Carbon Chemistry in IRC+10216: Infrared Detection of Diacetylene

Evolved stars are known to develop a circumstellar envelope surrounding their central objects. Around a third of the total number of molecules discovered in space are present in the envelope of this kind of star (e.g., Guélin et al. 1978, 1987; Hinkle et al. 1988; Bernath et al. 1989; Ohishi et al. 1989; Bell et al. 1993; Cernicharo & Guélin 1996; Cernicharo et al. 2000, 2015; Agúndez et al. 2014a, 2014b; Anderson & Ziurys 2014). Most of these molecules are formed in the outer shells of the envelopes due to the active chemistry triggered by the radicals and ions that arise after the dissociation of neutral molecules by the external ultraviolet (UV) radiation field (Millar & Herbst 1994; Millar et al. 2000; Cernicharo 2004; Agúndez et al. 2017). In particular, the abundances of polyynes (C_2nH₂) and cyanopolyynes (HC_2n+1N) that have been observed only in the C-rich proto-planetary nebula CRL618 to date can be explained by a photochemical model in which these molecules are formed by chemical reactions involving the radicals ${{\rm{C}}}_{2n}{\rm{H}}$ and ${{\rm{C}}}_{2n+1}{\rm{N}}$, giving raise to a polymerization mechanism that produces carbon-chain molecules (Woods et al. 2003; Cernicharo 2004). The abundance of these molecules decreases as their number of atoms increases, strongly depending on the temperature and the density of the gas (Cernicharo 2004).

To date, several members of the cyanopolyyne family have been detected in several evolved stars, including the very well known Asymptotic Giant Branch star IRC+10216 (HC_2n+1N, n = 0, ..., 4; Winnewisswer & Walmsley 1978; Henkel et al. 1985; Matthews et al. 1985; Guélin & Cernicharo 1991). Regarding the polyyne family, no member has been detected so far in this source, apart from C₂H₂, which was observed with a column density of ∼10¹⁹ cm⁻² (Cernicharo et al. 1999; Fonfría et al. 2008). This non-detection suggests low column densities for other polyynes (≲10¹⁶ cm⁻²), contrary to what happens in CRL618. Cernicharo et al. (2001) and Fonfría et al. (2011) clearly detected the C₄H₂ and C₆H₂ features produced in the photodissociation shells of the envelope of this proto-planetary nebula, with column densities ∼10¹⁷ cm⁻², similar to that of C₂H₂. This remarkable difference is an effect of the gas density in the photochemical evolution of the envelopes of evolved stars and the photopolimerization of C₂H₂ and HCN (Cernicharo 2004).

In this paper, we present the first detection of the spectrum of C₄H₂ toward the C-rich AGB star IRC+10216. The observations are described in Section 2. Section 3 contains the results of the data analysis. A discussion about them and their implications in the current chemical scenario of IRC+10216 can be found in Section 4. A brief summary of our work and the final conclusions are in Section 5.

The observations were carried out with the Texas Echelon-cross-Echelle Spectrograph (TEXES, Lacy et al. 2002) mounted on the 3 m Infrared Telescope Facility (IRTF) on 2008 May. TEXES was used in its High_Medium mode with a resolving power of R ≃ 85,000, which provides us with a spectral resolution of ≃3–4 km s⁻¹. IRC+10216 was nodded along the slit to allow for a better sky subtraction and an efficient on-source integration. The observations were corrected from the atmosphere with a blackbody-sky difference. The data were reduced with the standard TEXES pipeline. The baseline of each order was removed with an up to 10th order polynomial fit, taking care of excluding all the features in the spectrum in this process. The total spectrum was composed of 150 different segments that almost completely cover the spectral range of 7.9–9.1 μm. The part of the spectrum where the C₄H₂ lines are found roughly ranges from 8.0 to 8.1 μm (Figure 1). The noise rms is estimated to be ≃0.2% of the continuum.

