© The Authors 2025

1 Introduction

Low- and intermediate-mass stars in the later stage of the asymptotic giant branch (AGB) phase generate a substantial amount of stellar winds. These winds strip away the outer layers of the stars, thereby leading to significant mass loss. The ejected material gives rise to a circumstellar envelope (CSE), which is rich in molecules and dust and is considered a natural site for molecule synthesis. To date, over 100 gas-phase molecules and 15 solid-state species have been detected within CSEs (Decin 2021). The evolutionary processes of materials ejected from stellar atmospheres during the expansion of the CSE remain poorly understood and require further investigation. To address that, observational data from a statistically significant sample of CSEs spanning various evolutionary stages across a wide frequency range are essential. Such comprehensive spectral line surveys provide an unbiased perspective on the chemical processes occurring in CSEs (e.g., Cernicharo et al. 2011).

Our research group has initiated a long-term project of searching for circumstellar molecules utilizing single-dish telescopes (Zhang et al. 2008, 2009a,b; Chau et al. 2012; Zhang et al. 2013; Zhang 2020; Zhang et al. 2020; Qiu et al. 2022; Yang et al. 2023). This study represents a continuation of our ongoing series as it presents a comprehensive 3 mm spectral line survey of the archetypal carbon-rich AGB star IRC+10216. The observational data obtained through this survey enable a detailed comparative analysis of the chemical composition in CSEs across different evolutionary stages, providing critical insights into the molecular evolution processes during late stellar evolution.

IRC+10216, the nearest carbon-rich AGB star (120 pc, Groenewegen et al. 1998), has served as a cornerstone of circumstellar chemistry studies since its serendipitous discovery by Becklin et al. (1969). Millimeter-wave breakthroughs commenced with pioneering molecular line detections by Wilson et al. (1971) and Solomon et al. (1971), which established its status as a molecular factory. Subsequently, Wilson et al. (1971), Solomon et al. (1971), and others initiated the detection of specific molecules toward this source at the millimeter-wave band. Treffers & Cohen (1974) detected SiC dust in IRC+10216 through identification of its characteristic 11.3 μm emission band, with gas-phase molecules SiCSi and SiC₂ probably being the key precursors for the formation of SiC dust grains. Modern observations reveal complex morphodynamics. Near-infrared interferometry uncovers a bipolar peanut-shaped core (Menut et al. 2007), while mid-IR imaging exposes nested nonconcen-tric dust shells with significant density fluctuations (Decin et al. 2011). This points to episodic mass loss. The extreme massloss rate (2.4 × 10^-5 M_⊙ yr^-1, Fonfría et al. 2022) generates an envelope heavily enshrouded by dust. This dense dusty outflow creates a stratified chemical structure: ultraviolet photons from the interstellar radiation field are effectively attenuated in the outer envelope (Huggins & Glassgold 1982), while the inner regions sustain active chemistry with enhanced molecular abundances (Huggins & Glassgold 1982; Cherchneff & Glassgold 1993). IRC+10216 hosts an exceptionally rich molecular inventory (106 species, Tuo et al. 2024, hereafter T24), including not only carbon chains but also exotic metal-bearing species such as SiC₄, MgC₄H, and MgC₅N. These characteristics establish it as the archetypal laboratory for studying AGB circumstellar chemistry. However, its apparent chemical uniqueness raises a critical question: Can IRC+10216 truly represent the broader population of carbon-rich AGB stars? Resolving this requires systematic multifrequency observations to disentangle source-specific phenomena from universal chemical processes.

IRC+10216 has been established as a benchmark system for millimeter and submillimeter molecular line surveys; Johansson et al. (1984) detected 45 emission lines within the frequency range of 72.2-91.1 GHz using the Onsala Space Observatory (OSO). Subsequent observations conducted by Kawaguchi et al. (1995) utilized the 45-meter radio telescope at the Nobeyama Radio Observatory (NRO) to systematically survey the 2850 GHz frequency band. This spectroscopic campaign successfully detected 188 molecular lines belonging to 22 distinct chemical species, with notable detections including the long carbon-chain molecules HC₇N and HC₉N. Cernicharo et al. (2000) conducted comprehensive observations of this source within the frequency range 129.0-172.5 GHz using the Institut de Radioastronomie Millimétrique (IRAM) 30m telescope and discovered new species, including those containing rare isotopes of C, Mg, Si, S, and Cl. Employing the 10-meter Submillimeter Telescope at the Arizona Radio Observatory (ARO), He et al. (2008) conducted spectroscopic observations across the 131.2-160.3 GHz band. Concurrent observations using the Submillimeter Telescope (SMT) targeted the 219.5-245.5 GHz and 251.5-267.5 GHz ranges. This comprehensive study detected 377 molecular transitions, including the first detection of ¹³CCH and HN¹³C. Tenenbaum et al. (2010) conducted a line survey of the 214.5-285.5 GHz band using the 10-meter Submillimeter Telescope at the ARO. This study uncovered a significant number of unidentified spectral features, highlighting the critical need for complementary laboratory spectroscopic measurements. Using the Effelsberg 100-meter telescope, Gong et al. (2015) performed a spectroscopic survey across the 17.826.3 GHz frequency range. This observation campaign resulted in the detection of 78 spectral lines, 20 of which represent first-time detections in this source. Zhang et al. (2017) utilized the Tian Ma Radio Telescope to conduct a spectral line survey of the 13.3-18.5 GHz band. This effort successfully identified 35 molecular transitions from 12 distinct chemical species, contributing to the characterization of molecular abundances in the observed region. Zhang et al. (2020) carried out line surveys of a representative sample of CSEs, including IRC+10216, within the 20-25 GHz band using the NRO 45 m telescope and explored the circumstellar chemistry during different stages. These observations were designed to investigate the circumstellar chemistry across different evolutionary stages. Early molecular line surveys exhibited a notable spectral coverage deficit between 90-116 GHz. To fill this observational gap, T24 conducted an 84.5-115.8 GHz survey using the Purple Mountain Observatory’s 13.7 m radio telescope. These observations have been adopted as a fundamental benchmark to validate gas-phase chemical models (Millar et al. 2024). Recent observations from the Atacama Large Millimeter Array (ALMA) have raised higher requirements for the sensitivity of line surveys. Using ALMA, Unnikrishnan et al. (2024) conducted high-sensitivity observations of three carbon-rich AGB stars, IRAS 15194-5115, IRAS 15082-4808, and IRAS 07454-7112, in the frequency range of 85 to 116 GHz. While the authors compared molecular abundances derived from their observations with those documented for IRC+10216 in prior studies, the absence of a systematic spectral line census of IRC+10216 at equivalent sensitivity thresholds within this specific frequency range may compromise the robustness of their comparative analysis.

