Submillimeter spectra of 2-hydroxyacetonitrile (glycolonitrile; HOCH2CN) and its searches in GBT PRIMOS observations of Sgr B2(N)⋆
1 Laboratoire de Physique des Lasers, Atomes, et Molécules, UMR CNRS 8523, Université de Lille I, 59655 Villeneuve d’Ascq Cedex, France
2 National Radio Astronomy Observatory, Charlottesville, VA 22903, USA
3 Departamento de Química y Física Teóricas, Instituto de Estructura de la Materia, IEM-C.S.I.C., Serrano 121, 28006 Madrid, Spain
4 Institut des Sciences Chimiques de Rennes, École Nationale Supérieure de Chimie de Rennes, CNRS, UMR 6226, 11 allée de Beaulieu, CS 50837, 35708 Rennes Cedex 7, France
Received: 18 March 2016
Accepted: 10 October 2016
Context. Recent experimental works have studied the possible formation of hydroxyacetonitrile on astrophysical grains. It was formed from hydrogen cyanide (HCN) and formaldehyde (H2CO) in the presence of water under interstellar medium conditions. Because these precursor molecules are abundant, hydroxyacetonitrile is an excellent target for interstellar detection.
Aims. Previous studies of the rotational spectra were limited to 40 GHz, resulting in an inaccurate line list when predicted up to the millimeter-wave range. We measured and analyzed its spectra up to 600 GHz to enable is searches using cutting-edge millimeter and submillimeter observatories.
Methods. The molecule 2-hydroxyacetonitrile exhibits large amplitude motion that is due to the torsion of the hydroxyl group. The analysis of the spectra was made using the RAS formalism available in the SPFIT program with Watson’s S-reduction Hamiltonians.
Results. The submillimeter spectra of hydroxyacetonitrile, an astrophysically interesting molecule, were analyzed. More than 5000 lines were fitted with quantum number values reaching 75 and 25 for J and Ka, respectively. An accurate line list and partition function were provided. A search for hydroxyacetonitrile in publicly available GBT PRIMOS project, IRAM 30 m, and Herschel HEXOS observations of the Sgr B2(N) high-mass star-forming region resulted in a non-detection; upper limits to the column density were determined.
Key words: ISM: molecules / methods: laboratory: molecular / submillimeter: ISM / molecular data / line: identification
Hydroxyacetonitrile fit is only available at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (126.96.36.199) or via http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/601/A50
© ESO, 2017
More than 30 nitriles (organic cyanides) have been observed to date in the interstellar medium (ISM). Among these are the well-known family of cyanopolyynes, and isopropyl cyanide, the first branched hydrocarbon detected in the ISM (Belloche et al. 2014). This large dataset is useful in constraining models of nitrile formation in the ISM, and in predicting which other nitrile species might be present and detectable. However, many simple nitriles have yet to be detected. These non-detections could be due to a weak or null dipolar moment (for example cyanogen NCCN or 2-butyne-1,4-dinitrile NC–C≡C–CN) or to the lack of recorded microwave spectra caused by unknown or difficult syntheses of the target products. With the advent of state-of-the-art millimeter and submillimeter facilities like the Stratospheric Observatory for Infrared Astronomy (SOFIA) and the Atacama Large Millimeter/submillimeter Array (ALMA), laboratory studies of rotational spectra in this frequency range are critical for advancing the field. Such work will have to be followed by attempts, both in the laboratory and through theoretical models, to understand formation processes and observed interstellar abundances (or upper limits).
The simplest cyanohydrine, 2-hydroxyacetonitrile (glycolonitrile; HOCH2CN), is readily formed in the lab by the addition of hydrogen cyanide (HCN) to formaldehyde (H2CO) in water (Gaudry 1955). It is thought to play an important role in prebiotic chemistry, and could lead to the formation of a number of simpler organic molecules (Arrhenius et al. 1994). The formation of 2-hydroxyacetonitrile in astrophysical ices was first studied from a theoretical point of view (Woon 2001). Recently, its formation and photolytic decomposition on interstellar grains have been experimentally evaluated. These studies show that under these conditions, HCN reacts with H2CO in the presence of H2O to form HOCH2CN (Danger et al. 2012, 2014). Although not studied, the hydrogen-atom bombardment of formyl cyanide (HC(=O)CN) could also lead to the formation of HOCH2CN, as could do the reaction of CN or OH with H2COH or CH2CN, respectively. Photochemistry of pure HOCH2CN leads to the formation of HC(=O)CN, HCN, ketenimine (H2C=C=NH), H2CO, CO, and CO2 (Danger et al. 2013). It should be noted that most of the compounds cited as possible precursors or as photoproducts have been detected in the ISM.