Zoom In Zoom Out Reset image size

The spectrum was not corrected from telluric contamination since IRC+10216 is much brighter at 8 μm than any available calibrator. The telluric feature identification has been performed by comparing the observations with the Atmospheric TRANsmission (ATRAN)⁵ model (Lord 1992). The identification of the features coming from IRC+10216 has been performed with the aid of the data in the last version of the HIgh-resolution TRANsmission molecular absorption Database (HITRAN,⁶ Rothman et al. 2013). The Doppler shift affecting these lines was accurately removed by fitting the strong CS and SiO lines in the observed data. The analysis of the spectra of these molecules will be published elsewhere.

C₄H₂ is a molecule with five stretching modes and four doubly-degenerated bending modes. Only the fundamental modes ${\nu }_{4}({\sigma }_{u}^{+})$, ${\nu }_{5}({\sigma }_{u}^{+})$ (stretching bands at ≃3.0 and 5.0 μm), and the symmetric C ≡ C−H and C−C ≡ C bending modes ν₈(π_u) and ν₉(π_u) (bending bands at ≃15.9 and 45 μm) are infrared active, i.e., the rotational levels in the corresponding excited vibrational states are radiatively connected in the infrared with those of the vibrational ground state by electric dipole transitions. The anti-symmetric C ≡ C−H and C−C ≡ C bending modes ν₆(π_g) and ν₇(π_g) are inactive in the infrared. Vibrational modes ν₆ and ν₈ play the same role in C₄H₂ that ν₄ and ν₅ play in C₂H₂. The infrared spectrum of C₄H₂ comprises additional combination bands, such as ${\nu }_{6}+{\nu }_{8}({\sigma }_{u}^{+})$. The strongest of all these bands are ν₄ and ν₈, followed by ν₆ + ν₈ (Khlifi et al. 1995). ν₈ is blocked by the strong telluric ozone band around 15 μm and is unavailable from the ground (but was detected from space by Cernicharo et al. 2001). ν₄ is highly overlapped with strong bands of C₂H₂ and HCN, which are very abundant in IRC+10216 (Fonfría et al. 2008), and other hydrocarbons, such as C₂H₆, that are also present in the atmosphere. Thus, looking for the lines of band ν₆ + ν₈ from the ground is a very good choice in order to detect C₄H₂, in spite of the absence of the prominent Q branch⁷ typical of π − σ bands that is present, e.g., in the ν₈ band (e.g., Fonfría et al. 2008).

We have identified 24 lines of the R and P branches of the C₄H₂ band ν₆ + ν₈ above the detection limit. Most of these lines belong to o-C₄H₂ due to the spin statistics. The detected lines are weak and show narrow features with a peak width dominated by the spectral resolution. They do not show a noticeable emission component, probably because of the spectrum noise, which overlaps with other spectral features, and the existence of several de-excitation routes from the upper vibrational state, such as the hot band ν₆ + ν₈ − ν₆, besides other radiation transfer and chemical reasons (see the text below and Section 4). The intensity of their absorption component suggests a column density ≲10¹⁶ cm⁻². These line profiles are compatible with a molecule formed either in the outer envelope or as close to the star as ≃10–15 R_⋆, as it occurs with C₂H₄ (Fonfría et al. 2017). C₂H₄ shows absorption features produced by two different excitation populations that are compatible with a molecular species arising in the dust formation zone (r ≲ 20 R_⋆; Fonfría et al. 2008), where the gas is still being accelerated and the kinetic temperature is above ≃400 K, and a significant absorption is produced in the colder shells of the outer envelope. This abundance profile implies redshifted high excitation lines with respect to the low excitation ones. The velocity of the line peak absorptions of the strongest C₄H₂ lines is −13.3 ± 0.8 km s⁻¹, with respect to the systemic velocity (≃−26.5 km s⁻¹; Cernicharo et al. 2000), which is typical of ro-vibrational lines formed once the terminal gas expansion velocity in IRC+10216 has been reached (≃14.5 km s⁻¹; e.g., Cernicharo et al. 2000; Fonfría et al. 2008). However, the velocity of the absorption peaks of the weakest line profiles is closer to the systemic velocity, as is for C₂H₄ (Section 4).