We present a comprehensive 3 mm spectral line survey of IRC+10216 conducted with the upgraded ARO 12m telescope at Kitt Peak, now equipped with the ALMA prototype receivers that doubled sensitivity compared to pre-2013 configurations. The remainder of this paper is structured as follows. In Section 2, we provide a concise description of the observations and data reduction procedures. Section 3 presents the outcomes of the spectral line identification and rotation diagram analysis. In Section 4, we compare our observational results with those reported by T24 and discuss the implications of our findings for circumstellar chemistry. Finally, Section 5 offers a summary of our work.

2 Observations and data reduction

The observations were carried out during May-June 2015. Our target sample comprises two archetypal carbon-rich evolved stars: IRC+10216 (this work’s primary focus) and CIT 6. The λ = 3 mm line survey of CIT6 has been reported by Yang et al. (2023), and serves as a key reference for our comparative analysis with IRC+10216. The receivers were set to operate in the single sideband (SSB) dual polarization mode. For the lower sideband, the typical image rejection ratio was approximately 20 dB. In contrast, the upper sideband exhibited a lower image rejection ratio. We carried out a comprehensive examination of the image frequencies of strong lines in the image sidebands and identified several image features. The most prominent among them is located at 112820 MHz, which has been misclassified as a “U” line in Yang et al. (2023). In reality, this feature is a result of sideband leakage from the HC₃N J = 11 → 10 line in the image sideband. The spectrometer back ends of the ARO 12m telescope were two 500 MHz 256 channels filter banks (FBs) connected to a millimeter autocorrelator (MAC) with 3072 channels at 195 kHz per channel. The typical system temperatures (T_sys) range from 100 to 250 K, with the value depending on the observed frequency band. Observations were conducted in position switching mode with an azimuth beam throw of ±2″. The schedule of observations was primarily determined by the rise and fall times of IRC+10216 and CIT 6. Efforts were made to complete the observations of both sources within a specific frequency band before switching to another band, to reduce the calibration time and minimize comparison errors with the work of Yang et al. (2023). The spectra cover a frequency range from 90 to 116 GHz, and the corresponding half-power beam width (HPBW) ranges from 41″ to 71″. The HPBW is sufficiently large to accommodate the entire emission region of the objects. The typical on-source time is about 40 minutes for each band. The spectra were calibrated to the $T_{\mathrm{A}}^{\ast }$ scale with a correction for atmospheric attenuation, radiative loss, and rearward and forward scattering. The main beam temperature, T_mb, is derived through the expression of $T_{\mathrm{m}\mathrm{b}}=T_{\mathrm{R}}^{\ast }/{\eta }_{\mathrm{m}}^{\ast }$, where the corrected beam efficiency, n_m, varies with frequency and is about 0.95 in the 3 mm window.

Data reduction was performed using the CLASS software package in GILDAS¹. We identified and removed the bad scans that are seriously affected by bandpass irregularities and then combined the spectra from individual scans. In addition, a low-order polynomial fit was used to subtract the baseline. To improve the signal-to-noise ratio, the spectra were smoothed (combined channels) by a factor of4. The full spectra are shown in Figure 1.

Fig. 1 Full spectrum of IRC+10216 from 90 to 116 GHz with the strong features marked.

2 Results

3.1 Line identifications

Generally, we consider an emission line to be a real detection when its signal-to-noise ratio exceeds 3. However, certain lines are relatively weak, having a signal-to-noise ratio below 3. We consider these lines as valid detections when other transitions of the same molecule have been detected in IRC+10216 or are commonly observed in other evolved stars. The identification of emission lines is based on the Spalatalogue² database, which incorporates the catalogs from the Cologne Database for Molecular Spectroscopy catalogs (CDMS³) (Müller et al. 2005), the Jet Propulsion Laboratory Catalog⁴ (JPL) (Pickett et al. 1998), and Lovas⁵ (Lovas 2004). In this work, we detect a total of 214 emission lines, including four unidentified lines. The lines that have been identified belong to 43 molecules. With the exceptions of SiS and NaCl, all of these molecules contain carbon. Although we discovered no new molecules, we detected 28 molecular lines for the first time in this source. As shown in Figure 1, the CO J = 1 → 0 transition exhibits the highest intensity, followed by the CN J = 1 → 0 transition. The zoomedin spectra, smoothed using a four-channel boxcar filter, are presented in Figure 2. The line profiles of most detected lines are fitted using a stellar-shell model. The fitting results for each line, including velocity-integrated intensity, line width, and line center temperature, are listed in Table A.1. For lines with blended hyperfine structures, only the total velocity-integrated intensity and the intrinsic strength ratio of each line are presented. In what follows, a brief description of the detected molecules is provided.