The gas-phase, infrared spectrum of HOCH2CN has been recently reinvestigated (Chrostowska et al. 2015), while the microwave spectrum of HOCH2CN has not been treated since it was first observed by Cazzoli et al. (1973) from 8–40 GHz. The gauche rotamer has been observed. Doublets of the rotational transitions were observed arising from tunneling between the two equivalent gauche rotamers. The rotational constants indicate that the angle between the HOC and OCC planes at the potential minima is approximately 123° from the trans position, while the rotational constants calculated from a molecular model, and the potential functions, indicate that this angle is likely closer to 116°.
New spectra were measured in Lille in the range 150–600 GHz, and combined with high-level ab initio calculations for an accurate analysis. A search of the publicly available GBT PRIMOS project, IRAM 30 m, and Herschel HEXOS observations of Sgr B2(N) for the molecule found no evidence for a cold population of the species, and upper limits were established.
The measurements from 50–660 GHz were performed using the Lille spectrometer (Zakharenko et al. 2015). A quasi-optic dielectric hollow waveguide, 3 m in length, containing HOCH2CN was used as the sample cell in the spectrometer. The measurements were typically performed at pressures of 10 Pa and at room temperature. The frequency ranges of 50–315 and 400–550 GHz were covered with various active and passive frequency multipliers where an Agilent synthesizer (12.5–18.25 GHz) was used as the source of radiation. Estimated uncertainties for measured line frequencies are 30 kHz and 50 kHz depending on the observed signal-to-noise ratio and the frequency range.
High-level ab initio calculations were used to determine the geometries of the two conformers, gauche and trans, and to compute low-energy torsional levels. Electronic structure calculations were performed using both the MOLPRO (Werner et al. 2012) and GAUSSIAN packages (Frisch et al. 2009). The torsional energy levels were determined by variational calculations using the procedure previously employed for other non-rigid species (Senent 2004).
The equilibrium structures, gauche and trans, of HOCH2CN, as well as the most relevant spectroscopic parameters (equilibrium rotational constants and the one-dimensional potential energy surface, 1D-PES, for the hydroxyl torsion) were computed using explicitly correlated coupled cluster theory with single and double substitutions augmented by a perturbative treatment of triple excitations (CCSD(T)-F12b) (Knizia et al. 2009; Werner et al. 2007) with an aug-cc-pVTZ basis set (Kendall et al. 1992). Second-order Möller-Plesset theory (MP2) was employed to determine vibrational corrections for the potential energy surface and the vibration-rotation constants. For the explicitly correlated calculations, the MOLPRO default options were selected.
The ground vibrational state rotational constants of the two conformers were determined from the equilibrium parameters using Eq. (1)(cf. Senent et al. 2015) (1)where represents the core-valence correlation correction and ΔBvib is the vibrational contribution to the rotational constants derived from the vibration-rotation interaction parameters. The results are shown in Table 1.
Calculated ground-state rotational constants (MHz), energies and conformational barriers (cm-1), and dipole moments (Debye) for HOCH2CN.
The low torsional energy levels were calculated using a variational solution to the Hamiltonian given in Eq. (2) (Senent 1998a,b): (2)Here, α is the OH torsional coordinate, Bα are the kinetic energy parameters, and Veff(α) is the vibrationally corrected 1D-PES (Császár et al. 2004) shown in Fig. 1.
The ground vibrational state splits into two components E(0−) and E(0+) because of tunneling in the V(gauche-gauche) barrier. Ab initio calculations predict a splitting of 96 563.148 MHz (3.2210 cm-1). In addition, the first excited vibrational state shows two components lying at 223.05 cm-1 (1+) and 261.27 cm-1 (1−) over the ground state. The first trans level lies at 459 cm-1 (0).