The use of a ro-vibrational diagram allows us to estimate the rotational temperature in the vibrational ground state (Figure 2). By assuming the same rotational temperature for the excited vibrational states and a given vibrational temperature (see the text below), we can calculate the total partition function and then obtain the total column density of C₄H₂. The C₄H₂ abundance is expected to be low, so we can consider the observed lines as optically thin. Thus, the following formula holds for each line:

$$\begin{eqnarray}&&\mathrm{ln}\left[\displaystyle \frac{8\pi {\nu }^{2}{cI}}{{A}_{{ul}}{g}_{u}{N}_{\mathrm{col},0}}\right]\simeq \mathrm{ln}\left[\displaystyle \frac{{N}_{\mathrm{col}}}{{N}_{\mathrm{col},0}Z}\displaystyle \frac{{\theta }_{s}^{2}}{{\theta }_{s}^{2}+{\theta }_{b}^{2}}\right]-\displaystyle \frac{{{hcE}}_{\mathrm{low}}}{{k}_{{\rm{B}}}{T}_{\mathrm{rot}}}\end{eqnarray} \tag{ 1 }$$

where ν is the rest frequency (cm⁻¹), I is the integrated absorption, A_ul is the A-Einstein coefficient (s⁻¹), g_u is the degeneracy of the upper level, Z is the total partition function, N_col is the column density (cm⁻²), E_low is the energy of the lower ro-vibrational level (cm⁻¹), and T_rot is the rotational temperature (K). The integrated absorption was estimated by means of a Gaussian fit that chose the proper baseline. This baseline was the continuum emission for isolated C₄H₂ lines and the profile of a molecular feature around the C₄H₂ line if it took part of a blending. The total partition function was calculated by direct summation over all the available ro-vibrational levels. The lack of hot bands and of an emission component in the detected lines prevents us from deriving the vibrational temperature of C₄H₂, which is necessary to calculate the partition function and thus the column density. We have then assumed the vibrational temperature for C₄H₂ equals that of the C₂H₂ band ${\nu }_{4}+{\nu }_{5}({\sigma }_{u}^{+})$ (Fonfría et al. 2008). ${N}_{\mathrm{col},0}={10}^{15}\,{\mathrm{cm}}^{-2}$ is a fixed column density included for convenience to get dimensionless arguments for the logarithms. The factor ${\theta }_{s}^{2}/({\theta }_{s}^{2}+{\theta }_{b}^{2})$ is the Point Spread Function (PSF) filling factor, where θ_s is the angular size of the C₄H₂ absorption, and θ_b results from the quadratic addition of the telescope half power beam width and the atmospheric seeing. It was ≃09 at 8 μm during our observing run. The size of the C₄H₂ absorption can be roughly estimated with our radiative transfer code (Fonfría et al. 2008, 2014), assuming the rotational temperature derived from the ro-vibrational diagram. Thus, θ_s = 063 ± 019, and ${\theta }_{s}^{2}/({\theta }_{s}^{2}+{\theta }_{b}^{2})=0.33\pm 0.13$. The uncertainties have been calculated, propagating the noise rms of the observed spectrum.

Zoom In Zoom Out Reset image size

From this diagram, we derive the existence of two different C₄H₂ populations (cold and hot), with rotational temperatures of 47 ± 7 and 420 ± 120 K and column densities of (4.5 ± 2.9) × 10¹⁵ and (1.9 ± 1.2) × 10¹⁶ cm⁻², respectively. The total diacetylene column density is (2.4 ± 1.5) ×10¹⁶ cm⁻². This low column density, distributed into rotational levels with energies spanning along several hundred of K, explains the weakness of the observed lines in the mid-infrared.