CO. The J = 1 → 0 transitions of CO, ¹³CO, C¹⁷O, and C¹⁸O are detected, with the CO J = 1 → 0 line being the strongest. These observations probe gas components with varying optical depths, thereby uncovering distinct kinematic characteristics. The first detection of the CO J = 1 → 0 line in IRC+10216 was made by Solomon et al. (1971) using the 36-foot antenna of the National Radio Astronomy Observatory. As shown in Figure 3, the line profiles of the J = 1 → 0 transitions of CO and ¹³CO are parabolic and double-peaked respectively, which correspond to the optically thick and thin scenarios. For the J = 1 → 0 transitions of C¹⁷O and C¹⁸O, their profiles cannot be unambiguously determined because of blending with other lines. Nevertheless, it appears that they possess multiple spectral components, which is consistent with limb-brightened outflows. At 104.71 GHz, a weak spectral feature is observed, which precisely coincides with the frequency of the ¹³C¹⁸O J = 1 → 0 transition line. Nevertheless, if we assume that the intensity ratio of the C¹⁸O J = 1 → 0 to the ¹³C¹⁸O J = 1 → 0 lines follows the ¹²C/¹³C abundance ratio (45, Cernicharo et al. 2000), the integrated intensity of the ¹³C¹⁸O J = 1 → 0 line is expected to be 13mK kms^-1, which is approximately an order of magnitude lower than the value obtained from our measurements. Given this significant discrepancy, we conclude that the identification of this feature as the ¹³C¹⁸O J = 1 → 0 should be considered tentative, pending further confirmation.

C_nS. We detected the J = 2 → 1 transition of CS, six rotational transitions of C₂S, five rotational transitions of C₃ S, and a weak vibrationally excited transition in the ν = 1 state of CS J = 2 → 1. We also detected the isotopologues of CS, including ¹³CS, C³³S, and C³⁴S. As shown in Figure C.1, the line profile of CS J = 2 → 1 shows a parabolic shape with a slightly asymmetric wing, where the red-shifted side is brighter. The line profile of C₃₄S J = 2 → 1 shows a double-peaked shape. The C₂S (N, J) = (7,8) → (6,7) line is partly blended with a C₄H line, as shown in Figure C.2. The remaining lines present slight double-peaked profiles. High spatial resolution maps of IRC+10216 reveal that the CS emission is distinctly clumpy, with its substructures presenting in a pattern resembling concentric shells or circular arcs (Velilla-Prieto et al. 2019).

SiS. We detected a total of five emission lines of SiS and its isotopologues. The SiS J = 5 → 4 transition exhibits a distinct double-peaked morphology characterized by an asymmetrical velocity wing demonstrating enhanced emission on the red-shifted side, as clearly depicted in Figure C.3. In contrast, the J = 6 → 5 line displays a smooth parabolic-shaped profile. This contrasting behavior between adjacent rotational transitions aligns with similar spectral discrepancies previously reported by Velilla-Prieto et al. (2019).

SiC₂. We detected seventeen emission lines of SiC2 and its ¹³C isotopologues. Among these, we report on three lines for the first time in IRC+10216, while the rest were previously documented in earlier observations (Cernicharo et al. 1986a, 2018). As depicted in Figure C.4, the line profiles of SiC₂ closely resemble those of SiS, characterized by a double-peaked structure featuring a brighter red wing. These lines exhibit a relatively narrow linewidth, with the local standard rest velocity (u_LSR) distributed within the range of -40 to -10 km s^-1. Interferometric observations have revealed that the spatial distribution of SiC₂ takes the form of a shell encircling the central star. However, this shell is far from uniform; rather, it is clumpy and nonspher-ical, as previously reported in the literature (Gensheimer et al. 1992; Takano et al. 1992; Gensheimer et al. 1995). The nonuniform nature of the SiC₂ shell implies complex physical processes at play in the circumstellar environment, potentially involving inhomogeneous mass loss from the central star, or interactions between the circumstellar gas and the surrounding interstellar medium.

SiC₄. There are eight SiC₄ transitions within the frequency range of the current observations. Six of these, from J = 30 → 29 to 35 → 34, are detected with a peak temperature of about 10 mK. As shown in Figure C.5, the J = 30 → 29 and 33 → 32 are unresolved and blended with the emission lines of ²⁶MgNC and C₆H, respectively. The SiC₄ J = 36 → 35 and 37 → 36 lines are too weak to be detected.

H₂C₃ and H₂C₄. We detected three emission lines of H₂C₃ and, as depicted in Figure C.6, two of these lines are blended with one another. The H₂C₃ 5(1,4) → 4(1,3) line exhibits a doublepeaked profile. H₂C₃ is an isomer of c-H₂C₃, a molecule that has been widely detected in the interstellar medium. The first potential detection of c-H₂C₃ was reported by Cernicharo et al. (1991a). However, in our study, we did not observe any emission lines associated with c-H₂C₃. On the other hand, we detected nine spectral lines of H₂C₄. Among these, five had been previously reported in earlier observations (Cernicharo et al. 1991b; Pardo et al. 2018). As depicted in Figure C.7, the line profiles of H₂C₄ exhibit both double-peaked and flat-topped shapes.

