit parameters. The beam-averaged best-fit column density is consistent with the Weeds pencil beam value listed in Table 4.

In the text