The lack of a permanent dipole moment of C₄H₂ prevents it to have a rotational spectrum. Therefore, it is rotationally under LTE, or very close to it in high density environments, but the population of the rotational levels can be significantly affected by the infrared continuum in low density ones. This effect is difficult to quantify so, to a first approximation, we can place the shell where the observed lines are formed using the kinetic temperature radial profile recently derived by Guélin et al. (2017) from high spatial resolution data of ¹²CO (T_k ≃ 257(r/0.8)^−0.675 K, if r ≲ 15'' and T_k ≃ 35 K beyond). Hence, the cold population of C₄H₂ is at ${10.0}_{-1.8}^{+2.7}$ arcsec from the star (${500}_{-90}^{+140}\,{R}_{\star }=({1.8}_{-0.3}^{+0.5})\times {10}^{16}\,\mathrm{cm}$, if R_⋆ = 002 = 3.7 × 10¹³ cm; Ridgway & Keady 1988; Fonfría et al. 2017), while the hot population arises at ${0.40}_{-0.14}^{+0.30}$ arcsec from the star (${20}_{-7}^{+15}\,{R}_{\star }=({7}_{-3}^{+6})\times {10}^{14}\,\mathrm{cm}$). We estimate the C₄H₂ abundance, with respect to H₂ to be 6 × 10⁻⁷ and 8 × 10⁻⁶, for the hot and cold populations, respectively, with an uncertainty of roughly a factor of 2.

3.1. Searching for Other Polyynes

The preliminary results derived from the ro-vibrational diagram (Section 3) suggest that the C₄H₂ hot population with a rotational temperature of ≃400 K arises from regions located around 20 R_⋆ from the star, which is similar to the case of C₂H₄ (Fonfría et al. 2017). The gas in this region of the envelope could expand at a lower velocity than the terminal velocity (≃11 versus ≃14.5 km s⁻¹), although this is still under debate (e.g., Decin et al. 2015; Fonfría et al. 2015). By adopting this gas expansion velocity field in the dust formation zone, the lines formed at r ≲ 20 R_⋆ are expected to show a velocity shift of a few km s⁻¹ compared to those that formed in the outer shells of the envelope. To explore this effect, we have divided the observed lines into three different groups: cold (E_low ≲ 40 K), hot (E_low ≳ 240 K), and transition (40 K ≲ E_low ≲ 240 K), stacking the previously scaled lines of each group to improve the signal-to-noise ratio and reduce the random shift of the absorption peak due to the spectral noise and the wavelength calibration uncertainties (Figure 3). The peak absorption of the cold stack, comprised of lines formed in the outer envelope, is shifted ≃1 km s⁻¹ with respect to the hot stack, which was formed by averaging the lines that are supposed to arise in the dust formation zone. The absorption peak of the transition stack, which is composed of the lines mostly formed around the acceleration shell at 20 R_⋆, is placed between both. This scenario supports the chosen gas expansion velocity profile and the formation of C₄H₂ in the dust formation zone, which is something unpredicted by the most commonly accepted photochemical models (Millar & Herbst 1994; Millar et al. 2000; Agúndez et al. 2017).