I-C₃H. The spectral analysis reveals complex blending in the I-C₃H transitions due to unresolved adjacent hyperfine components. We detected eight hyperfine lines of I-C₃H, two of which blended together and remained unresolved as evidenced by asymmetric profiles in Figure C.8. The double-peaked line profiles indicate optically thin emission from an expanding circumstellar envelope. Observations of these lines have been previously reported (Thaddeus et al. 1985; Cernicharo et al. 1986b). Comparative analysis with its cyclic isomer c-C₃H reveals significant chemical differentiation processes (Sheehan et al. 2008). Although many transition lines of c-C₃ H lie within our observed frequency range, no lines are detected. Therefore, we conclude that the abundance ofc-C₃H in IRC+10216 is substantially lower than that of its linear isomer. This may arise from the fact that linear chains preferentially form through radical-radical recombination in the outer CSE, while cyclic isomers require a higher density environment that is unavailable in IRC+10216’s extended envelope.

C₄H. Our observations revealed 27 rotational transitions of C₄H. Each rotational transition resolves into two hyperfine components with a similar intensity. Ground-state transitions display double-peaked profiles with slight red-wing enhancement, as shown in Figure C.9. The C₄H v₇ =2⁰, N = 10 → 9, J = 19/2 → 17/2 line is badly blended with the C₆H²Π₃/₂J = 69/2 → 67/2 f line. Spatially resolved ALMA observations (Agúndez et al. 2017) constrain the C₄H distribution to a hollow spherical shell.

C₅H. Our observations detected 20 rotational transitions of the C₅H radical, including two newly detected transitions. The remaining lines are consistent with previous detections (Cernicharo et al. 1986b; Ziurys et al. 1995; Pardo et al. 2018). As shown in Figure C.10, each line splits into two unresolved blended hyperfine structures.

C₆H. Among all identified molecules, C₆H exhibits the highest number of spectral lines, totaling 33. The C₆H₂Π₁/₂, J = 75/2-73/2e, f transition, which falls within our observed frequency range but is blended with C₄H lines, could not be conclusively detected. The ₂Π₁/₂, J = 81/2 → 79/2e and ²Π₁/₂, J = 81/2 → 79/2f transitions represent new detections, while the remaining lines have been previously reported (Guelin et al. 1987; Cernicharo et al. 1987; Ziurys et al. 1995; Agúndez et al. 2010). Similar to C₄H, each rotational transition of C₆H undergoes splitting into two Λ doubling components. As illustrated in Figure C.11, these lines display distinct double-peaked profiles, which suggests that C₆H is likely to be optically thin.

CN. The significant abundance of CN observed in CSE and the interstellar medium arises from its efficient formation via HCN photodissociation and resilience in UV-irradiated environments. Its complex hyperfine structure arises from the spinrotation interaction and the hyperfine structure due to the nuclear spin of nitrogen (I = 1), which produces nine measurable hyperfine structure components in the N = 1 → 0 transition. Our observations detected all nine components, as shown in Figure C.12. The rarer isotopologue ¹³CN shows more intricate splitting, with 14 hyperfine components in its N = 1 → 0 transition (113.488 GHz), which are categorized into three distinct groups. This substructure emerges from combined ¹³C nuclear spin (I = 1/2) and ¹⁴N quadrupole moment (I = 1) interactions, creating additional splitting compared to the main isotopologue. Previous observations by T24 detected all the CN lines but failed to observe ¹³CN.

C₃N and C₅N. Our observations revealed four ground-state rotational transitions and one ν₅ = 1 vibrationally excited state transition of C₃N. The ground-state transitions exhibit characteristic double-peaked profiles with enhanced red-wing emission (Figure C.13), which is consistent with optically thin radiation from an expanding circumstellar envelope. While these groundstate lines were previously observed by T24, we report the first detection of the ν₅ = 1 excited state transition, which indicates non-local thermodynamic equilibrium (non-LTE) excitation in the inner envelope. C₅N was first detected in IRC+10216 by Guelin et al. (1998). Subsequent observations by Pardo et al. (2022) identified the N = 14 → 13 and 15 → 14 transitions. In this work, we detected five new hyperfine components of the N = 40 → 39, J = 79/2 → 77/2 transition. These hyperfine structures show severe blending that prevents reliable profile fitting.

HC₃N and HC₅N. We detected nine emission lines of HC₃N, consisting of three ground-state lines and six v₇ = 1 vibrationally excited lines. As depicted in Figure C.14, the parabolic profiles of the ground-state lines indicate that the HC₃N lines are optically thick. For the lines of HCC¹³CN and HC¹³CCN, which are from the same energy level, a small frequency difference leads to partial blending. As shown in Figure C.15, we detected ten ground-state transitions of HC₅N, ranging from J = 34 → 33 to J = 43 → 42. Their flat-topped line profiles suggest an optically thin and spatially unresolved scenario.

MgNC and MgCN. We detected four MgNC lines, one ²⁵MgNC emission line, and two ²⁶MgNC lines. All of these lines had been previously reported in studies (Guelin et al. 1995). As depicted in Figure C.16, the line profiles of MgNC (N, J) = (8,15/2) → (7,13/2) and (8,17/2) → (7,15/2) exhibit a double-peaked shape. The (9,17/2) → (8,15/2) and (9,19/2) → (8,17/2) lines are blended with each other, making it impossible to determine their individual profiles. The rotational transitions of MgCN (N = 9 → 8,10 → 9, and 11 → 10) at 91.7, 101.9, and 112.07 GHz, respectively, fall within our observational frequency range. Among these, the N = 10 → 9 is distinctly detected in the blue wing of a C₆H line. However, the N = 9 → 8 and 11 → 10 transitions are heavily obscured by noise, rendering them undetectable in our data. This molecular species was first identified in IRC+10216 through 3 mm wavelength observations by Ziurys et al. (1995). A direct comparison of our spectra with their published results demonstrates remarkable consistency in the detected line profiles and intensities, thereby corroborating the presence of MgCN in this object.

