Interstellar nitrile anions: Detection of C₃N⁻ and C₅N⁻ in TMC-1

Abstract

We report on the first detection of C₃N⁻ and C₅N⁻ towards the cold dark core TMC-1 in the Taurus region, using the Yebes 40 m telescope. The observed C₃N/C₃N⁻ and C₅N/C₅N⁻ abundance ratios are ~140 and ~2, respectively; that is similar to those found in the circumstellar envelope of the carbon-rich star IRC +10216. Although the formation mechanisms for the neutrals are different in interstellar (ion-neutral reactions) and circumstellar clouds (photodissociation and radical-neutral reactions), the similarity of the C₃N/C₃N⁻ and C₅N/C₅N⁻ abundance ratios strongly suggests a common chemical path for the formation of these anions in interstellar and circumstellar clouds. We discuss the role of radiative electronic attachment, reactions between N atoms and carbon chain anions C_n ⁻, and that of H⁻ reactions with HC₃N and HC₅N as possible routes to form C_nN⁻.

The detection of C₅N⁻ in TMC-1 gives strong support for assigning to this anion the lines found in IRC +10216, as it excludes the possibility of a metal-bearing species, or a vibrationally excited state. New sets of rotational parameters have been derived from the observed frequencies in TMC-1 and IRC +10216 for C₅N⁻ and the neutral radical C₅N.

1. Introduction

The importance of anions in the chemistry of interstellar clouds was analyzed in the early years of astrochemistry by Dalgarno and McCray (1973). The presence of carbon chain negative ions in space was predicted on the ground that electron radiative attachment is efficient for molecules with large electron affinities and dense vibrational spectra (Sarre, 1980; Herbst, 1981).

The first anion detected in space, C₆H⁻, was observed towards TMC-1 (McCarthy et al., 2006). Lines from this species were already reported as unidentified features in the line survey of IRC +10216 performed with the Nobeyama 45 m telescope by Kawaguchi et al. (1995). Their assignation to C₆H⁻ was not possible until the laboratory observations of McCarthy et al. (2006) became available. Nevertheless, Aoki (2000) suggested that the carrier of these lines was C₆H⁻ from ab initio calculations. The laboratory and space detection of this species prompted attention to the abundance of hydrocarbon anions in interstellar and circumstellar clouds. C₄H⁻ was first discovered in the circumstellar cloud IRC +10216 by Cernicharo et al. (2007) and then in the interstellar clouds L1527 and TMC-1 (Sakai et al., 2008; Agúndez et al., 2008a; Cordiner et al., 2013). C₆H⁻ was also detected towards other interstellar sources (Sakai et al., 2007; Gupta et al., 2009; Cordiner et al., 2011, 2013). Following the observation of C₈H⁻ in the laboratory (Gupta et al., 2007) this anion was found in TMC-1 (Brünken et al., 2007) and IRC +10216 (Kawaguchi et al., 2007; Remijan et al., 2007).

The nitrile anions CN⁻, C₃N⁻ and C₅N⁻ were first detected in the circumstellar envelope of the carbon-rich star IRC +10216 (Agúndez et al., 2010; Thaddeus et al., 2008; Cernicharo et al., 2008). While accurate laboratory frequencies were available for CN⁻ and C₃N⁻(Gottlieb et al., 2007; Thaddeus et al., 2008; Amano, 2008), the assignment of C₅N⁻ was based on the co-incidence of the observed rotational constants with ab initio calculations by Botschwina and Oswald (2008) and Aoki (2000). Although this species is the best candidate for this identification, the lack of precise laboratory frequencies prevents us from ruling out other species involving metals, which are present in IRC +10216. For example, MgC₃N and MgC₄H have been recently detected in IRC +10216 based on ab initio calculations (Cernicharo et al., 2019) and they have rotational constants B just a few megahertz below that of C₅N⁻. Moreover, there is controversy about the formation of C_nN⁻ anions through radiative electron attachment to C_nN radicals, for which calculated rate constants differ by orders of magnitude (Walsh et al., 2009; Khamesian et al., 2016; Millar et al., 2017). Hence, the detection of these anions in cold dark clouds and the determination of their abundances are an important step forward in understanding their chemistry in different astronomical environments.