Zoom In Zoom Out Reset image size

The observed C₄H₂ lines can be roughly modeled with the code employed by Fonfría et al. (2017) to model the C₂H₄ spectrum toward IRC+10216. In this model, we adopted (1) the same kinetic and vibrational temperatures used during the analysis of the observational ro-vibrational diagram (Figure 2), (2) a mass-loss rate of 2.7 × 10⁻⁵ M_⊙ yr⁻¹ (Guélin et al. 2017), (3) a distance of 123 pc (Groenewegen et al. 2012), and (4) the gas expansion velocity proposed by Fonfría et al. (2015), i.e., 1 + 2.5(r/R_⋆ − 1) km s⁻¹ if 1 ≤ r/R_⋆ < 5, 11 km s⁻¹ if 5 ≤ r/R_⋆ < 20, and 14.5 km s⁻¹ if r/R_⋆ ≥ 20. We considered two different models that are compatible with the diacetylene abundance distribution derived in Section 3 to explore how its variation affects the ro-vibrational diagram: (1) C₄H₂ is distributed in two isolated shells ranging from 15 to 20 R_⋆ and from 400 to 1000 R_⋆, and (2) C₄H₂ is formed at 15 R_⋆ with a constant abundance up to 400 R_⋆, adopting another constant value beyond. We chose a distance of 15 R_⋆ as the inner radius of the abundance distribution because a value of 20 R_⋆ results in ro-vibrational diagrams with significantly steeper slopes at high E_low than the observed data suggest. Noticeable emission components arose in the synthetic lines after assuming shorter distances, which was something unobserved in our spectrum. The results are plotted in Figure 2 as the solid and dashed magenta curves. Both models are compatible with the observational ro-vibrational diagram, but the data seems to be better reproduced with model 1 (plotted in Figure 1). However, the transition lines (E_low ≃ 100–200 K) can also be slightly influenced by a low-abundance C₄H₂ contribution with a rotational temperature of a few hundred K located between ≃20 and 400 R_⋆.

The photochemical models for the outer envelope reproduce reasonably well the abundances of the molecules formed in the shells of C-rich evolved stars, such as IRC+10216 irradiated by dissociating radiation, in particular cyanopolyynes ${\mathrm{HC}}_{2n+1}{\rm{N}}$ (Millar & Herbst 1994; Millar et al. 2000; Cernicharo 2004; Agúndez et al. 2010, 2017). These models predict that C₄H₂ (and larger polyynes) arises mostly due to the reaction C₂H₂+C₂H→C₄H₂+H that can happen after the formation of C₂H, which results from the photodissociation of C₂H₂ (Figure 4). This polymerization process produces polyynes of an increasing length, as can be seen in Figure 5, where we show the radial abundance profiles calculated with the photochemical model of Agúndez et al. (2017). Briefly, this model calculates the chemical evolution of the isotropically expanding gas around a cold star starting at ≃5 R_⋆ from its center, where the considered parent molecules (H₂, CO, C₂H₂, CH₄, C₂H₄, H₂O, N₂, HCN, NH₃, CS, H₂S, SiS, SiO, SiH₄, PH₃, and HCP) were supposed to be already formed. To reproduce their Atacama Large Millimeter/submillimeter Array observations of carbon chains in IRC+10216, the authors assumed a smooth envelope externally illuminated by the local UV radiation field of Draine (1978) and a ratio N_H/A_V 1.5 times lower than the canonical value of 1.87 ×10²¹ cm⁻² mag⁻¹, derived by Bohlin et al. (1978) for the local interstellar medium. This model also takes into account the molecular ionization process triggered by cosmic-rays, which is known to be significant not only for H₂ but also for C₂H₂ as well (Gredel et al. 1989). The authors adopted a large network of around 8,300 chemical reactions that are mainly taken from the literature on gas-phase chemical kinetics and the University of Manchester Institute of Science and Technology (UMIST) and KInetic Database for Astrochemistry (KIDA) databases (McElroy et al. 2013; Wakelam et al. 2015). More details about the model can be found in Agúndez et al. (2017).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The predicted column density associated to the cold population (≃1.7 × 10¹⁵ cm⁻²) compared to our estimate ((4.5 ± 2.9) × 10¹⁵ cm⁻²) is particularly satisfactory, lying inside the 1σ error interval. This means that the predicted column density of C₄H, which is about half the C₄H₂ column density, is well estimated, and its spectrum could be detected in the infrared. The model also fairly well predicts the position of the shell where C₄H₂ arises and its abundance, which is compatible with our estimate regarding the cold population within a factor of 2. Nevertheless, this does not occur with the abundance of the hot population, which is several orders of magnitude higher than the model prediction in the shell where C₄H₂ forms. It is noteworthy that, contrary to what occurs with the column density, which is lower for the cold C₄H₂ population with respect to the hot one, the abundance (with respect to H₂ shown in Figure 5) is higher in the outer envelope than in the inner envelope. This apparent incompatibility is explained by the fact that the gas density in the outer shells (≃500 R_⋆) is well below its value at ≃20 R_⋆ from the star.