CCSD(T)-F12 potential energy as a function of the hydroxyl group internal rotation coordinate.
|Open with DEXTER|
We have analysed the spectra of the gauche conformer, which is the more stable of the two. This analysis was not trivial because the gauche rotamer exhibits one large amplitude motion (LAM) with a symmetric two minima potential due to the two equivalent gauche rotamers. As a result of tunneling through the barrier to OH-group torsion, the ground state is split into two substates: 0− and 0+. This rotamer is very close to a symmetric prolate rotor with an asymmetry parameter κ = −0.968. Like other nitriles, the dipole moment is large; the values we determined from MP2/aug-cc-pVTZ ab initio calculations and are given in Table 1. The component along the c-axis of the dipole moment is linked to torsional motion, giving rise to transitions between the two torsional substates 0− and 0+.
An initial prediction of the spectra was made for the two substates 0− and 0+ separately with Watson’s Hamiltonian in the Ir representation using the rotational parameters from Cazzoli et al. (1973) and the quartic centrifugal distortion constants obtained from the harmonic force field at the MP2/aug-cc-pVTZ level of theory. We first assign the A-type, R-branch transitions, which are the brightest in our spectra. Owing to the rather strong interaction between these two sublevels, it was not possible to assign and fit data even for Ka = 0 and 1 transitions in the lower range of the spectra 150–315 GHz. Some less perturbed levels were found above 400 GHz, and an assignment and fit was possible by treating the two sublevels separately for Ka values up to 2 in the 400–500 GHz range, improving the values of the rotational and the centrifugal distortion constants. The two sublevels were then merged in a unique fit, using the energy difference between the 0− and 0+ levels determined by Cazzoli et al. (1973) as the starting point. We note that this value, 110 700 MHz, is close to the one we determined, 112 672.5526(30) MHz.
For the global fit, a model based on the reduced axis system (RAS) approach proposed by Pickett (1972) was implemented in the widely used SPFIT/SPCAT programs for fitting and predicting molecular spectra. We have successfully employed this method in previous studies showing the same kind of double minimum LAM (Motiyenko et al. 2010, 2015; Smirnov et al. 2013). Is has been shown by Christen et al. (2002) that the RAS formalism is equivalent to the IAM approches developed by Hougen and coworkers for molecules with several large amplitude motions (Hougen 1985; Coudert & Hougen 1988).
The perturbation terms in the RAS Hamiltonian are equivalent in this case to off-diagonal Coriolis coupling terms; the Hamiltonian used has the form given in Eq. (3): (3)More details about the exact expression of the Hamiltonian can be found in Christen & Müller (2003). In that work, they also proposed the use of average rotational constants for both substates introducing a Hamiltonian for centrifugal distortion corrections, given in Eq. (4), (4)where HR is the standard Watson S-reduction Hamiltonian in the Ir coordinate representation and HΔ defined in Eq. (5), (5)where E∗ is a half energy difference between two A or B levels (ΔE = 2E∗), and J, Jz, and J± = Jx ± iJy are the rotational angular momentum operator and its components. This procedure has two main advantages. First, a unique set of rotational and centrifugal distortion parameters permits an easier comparison with ab initio calculations. Second, this method is more robust and avoids correlations between different rotational and Coriolis coupling parameters.
Spectroscopic parameters of the gauche conformer of 2-hydroxyacetonitrile.
Including the F2ac parameter slightly decreases the standard deviation of fit from 29.2 kHz (wrms: 0.931) to 28.5 kHz (wrms: 0.907), and it seems to be determined: −0.0967(82). Looking closely, however, including F2ac has several negative consequences on the fit. Ten lines were rejected from the fit after including F2ac. We also notice a strong influence on FacK: its values change from 0.009943(25) to 0.00304(59) when fitting F2ac. This influence is confirmed by looking at the correlation coefficient; there is almost full correlation with Fac, FacJ, and FacK, with coefficients of −0.996, −0.990, and 0.999, respectively. We have thus decided to keep F2ac fixed to zero in the final fit. The final set of parameters obtained are in Table 2. In total, 5128 distinct lines were assigned with maximum values for J and Ka of 75 and 25, respectively. Part of the new measurements are in Table 3. Owing to its large size, the complete version of the global fit Table (S1) is supplied at the CDS1. The fitting files .lin (S2), .par (S3), and the prediction .cat (S4) are also available at CDS.
Measured frequencies of the gauche conformer of 2-hydroxyacetonitrile and residuals from the fit (full fit is available at the CDS: S1).