NaCN and NaCl. We detected two sodium-containing species, NaCN and NaCl, toward IRC+10216. The NaCl 7 → 6 and 8 → 7 lines fall within our frequency coverage and are both detected, as depicted in Figure C.17. In total, eight NaCN lines are detected, as shown in Figure C.18, and all of them have been previously reported (Turner et al. 1994; Agúndez et al. 2012; Cernicharo et al. 2019).

CH₃CN. We detected seven rotational transitions of CH₃ CN, all of which had been previously reported (Agúndez et al. 2014; Quintana-Lacaci et al. 2017). Each J + 1 → J rotational transition undergoes centrifugal distortion, splitting into J hyperfine components that become spectrally blended (Figure C.19), which prevents individual line profile analysis. ALMA mapping observations toward IRC+10216 revealed that the majority of CH₃CN is distributed in a hollow shell approximately 2″ from the star, with a central void that has a radius of up to 1″ (Agúndez et al. 2015). This distribution contradicts the standard chemical model of IRC+10216 and significantly differs from that of other molecules. The observed discrepancy may be attributed to the clumpy structure of the envelope, which enables UV photons to penetrate the inner regions.

Unidentified lines. We detected four emission lines that could not be correlated with any potentially detectable transitions listed in the spectroscopic catalog. In Table A.1, these lines are marked with “U”. The U line at 98.05 GHz coincides with CH₃₃CH₂CN 11_5,7 → 10_5,6 observed in Orion KL (Demyk et al. 2007). However, we exclude the molecule as the carriers based on the non-detection of it’s numerous predicted transitions within our spectral coverage. The U line at 112.04 GHz was also detected by Ziurys et al. (1995). These U lines likely originate from novel circumstellar molecules, known moelcules with complex internal degrees of freedom (e.g., internal rotation in HC_2n+1 N polymers), among other possibilities. Future experimental investigations are crucial to characterize the nature of these unassigned U lines.

Fig. 2 Zoom-in spectra of IRC+10216 with all detected lines labeled. To provide a clearer visualization, the spectra have been smoothed to a resolution of 3 MHz. “SL” denotes that the spectral feature arises from sideband leakage rather than a genuine detection. The complete set of spectra is shown in Figure B.1.

Fig. 3 Line profiles of CO and its isotopologues. The red and green curves represent the multiple fitting profiles obtained through the application of the stellar-shell model. The vertical solid lines indicate the wavelengths and relative intensities of the hyperfine components. Note that the identification of the 13C18O J = 1 → 0 line is tentative. The complete figure set is shown in Figures C.1-C.19.

3.2 Rotational temperatures and column densities

The rotational diagram method is used to calculate the excitation temperatures (T_ex) and column densities (N) of the molecules detected in our observations. We assume that the emission lines are optically thin and that the level populations of the molecules are in LTE. The expression of rotation diagram is $$\mathrm{l}\mathrm{n}\left(\frac{3k\int T_{\mathrm{s}}\mathrm{d}\mathrm{v}}{8{\pi }^3\nu {\mu }^{2}S}\right)=\mathrm{l}\mathrm{n}\left(\frac{N_{\mathrm{u}}}{g_{\mathrm{u}}}\right)=\mathrm{l}\mathrm{n}\left(\frac{N}{Q}\right)-\frac{E_{\mathrm{u}}}{k{T}_{\mathrm{e}\mathrm{x}}},$$(1)

where ∫T_s dv represents the integration of the source brightness temperature (T_s) over the velocity, ν is the rest frequency of each transitions, μ is the permanent dipole moment, S is the transition moment, g_u is the degeneracy of the upper state, E_u is the upper level energy of the transition, and Q is the partition function that varies as a function of excitation temperature. If multiple distinct transitions of a given molecule are detected, a straight line is utilized to fit them by means of Equation (1), where $\ln \frac{N_{\mathrm{u}}}{g_{\mathrm{u}}}$ represents the ordinate and $\frac{E_{\mathrm{u}}}{k}$ serves as the abscissa. We can determine T_ex and N from the slope and intercept of the fitting line. To account for beam dilution effects in our single-dish observations, we applied the standard correction formula $T_{\mathrm{s}}=T_R({\theta }_{\mathrm{s}}^2+{\theta }_{\mathrm{b}}^2)/{\theta }_{\mathrm{s}}^2$, where θ_b, (55″-70″) represents the frequencydependent HPBW. Although molecular emission regions exhibit θ_s variations (T24), we adopted a uniform value, θ_s = 30″. For optically thin transitions, rotational diagrams constructed via Equation (1) show enhanced reliability with increasing numbers of detected transitions. Deviations from linearity may arise from non-LTE excitation processes (e.g., radiative pumping in vibrationally excited states), moderate optical depth effects, line blending, or molecular misassignments.

Figure D.1 illustrates the rotational diagrams of 16 molecules detected in our observations. While rotational diagrams for most species in our sample are well fitted by a single straight line, C₆H, HC₃N, and HC₅N exhibit significant deviations requiring dual excitation temperature components. The spatial thermal structures of these molecules are likely to be stratified. Among the detected molecules, the rotational diagram of HC₅N shows the most distinct thermal stratification. As shown in Figure D.1, based on our observations, the rotational temperature of HC₅N is 18.7 K, with higher-J lines tracing hotter regions (Cernicharo et al. 2000), while lower-J transitions probe colder regions (Pardo et al. 2022). The derived T_ex and N of 16 molecules are listed in Table A.2.