The disagreement in the abundance of the hot population found between the observations and the model results can be explained in two different ways. First, the high excitation C₄H₂ lines in the spectrum could be blended with stronger, unidentified lines. However, the intensity of these lines is compatible with the lower excitation lines (Figure 2), which is something that would not happen if they were from other molecules. This leads us to the second way, which suggests that the photochemistry model underestimates the abundance of C₄H₂ in the inner envelope. This would be in line with the idea that the dissociating external radiation field can reach shells that are significantly closer to the star than what is usually accepted. It is observationally supported by the works based on data acquired in the visible and the far-UV by Leão et al. (2006), Kim et al. (2015), and Matthews et al. (2015). Since the bulk opacity for this external dissociating radiation field is the dusty component of the envelope, dissociating photons can go deeper into the envelope if dust grains are inhomogeneously distributed in clumpy shells. Several authors have demonstrated that the circumstellar chemistry can be significantly modified if a higher density, higher temperature photochemistry and clumpiness are considered (Redman et al. 2003; Woods et al. 2003; Cernicharo 2004; Agúndez et al. 2010). In this scenario, the abundance of C₄H₂ is naturally explained at the same time as the existence of H₂O in the envelope of a C-rich star (Melnick et al. 2001; Agúndez & Cernicharo 2006; Agúndez et al. 2010; Neufeld et al. 2013) and the discovery of vibrationally excited C₄H and C₂H₄, which are as close to the star as ≃10 R_⋆ (Yamamoto et al. 1987; Cooksy et al. 2015; Fonfría et al. 2017).

In this paper, we have presented for first time 24 features of the C₄H₂ fundamental band ${\nu }_{6}+{\nu }_{8}({\sigma }_{u}^{+})$ observed with a high spectral resolution (R ≃ 85,000) toward the C-rich star IRC+10216 with the TEXES spectrograph mounted on the 3 m telescope IRTF. From the analysis of this spectrum, we conclude that:

• 1.  
  There are two C₄H₂ populations with different rotational temperatures (420 ± 120 and 47 ± 7 K). We estimate that these rotational temperatures are typical of shells at ≃04 ≃ 20 R_⋆ ≃ 7 × 10¹⁴ cm and ≃10'' ≃ 500 R_⋆ ≃1.8 × 10¹⁶ cm from the central star.
• 2.  
  The total C₄H₂ column density is (2.4 ± 1.5) × 10¹⁶ cm⁻². Only about 20% of it is located at the outer envelope, where the external dissociating radiation field is able to dissociate parent molecules coming from the inner layers of the envelope. The rest (80%) corresponds to C₄H₂ formed in the dust formation zone (r ≲ 20 R_⋆).
• 3.  
  The underestimation of the C₄H₂ abundance predicted by our photochemical model suggests that the molecules in the envelope are photodissociated in shells closer to the star than is commonly assumed. The easiest scenario in which this could happen would involve a clumpy outer envelope where the dust grains density undergo significant variations between different places.

We thank the anonymous referee for his/her comments about the manuscript. Development of TEXES was supported by grants from the NSF and USRA. The research leading to these results has received funding support from the European Research Council under the European Union's Seventh Framework Program (FP/2007-2013)/ERC Grant Agreement No. 610256 NANOCOSMOS. J.C. and M.A. thank Spanish MINECO through grants AYA2012-32032 and AYA2016-75066-C-1-P. M.A. also thanks funding support from the Ramón y Cajal program of Spanish MINECO (RyC-2014-16277).

Facility: IRTF(TEXES) - Infrared Telescope Facility.