In this Letter we present the detection of C₃N⁻ and C₅N⁻ in the cold dark core TMC-1. This is the first time nitrile anions are observed in the interstellar medium. The C₃N/C₃N⁻ and C₅N/C₅N⁻ abundance ratios derived are discussed within the frame of a chemical model of a cold dense cloud.

2. Observations

The Q-band (31.0-50.3 GHz) observations of TMC-1 (α₂₀₀₀=4^h41^m42.0^s, δ₂₀₀₀=25° 41′ 27.6″) were carried out during the winter 2019/2020 with the 40 m radio telescope of the Yebes observatory (IGN, Spain), hereafter Yebes 40 m. Observations of IRC +10216 in the same frequency band were performed in spring 2019 and have been previously described by Cernicharo et al. (2019) and Pardo et al. (2020). New receivers were built within the Nanocosmos project¹ and installed at the telescope (Tercero et al., 2020). They were used for the observations presented in this work. The Q-band receiver consists of two HEMT cold amplifiers covering the 31.0-50.3 GHz band with horizontal and vertical polarisations. Receiver temperatures vary from 22 K at 32 GHz to 42 K at 50 GHz. The backends are 16 × 2.5 GHz fast Fourier transform spectrometers (FFTS) with a spectral resolution of 38.1 kHz providing the whole coverage of the Q band in both polarisations. The main beam efficiency varies from 0.6 at 32 GHz to 0.43 at 50 GHz. Pointing corrections obtained by observing strong nearby quasars and were always within 2-3^″.

For IRC +10216 the observing mode was position switching with an off position at 300^″ in azimuth. The final spectra were smoothed to a resolution of 0.15 MHz, that is a velocity resolution of ≈1.5 and 0.9 km s⁻¹ at 31 and 50 GHz, respectively. The sensitivity of the final spectra varies between 0.4 mK and 1 mK per 0.15 MHz channel across the Q band, which is a factor of ≈10 better than previous observations in the same frequency range with the Nobeyama 45 m telescope taken with a spectral resolution of 0.5-0.625 MHz (Kawaguchi et al., 1995).

The TMC-1 Q-band observations were performed using the frequency switching technique with a frequency throw of 10 MHz. The nominal spectral resolution of 38.1 kHz was left unchanged for the final spectra because of the low temperature of this source and, therefore, narrowness of its lines. The average noise at this spectral resolution in the Q band ranges from ∽0.7 mK at 31 GHz to ∽2 mK at 49.5 GHz, which considerably improves previous line surveys in this frequency range for this source (Kaifu et al., 2004).

In order to improve the rotational constants of C₅N⁻ we used all lines of this species observed with the IRAM 30 m telescope since its detection (Cernicharo et al., 2008). These observations in the λ 3 mm band have been described in detail by Cernicharo et al. (2019). They correspond to observations acquired during the last 35 years covering the 70-116 GHz range with very high sensitivity (1-3 mK) over 1 MHz wide channels. Examples of these data can be found in Cernicharo et al. (2004, 2007, 2008,2019) and Agúndez et al. (2008b, 2014).

The beam size of the Yebes 40 m in the Q band is in the range 36-56^″, while that of the IRAM 30 m telescope in the 3 mm domain is 21-30^″. Pointing corrections were obtained by observing strong nearby quasars or SiO masers for both sources. Pointing errors were always within 2-3^″. The intensity scale for the observations with both telescopes, antenna temperature $\left({\text{T}}_A^{*}\right),$ was obtained after a calibration procedure that uses two absorbers at different temperatures and the atmospheric transmission model (ATM; Cernicharo 1985; Pardo et al. 2001). Calibration uncertainties are ~10 %. Additional uncertainties could arise, in the case of IRC +10216, from the line intensity fluctuation induced by the time variation of the stellar infrared flux (Cernicharo et al., 2014; Pardo et al., 2018). All data were analyzed via the GILDAS package².