CN ranks among the most abundant molecules in IRC+10216, with an abundance approximately five times lower than that of HCN (Dayal & Bieging 1995). Our observations reveal two fine-structure components of the CN (N = 1-0) transition at approximately 113.1 and 113.5 GHz, which allows us to derive their optical depths. The intrinsic intensity ratio of these two components is two. However, a statistical survey of 42 carbon-rich stars by Bachiller et al. (1997) reported an average ratio of 1.3, which suggests significant optical thickness effects. By defining the intensity ratio between the stronger and weaker components as R, the optical depth of the stronger component (τ_s) can be determined using the relation $R=\frac{1-\mathrm{e}\mathrm{x}\mathrm{p}(-{\tau }_{\mathrm{s}})}{1-\mathrm{e}\mathrm{x}\mathrm{p}(-{\tau }_{\mathrm{s}}/2)}$. For IRC+10216, our analysis yields an optical depth of τ_s = 0.55 for the stronger component.

Fig. 4 Number of detected spectral lines above a given ∫TRdv value. The maximum ∫TRdv value has been normalized to 1 (left panel). Comparison of the line intensities of molecules detected in this work and T24. The dashed line represents x = y (right panel).

3.3 Fractional abundances relative to H₂

Assuming that the CSE is a spherical shell formed by a constant mass-loss rate, the molecular emission is optically thin, and a uniform T_ex, we calculated the fractional abundances of the molecules relative to H₂ (f) using the formula given by Olofsson (1997), $$f_{\mathrm{X}}=1.7\times {10}^{-28}\frac{v_e{\theta }_{\mathrm{b}}D}{{\dot{M}}_{{\mathrm{H}}_2}}\frac{Q(T_{\mathrm{e}\mathrm{x}}){\nu }^2}{g_{\mathrm{u}}{A}_{\mathrm{u}\mathrm{l}}}\frac{e^{E_{\mathrm{l}}/k{T}_{\mathrm{e}\mathrm{x}}}\int T_R\mathrm{d}\mathrm{v}}{{\int }_{x_i}^{x_e}{e}^{-4x^{2}ln2}\mathrm{d}\mathrm{x}},$$(2)

where v_e is the expansion velocity of the CSE, with the measured line width taken as the value, in unit of km s^-1 , D is the distance in pc (120 pc; Groenewegen et al. 1998), M_H2 is the mass-loss rate in M_⊙ yr^-1, g_u is the statistical weight of the upper level, A_ul is the Einstein coefficient for the transition, and x_i,e is calculated by using the formula x_i,e = R_i,e/(Dθ_b), where R_i and R_e are the inner and outer radii of the shell obtained by Woods et al. (2003). The mass-loss rate is obtained via the CO J = 1 → 0 line using the prescription from Winters et al. (2002), with a CO abundance assumption of f_CO = 1 × 10^-3. We obtain M_H2 = 6.7 × 10^-6. The calculated molecular abundances are presented in Table A.2. We compared our results with those of previous studies by He et al. (2008), Gong et al. (2015), Zhang et al. (2017), and Pardo et al. (2022). Our findings show a good agreement with the earlier results.

4 Discussion

4.1 Comparison with T24

The comparative analysis between our 3 mm spectral survey and the prior work of T24 (Figure 4) reveals both consistency and advancement in probing IRC+10216’s molecular inventory.

Within the overlapping 90-115.8 GHz spectral range, both surveys demonstrate robust agreement for bright emission lines. We compared the integrated intensities of the common emission lines detected in our observations and those in T24, as depicted in Figure 4. The data have been corrected for the impact of beam dilution. Overall, the ratios of line intensities between our results and those of T24 are close to unity. This indicates a high degree of consistency in the intensities of the emission lines between the two datasets. This mutual validation strengthens confidence in line identifications for 57 previously cataloged transitions. Our observations achieve a seven-fold sensitivity improvement. This enables the detection of 157 additional weak lines.

However, a more detailed analysis reveals variations in the beam-dilution corrected intensity ratios of individual molecules between T24 and this work [I_int (Tuo et al. 2024) versus I_int (this paper)], as shown in Figure E.1. During beam dilution corrections, an identical source size was assumed for both datasets. Given the larger telescope beam of our observations, our measurements preferentially sample more extended spatial regions compared to T24. Consequently, if species exhibit differing spatial distributions, the I_int (Tuo et al. 2024)/I_int (this paper) ratios may deviate from unity.

4.2 Isotopic ratio

The central star of IRC+10216 is thought to have evolved to the late AGB phase (Cernicharo et al. 2000), where the third dredge-up processes have transported nucleosynthetic products from the stellar interior to the surface through convective mixing, subsequently expelling them into the CSE via stellar winds. Isotopic abundance ratios of molecular species in the CSE serve as critical tracers of these internal nucleosynthetic and mixing processes (Herwig 2005). Based on observed molecular transition lines, we derived the carbon and silicon isotopic ratios in IRC+10216. The results are listed in Table 1 alongside values for the Sun, CIT 6, and CRL 3068.

Carbon isotopic ratios prove particularly sensitive to stellar processing. The third dredge-up typically enhances surface ¹² C/¹³ C ratios, while cool bottom processing (CBP) in low-mass red giants (M < 2 M_⊙) counteracts this trend through extra mixing (Charbonnel 1995; Charbonnel et al. 1998). We derived the ¹²C/¹³C ratio through five ¹³C-bearing species (¹³CO, ¹³CS, Si¹³CC, H¹³CCCN, HC¹³CCN), employing line intensity ratios as shown in Figure 5. Due to optical thickness in CO and CS lines, the derived ¹²CO/¹³CO and ¹²CS/¹³CS ratios represent lower limits. The derived ¹²C/¹³C ratio is consistent with that reported by Cernicharo et al. (2000), while showing substantial depletion relative to solar (~89) and Orion Bar (~67) values (Keene et al. 1998). This depletion supports CBP-induced mixing in IRC+10216’s progenitor.