3. Results

One of the most remarkable results from the Q-band observations in TMC-1 and IRC +10216 is the presence of a forest of weak lines. Most of these can be assigned to known species and only a few remain unidentified in IRC +10216 (Cernicharo et al., 2019; Pardo et al., 2020) and TMC-1 (Marcelino et al., in preparation). For both sources the level of sensitivity in this work has been increased, as commented previously, by a factor 5-10 with respect to previous works with other telescopes at the same frequencies (Kawaguchi et al., 1995; Kaifu et al., 2004). This high sensitivity has also allowed us to detect all known C_2nH and C_2n+1N anions in TMC-1. In this paper we focus on C₃N⁻ and C₅N⁻ lines in TMC-1. In addition, we use the observed frequencies of C₅N⁻ transitions towards IRC +10216 for an improved determination of the rotational and distortion constants of this anion.

3.1. C₃N⁻

Laboratory work exists for this anion and frequencies are well determined with accuracies of ⋍2 kHz (Thaddeus et al., 2008). This species has two rotational transitions in the 31.0-50.3 GHz frequency range. Figure 1 shows both of these transitions towards TMC-1, while the derived line parameters are given in Table 1. This species was searched towards TMC-1 by Thaddeus et al. (2008) without success, and to the best of our knowledge, this is the first time this anion has been detected in an interstellar cloud.

Figure 1.

Lines of C₃N⁻ observed towards TMC-1 in the 31.0-50.3 GHz frequency range. The abscissa corresponds to the local standard of rest velocity in km s^{− 1}. Frequencies and intensities for the observed lines are given in Table 1. The ordinate axis represents the antenna temperature in mK corrected for atmospheric and telescope losses.

Table 1. Observed line parameters for C₃N⁻ in TMC-1.

Transition	νarest (MHz)	vLSR (kms-1)	Δvb (kms-1)	∫T*Advc (mK kms-1)	T*A (mK)
J=4-3 F=4-3+F=5-4	38812.810	5.82±0.05	0.51±0.13	2.4±0.3	4.5
J=4-3 F=3-2	38812.728	5.84±0.10	0.68±0.16	2.3±0.4	3.3
J=5-4	48515.874	5.90±0.03	0.50±0.07	3.5±0.3	6.5

^(a)Rest frequencies as provided by the MADEX catalogue (Cernicharo, 2012). Typical uncertainties are 2 kHz.

^(b)Line width at half intensity derived by fitting a Gaussian line profile to the observed transitions.

^(c)Integrated line intensity.

3.2. C₅N⁻

The detected lines of C₅N⁻ towards TMC-1 are shown in Fig. 2. Only the J=18-17 transition at 49998.4 MHz has not been detected. This is, however, justified since the sensitivity of the data at 50 GHz is 3 mK, whereas the expected J=18-17 line intensity is below 5 mK (see also Fig. 2). No C₅N⁻ lines were detected with the Nobeyama 45 m telescope by Kaifu et al. (2004).

Figure 2.

Same as Fig. 1 but for C₅N⁻. Observed frequencies and intensities are given in Table 2.

In order to derive precise frequencies for C₅N⁻ in TMC-1 we fitted, in our Yebes 40 m data, the central v_LSRof the line emission from well-known species for which accurate laboratory rotational frequencies exist. From the seven HC₅N rotational transitions in the Q band (J_up=12 to J_up=18), Cernicharo et al. (2020) derive a v_LSRof 5.83 0.01 km s⁻¹. From the ¹³C and ¹⁵N isotopologues of HC₅N, they obtain v_LSR= 5.84 ± 0.01 km s⁻¹. Hence, we adopt a v_LSRof 5.83 km s⁻¹ for further frequency determinations in TMC-1. The value given by Kaifu et al. (2004) is 5.85 km s⁻¹, which is practically identical to our result within the uncertainties. Derived line parameters for C₅N⁻ in TMC-1 are given in Table 2.

Table 2. Observed line parameters for C₅N⁻ in TMC-1.

Ju	νaobs (MHz)	νo -ν b c (kHz)	Δvc (kms-1)	∫T*Advd (mK kms-1)	T*A (mK)
12	33332.572	1.7	0.75±0.08	5.0±0.2	6.2
13	36110.244	5.5	0.73±0.07	5.5±0.2	7.1
14	38887.893	-2.6	0.54±0.09	3.7±0.3	6.4
15	41665.548	7.1	0.50±0.16	2.4±0.41	4.6
16	44443.178	4.5	0.63±0.15	3.3±0.5	4.9
17	47220.754	-38.8	0.76±0.19	2.0±0.5	2.5

^(a)Observed frequencies for a v_LSR of 5.83 km s⁻¹. The uncertainty is 10 kHz for all lines except for J=17-16, for which it is 20 kHz.

^(b)Observed minus calculated frequencies.

^(c)Line width at half intensity derived by fitting a Gaussian line profile to the observed transitions.

^(d)Integrated line intensity.

The characteristic U-shaped line profiles exhibited by molecular lines in IRC +10216 allow an accurate central frequency determination, within ≃50 kHz (Cernicharo et al., 2018), in spite of the broad emission that covers 29 km s⁻¹(Cernicharo et al.,2000). The v_LSRof the source, -26.5 km s⁻¹, has been well determined from the observation of hundreds of lines (Cernicharo et al., 2000). The observed C₅N⁻ lines in IRC +10216 at λ ~7 mm (Q band) are shown in Fig. A.1. Frequencies and other line parameters are given in Table A.1. This table also includes lines observed in the 3 mm data obtained after the detection of this species in 2008. All lines at 3 mm re-observed after Cernicharo et al. (2008) have a spectral resolution of 0.198 MHz; these lines are shown in Fig. A.2.

3.3. New rotational constants for C₅N⁻

The frequencies of a linear molecular species can be fitted to this standard expression involving the rotational quantum number J, rotational constant B ₀, and the distortion constant D ₀:

$$v(J\to J-1)=2B_0\text{J-4}{D}_0{\text{J}}^3.$$

Using all observed C₅N⁻ lines in both sources, the fit provides the following results:

$$\begin{array}{l}{B}_0=1388.86681(19)\text{MHz}\\ D_0=34.44(13)\times {10}^{-6}\text{MHz,}\end{array}$$

where values between parentheses represent the 1σ uncertainty for the fitted parameters. The data were weighted in the fit according 1/δv ², where δv is the estimated uncertainty on the observed frequencies. The correlation coefficient between B ₀ and D ₀ is 0.738 and the standard deviation between the predicted and observed frequencies is 66 kHz. These values significantly improve those derived from previous observations by Cernicharo etal (2008), B ₀=1388.860(2) and D ₀=33(1)× 10⁻⁶ MHz. We tried to fit the distortion constant of order six (H ₀), but the derived value is only marginally significant. Its inclusion reduces the weighted deviation from 1.03 to 0.89 and the standard deviation from 66 to 53.7 kHz. However, it increases the correlation between the fitted parameters. The derived values in this case, are written as

$$\begin{array}{l}{B}_0=1388.86734(25)\text{MHz}\\ D_0=35.68(44)\times {10}^{-6}\text{MHz}\\ H_0=5.0(1.7)\times {10}^{-10}\text{MHz}.\end{array}$$

The differences between observed and calculated frequencies are given in Tables 2 and A.1 for TMC-1 and IRC+10216, respectively. New rotational parameters were also derived for C₅N (see Section A.1).

4. Discussion

Only two rotational C₃N⁻ lines have been detected in TMC-1. We assumed a volume density of H₂ of 4 10⁴ cm⁻³ (see e.g. Fossé et al. 2001), a kinetic temperature of 10 K, a dipole moment for the molecule of 2.27 D (Pascoli and Lavendy, 1999), and the collisional rates of C₃N⁻/H₂ from Lara-Moreno et al. (2019). For simulating the source we assumed a circular uniform brightness distribution with a radius of 40^″ (see e.g. the intensity maps of different carbon chains presented by Fossé et al., 2001). With this assumed size the source completely covers the main beam of the Yebes 40 m at 50 GHz.