The ¹⁷O/¹⁸O ratio in AGB envelopes can be significantly enhanced through hot bottom burning (HBB) nucleosynthesis. This process occurs when the base of the convective envelope reaches temperatures exceeding 10⁷ K, which enables proton-capture reactions that preferentially produce ¹⁷O via ¹⁶O(p, γ)¹⁷F(β⁺ v)¹⁷O while depleting ¹⁸O through ¹⁸O(p, α)¹⁵N. Using the integrated intensity ratio of C¹⁷O J = 1 → 0 to C¹⁸O J = 1 → 0 rotational transitions, combined with the isotopic abundance calculation framework established by De Nutte et al. (2017), we derive ¹⁷O/¹⁸O = 1.51 ± 0.16 for IRC +10216. This measurement shows reasonable consistency with the value of 1.19 ± 0.05 reported in De Nutte et al. (2017), yet demonstrates a striking contrast to the solar value of 0.19 ± 0.02. The observed ¹⁷O/¹⁸O enhancement provides direct observational evidence for the operation of HBB in this object.

Silicon isotopic ratios remain largely unaffected by AGB nucleosynthesis. Our detection of four silicon isotopologues (²⁹SiS, ³⁰SiS, ²⁹SiCC, ³⁰SiCC; Figure F.1) reveals silicon isotopic ratios consistent with solar system abundances. The low ^29,30Si abundances minimize optical depth effects, ensuring robust determination of primordial silicon isotope ratios inherited from the interstellar medium.

Fig. 5 12C/13C isotopic ratios derived from CO, CS, SiCC, and HC3N. The ordinate represents the line intensity ratio derived from two distinct isotopologues. The abscissa denotes the upper-level energy of the transitions.

4.3 Comparison with CIT 6

CIT 6 is the second most infrared-bright carbon-rich star, immediately following IRC+10216. To investigate whether IRC+10216 represents a typical carbon-rich AGB star, we compared its 3 mm spectrum with that observed for CIT 6, which was acquired under identical instrumental setup (Yang et al. 2023). Every emission line detected in CIT 6 was also observed in IRC+10216, which suggests a similarity in their underlying physical and chemical processes. The beam-dilution corrected integrated line intensity ratios between the two sources, I_int(CIT 6)/I_int(IRC+10216), are presented in Figure 6, which can provide insights into differences in molecular abundances, excitation conditions, or the physical structure of their circumstellar environments. The mean value of log[I_int(CIT6)/I_int(IRC+10216)] is about -0.95, with all emission lines displaying values distributed around this mean. The ratio of distances between CIT 6 and IRC+10216 is approximately 3. Given the inverse-square relationship between intensity and distance, the observed differences in line intensities can be attributed to their varying distances, with IRC+10216 being located at a closer proximity, rather than intrinsic source properties.

Although the principal variations in line intensities are mainly attributed to the difference in distance between these sources, a careful scrutiny of Figure 6 discloses subtle disparities between the two. The intensity ratios of the SiS, C₄H, and C₃N lines are beneath the mean value, while those of SiC₂, CN, and HC₅N are higher. These discrepancies could serve as tracers of CSE evolution, specifically probing the chemical processing occurring within these environments. SiC₂, as a daughter molecule of SiS (Feng et al. 2023), could experience an increasing SiC₂/SiS abundance ratio over the course of its evolution.

Cyanopolyynes are hypothesized to be synthesized through a bottom-up process similar to that of polyynes in circumstellar environments. Consider the following potential reactions for the formation of HC₅N: $${\mathrm{C}}_2\mathrm{H}+\mathrm{H}{\mathrm{C}}_3\mathrm{N}⟶\mathrm{H}{\mathrm{C}}_5\mathrm{N}+\mathrm{H},$$(3) $$\mathrm{C}\mathrm{N}+{\mathrm{C}}_4{\mathrm{H}}_2⟶\mathrm{H}{\mathrm{C}}_5\mathrm{N}+\mathrm{H},$$(4)

and $${\mathrm{C}}_3\mathrm{N}+{\mathrm{C}}_2{\mathrm{H}}_2⟶\mathrm{H}{\mathrm{C}}_5\mathrm{N}+\mathrm{H}.$$(5)

The activation barrier of the reaction described in Equation 3 is approximately 800 K. As a result, this reaction has a relatively low rate in the environment of IRC+10216 (Agúndez et al. 2017). Although chemical equilibrium calculations indicate that both Equations (4) and (5) proceed rapidly at low temperatures and are considered equally significant for the formation of HC₅N, a comparison between theoretical calculations and observational data reveals that the reaction in Equation (5) predominantly occurs in the inner regions of the CSE, where the abundance of C₃N is relatively low. Consequently, the formation of HC₅N is mainly attributed to the reaction described in Equation (4). Moreover, the formation of HC₃N takes place closer to the inner regions of the CSE and at an earlier stage compared to HC₅N. As depicted in Figure 6, we observe that the abundance of HC₅N in CIT 6 is obviously higher than that in IRC+10216, while the abundances of HC₃N in both objects are nearly identical.