Using the large velocity gradient approximation (LVG) implemented in MADEX (Cernicharo, 2012), and correcting the intensities of all observed transitions for the beam efficiency and the source beam dilution, we derive a column density of N(C₃N⁻)=1.3 × 10¹¹ cm⁻².

For C₅N⁻ we have enough observed lines to build the rotational diagram shown in Fig 3. A dipole moment of 5.2 D was adopted for this species (Botschwina and Oswald, 2008). A source size identical to that of C₃N⁻ was adopted. We derive a rotational temperature of 6.4 ± 0.6 K and a column density of (2.6 ± 0.9) 10¹¹ cm⁻². We also performed an LVG calculation with MADEX adopting for C₅N⁻ the collisional rates of C₆H⁻/p-H₂ (Walker et al., 2017). For a kinetic temperature of 10 K, the observed intensities corrected for beam efficiency and source beam dilution can be reproduced with a column density of 9 × 10¹¹ cm⁻². If the collisional rates of HC₅N/p-H₂ are adopted (F. Lique, private communication), then the column density is 1.5 × 10¹¹ cm⁻². Hence, the difference by a factor two of the latter value with respect to that from the rotational diagram is due to the uncertainties on both the collisional rates and the assumed H₂ volume density. We adopt for C₅N⁻ the column density derived from the rotational diagram of Fig. 3.

Figure 3.

Rotational diagram for the observed lines of C₅N⁻ in TMC-1.

The derived line parameters for C₃N and C₅N are given in Tables A.2 and A.3. For C₃N collisional rates are not available and we assumed a rotational temperature of 7 K (that is similar to that of C₅N⁻) to compute line intensities assuming local thermodynamical equilibrium conditions (LTE). Assuming the same source size as for the anions, we derive N(C₃N)=1.8×10¹³ B ₀=1388.86681(19) MHz cm⁻². For C₅N we can build a rotational diagram based on D ₀=34.44(13)×10⁻⁶ MHz, the data provided in Table A.3, which yields T_rot =10.1±1.90 K and N(C₅N)=(6.0±2.5)×10¹¹ cm⁻². The adopted dipole moment for this species is 3.385 D (Botschwina, 1996). Hence, the N(C₃N)/N(C₃N⁻) and N(C₅N)/N(C₅N⁻) abundance ratios in TMC-1 are ~140, and ~2.3, respectively.

In IRC +10216 Thaddeus et al.(2008) derived a N(C₃N)/N(C₃N⁻) abundance ratio of 194, and Cernicharo et al. (2008) obtained a N(C₅N)/N(C₅N⁻) abundance ratio around 2. Hence, the observed abundance ratios between neutral radicals C_nN and their anions are very similar in interstellar and circumstellar clouds.

One of the problems in obtaining the C_nN/C_nN⁻ abundance ratio is the assumed permanent dipole moment for the neutrals. In the case of C₅N, the dipole moment for the ground ²Σ electronic state was calculated by Botschwina (1996). However, as discussed by Cernicharo et al. (2008), the C₅N radical has a low lying ²Π electronic state with a dipole moment of ~1 D (Pauzat et al., 1991). Detailed calculations by Botschwina (1996) indicate that it lies 500 cm⁻¹ above the ²Σ ground state. Hence, C₅N could have a dipole moment between these two values in the case of admixing between the ²Σ and the ²Π states. A dipole moment averaged over both electronic states (i.e., twice as small as that calculated for the unperturbed ²Σ state) would raise the C₅N/C₅N⁻ abundance ratio to ~8 in TMC-1 and IRC +10216, which is very similar to the C₆H/C₆H⁻ ratio in IRC +10216 (Cernicharo et al., 2007).

The same situation applies to C₄H and C₄H⁻. The neutral radical also has a ²Σ⁺ ground electronic state with a first excited ²Π electronic state very close in energy. The case has been recently discussed by Oyama et al. (2020) who computed a dipole moment for C₄H 2.4 times larger than the value of the ²Σ⁺ state alone. The effect is a decrease by a factor ~6 on the derived C₄H column densities and an increase of the C₄H/C₄H⁻ abundance ratio by the same factor.

The chemistry of negative ions in cold interstellar clouds has been discussed by Walsh et al. (2009) and Millar et al. (2017). In the light of the discovery of interstellar C₃N⁻ and C₅N⁻, we revisit the chemistry of these species in this work. We performed a chemical model of a cold dark cloud with typical physical parameters (T_k= 10 K, n _H = 2 10⁴ cm⁻³, ζ = 1.3 10⁻¹⁷ s⁻¹, A_V= 30 mag) and “low metal” elemental abundances (see Agúndez & Wakelam 2013). We adopted the University of Manchester Institute of Science and Technology RATE12 reaction network (McElroy et al., 2013), with a subset of reactions involving HCCN from Loison et al. (2015). According to the model, formation of C₃N⁻ does not occur via radiative electron attachment to C₃N, which is slow (Petrie & Herbst, 1997), but through reactions between N atoms and bare carbon-chain anions C⁻ n(with n ≥ 6), which are rapid and produce several nitrile radical and anions (Eichelberger et al., 2007). In the case of C₅N⁻, the reaction of radiative electron attachment to C₅N is rapid (Walsh et al., 2009) and dominates the synthesis of this anion. Formation of C₃N⁻ through dissociative electron attachment to metastable isomers such as HNC₃(Harada & Herbst, 2008) is included and occurs to some extent, although most of C₃N⁻ is formed by N + C_n ⁻ reactions. Reactions of H⁻ with HC₃N and HC₅N could be a source of C N⁻ and C N⁻, although they are likely to be too slow at 10 K based on calculations for similar reactions of H⁻ with C₂H₂ and C₄H₂(Gianturco et al., 2016). Destruction of C₃N⁻ and C₅N⁻ is dominated by reactions with neutral atoms H, O, and C in cold dark clouds. Nitrile anions have been shown to react rapidly with polar molecules at low temperatures (Joalland et al., 2016). However, it is unlikely that this is a major loss channel for anions in cold dense clouds because, apart from CO, molecules have much lower abundances than neutral atoms.

The much larger rate constant of radiative electron attachment to C₅N compared to C₃N is the main reason why C₅N⁻ is calculated to be much more abundant with respect to the neutral than C₃N⁻ (see Fig. 4). The results of our model are similar to the theoretical predictions of Walsh et al. (2009). Calculated neutral-to-anion abundance ratios during the so-called early time (10⁵-10⁶ yr), at which calculated abundances agree better with observations (see e.g. Agúndez & Wakelam 2013), agree well with the values retrieved from observations for C₃N⁻, although they are somewhat higher for C₅N⁻ (see Fig. 4). We note that, as discussed above, the observed C N/C N⁻ ratio could be as high as ∼ 8, in which case model and observations would agree much better. Finally, we point out that the chemical model predicts an abundance of ∼ 3 × 10⁻¹¹ relative to H₂ for CN⁻. Adopting a column density of 10²² cm⁻² for H₂(Cernicharo & Guélin, 1987), the predicted column density of CN⁻ would be 3 10¹¹ cm⁻², which is below the 3σ upper limit of 1.4 10¹² cm⁻² derived by Agúndez et al. (2008a).

Figure 4.

Neutral-to-anion abundance ratios calculated with the chemical model are shown as a function of time and compared with observed values. The early time (10⁵-10⁶ yr) at which calculated abundances agree better with observations (see e.g. Agúndez & Wakelam 2013) is indicated. A conservative uncertainty of a factor of 2 is adopted for the observed ratios.

This work definitively excludes metal-bearing species, or vibrationally excited states of other known species, as carriers for the series of lines assigned to C₅N⁻ by Cernicharo et al. (2008), and gives strong support to this identification.