Therefore, we conclude that the higher abundances of SiC₂ and HC₅N in CIT 6 imply a more advanced evolutionary stage of CIT 6 relative to IRC+10216. This finding is in accordance with the results of previous studies (Zhang et al. 2009a; Yang et al. 2023). Additionally, we note that the line intensities of C₄H and C₃N in IRC+10216 are anomalously high compared to those in CIT 6. As proposed by Agúndez et al. (2017), HC₃N is considered a first-generation daughter molecule in the C₂H₂ reaction chain, while C₄H and C₃N are second-generation daughter molecules. Consequently, HC₃N is expected to form earlier. Nevertheless, the results presented in Figure 6 seem to show the opposite trend. This discrepancy indicates that other factors likely influence the abundances of C₄H and C₃N. Given that the formation of both C₄H andC₃Nis driven by UV photons, we infer that the UV-induced reactions are more pronounced in IRC+10216. Some studies (e.g., Cernicharo et al. 2015; Siebert et al. 2022) have suggested the presence of a companion star within IRC+10216. The UV radiation from the hot companion star would enhance the overall UV radiation field in the envelope, which would potentially explain the abundances patterns of C₄H and HC₃N. However, this conclusion remains tentative and requires further observational confirmation in the future.

Fig. 6 Integrated intensity ratios of the λ = 3 mm molecular lines between IRC+10216 and CIT 6. The dashed line represents the mean value.

5 Summary

This paper presents a high-sensitivity spectral line survey of the carbon-rich AGB star IRC+10216, carried out within the frequency range of 90-116 GHz. In total, 214 emission lines were detected, among which 28 are newly reported. We identified 211 spectral lines corresponding to 43 distinct molecules. However, four “U” lines could not be assigned to any known molecule. With the exception of SiS and NaCl, all detected molecules are carbon-bearing species. C₆H exhibits the largest number of emission lines, with 32 lines detected, followed by C₄H with 27 lines. For molecules with optically thin emission lines, we employed the rotational diagram method to calculate their excitation temperatures, column densities, and fractional abundances. Based on the line intensity ratios, we determined the isotopic ratios of carbon, oxygen, and silicon elements. Our findings indicate that the ¹²C/¹³C ratio in IRC+10216 is characteristic of evolved stars and is significantly lower than the solar value.

Through a comparison of the overlapping spectra between our observations and those of T24, we determined a high degree of consistency in the detection of stronger lines. Specifically, every line identified by T24 was also detected in our study. Moreover, our observations revealed a multitude of weaker lines that evaded detection in the work of T24, underscoring the enhanced sensitivity of our observational setup. Regarding the line intensity ratios between the two observational datasets, they are found to be close to unity. The moderate deviations from unity can be ascribed to variations in the spatial extents of different molecular species. Our high-sensitivity spectra establish a robust spectral reference framework for comparative analysis of ALMA observations toward carbon-rich evolved stars.

Our systematic comparison of IRC+10216 and CIT 6 reveals critical insights into chemical evolution during the AGB phase. While both sources exhibit remarkably similar molecular inventories, with line intensity differences primarily attributable to their different distances, we have identified crucial evolutionary indicators. The relatively weaker SiC₂ and HC₅N line intensities in IRC+10216 suggest advanced evolutionary processing. Conversely, the enhanced C₄H and C₃N intensities in IRC+10216 defy standard chemical evolution models. Their photodissociation-driven formation pathways implicate a stronger UV radiation field potentially generated by a binary system.

Data availability

The appendices are available on Zenodo at https://zenodo.org/records/15739337

Acknowledgements

We sincerely express our gratitude to an anonymous reviewer for his/her valuable comments, which have significantly enhanced the quality of this paper. The financial supports of this work are from the National Natural Science Foundation of China (NSFC, No. 12473027, 12333005, 11973099, and 12463006), the Guangdong Basic and Applied Basic Research Funding (No. 2024A1515010798), the Greater Bay Area Branch of the National Astronomical Data Center (No. 2024B1212080003), and the science research grants from the China Manned Space Project (NO. CMS-CSST-2021-A09, CMS-CSST-2021-A10, etc). Y.Z. thanks the Xinjiang Tianchi Talent Program (2023). X.H.L. acknowledges support from the Natural Science Foundation of Xinjiang Uygur Autonomous Region (No. 2024D01E37) and the National Science Foundation of China (12473025).

References

All Tables

All Figures

	Fig. 1 Full spectrum of IRC+10216 from 90 to 116 GHz with the strong features marked.
In the text

	Fig. 2 Zoom-in spectra of IRC+10216 with all detected lines labeled. To provide a clearer visualization, the spectra have been smoothed to a resolution of 3 MHz. “SL” denotes that the spectral feature arises from sideband leakage rather than a genuine detection. The complete set of spectra is shown in Figure B.1.
In the text

Fig. 3 Line profiles of CO and its isotopologues. The red and green curves represent the multiple fitting profiles obtained through the application of the stellar-shell model. The vertical solid lines indicate the wavelengths and relative intensities of the hyperfine components. Note that the identification of the 13C18O J = 1 → 0 line is tentative. The complete figure set is shown in Figures C.1-C.19.

In the text

	Fig. 4 Number of detected spectral lines above a given ∫TRdv value. The maximum ∫TRdv value has been normalized to 1 (left panel). Comparison of the line intensities of molecules detected in this work and T24. The dashed line represents x = y (right panel).
In the text

	Fig. 5 12C/13C isotopic ratios derived from CO, CS, SiCC, and HC3N. The ordinate represents the line intensity ratio derived from two distinct isotopologues. The abscissa denotes the upper-level energy of the transitions.
In the text

	Fig. 6 Integrated intensity ratios of the λ = 3 mm molecular lines between IRC+10216 and CIT 6. The dashed line represents the mean value.
In the text