We have conducted a search of the publicly available Prebiotic Interstellar Molecular Survey (PRIMOS) project observations of the high-mass star-forming region Sgr B2(N) for a signal from HOCH2CN. PRIMOS is a near-frequency continuous, 1–50 GHz broadband molecular line survey conducted with the Robert C. Byrd Green Bank Telescope (GBT), and centered on the Large Molecular Heimat at (J2000) , δ = −28°22′17′′. Full observational and data-reduction details are given in Neill et al. (2012); the entire dataset is accessible online2. The PRIMOS observations vary in RMS noise level from ~3–8 mK across the band; however, we find no evidence for emission or absorption from HOCH2CN in the survey. Below, we establish upper limits to the column density for a variety of physical conditions, and briefly discuss the possibilities for future searches for this molecule.
We follow the convention of Hollis et al. (2004) using Eq. (6)to calculate an upper limit to column density NT, given a rotational partition function Q, upper state energy EU, rotational temperature Tr, transition frequency ν, intrinsic line strength Sμ2, observed intensity , linewidth ΔV, telescope efficiency ηB, and background temperature Tbg: (6)The structure of Sgr B2(N) is complex, with a compact (~5′′) hot molecular core surrounded by a more extended, colder molecular shell (Hollis et al. 2007). The background continuum structure against which molecules in this shell absorb is ~20′′ in diameter (see, e.g., Fig. 3b of Mehringer et al. 1993). For the purposes of this study, we assume these same source sizes for warm molecular material and cold molecular material and explicitly calculate the spatial overlap of these regions with the GBT beam as its size varies across the frequency coverage.
Hollis et al. (2007) establish the background continuum temperature at 85 points across the PRIMOS frequency coverage. These continuum measurements range from more than 100 K at low frequencies to just above the CMB at the higher frequencies, although these measurements are not corrected for any assumed source geometry or beam dilution effects. The authors attribute this to non-thermal continuum in the source. We adopt their measurements of the continuum for these calculations, but correct them for the overlap of the assumed source size of the continuum-emitting region with the GBT beam in the PRIMOS observations.
The tabulated values for the partition function are given in Table 4, with Qtot(T) = Qvib(T)Qrot(T). The rotational partition function is calculated at each temperature by direct state counting according to Eq. (7)(Gordy & Cook 1984): (7)The vibrational partition function was calculated with respect to the zero-point level using the expression: (8)The four lowest vibrational excited states levels were considered. The remaining ones above 885 cm-1 were found to have no influence on the partition function calculations. Their frequencies are respectively 228 cm-1 (328 K), 246 cm-1 (354 K), 343 cm-1 (494 K), and 576 cm-1 (829 K); these values were taken from Chrostowska et al. (2015).
Finally, we assume that the excitation of all HOCH2CN can be described by a single rotational temperature. We do not, however, make the assumption that this temperature is described by local thermodynamic equilibrium (LTE), meaning that the excitation temperature is not thermalized to the kinetic temperature of the gas. This approach has previously proven to be successful in modeling both the emission and absorption of molecules in PRIMOS observations (Hollis et al. 2004; Loomis et al. 2013; McGuire et al. 2016). We note, however, that several other molecules exhibit populations which cannot be described by a single excitation temperature (McGuire et al. 2012; Faure et al. 2014), and thus care should be taken to examine any single-excitation temperature results on a case-by-case basis.
In addition to the PRIMOS observations, we have also examined publicly available surveys of Sgr B2(N) from Belloche et al. (2013) [IRAM 30 m: 80–116 GHz] and Neill et al. (2014) [Herschel HEXOS: 480–600 GHz]. No compelling evidence for HOCH2CN is found in either survey. With the exception of the warm (200 K), extended (200′′) population, the upper limits established by the PRIMOS data are not changed by more than ~10% due to greater noise in the higher-frequency data, compared to PRIMOS.
Table 5 shows the upper limits calculated here for several different common populations of species. The given values represent the best limits achievable across all three datasets. It is immediately clear that the PRIMOS observations are most sensitive to cold populations of HOCH2CN, with the established upper limits several orders of magnitude lower than for warmer excitation conditions. This occurs because for relatively large, complex species such as HOCH2CN, which have large partition functions, any significant level of excitation pushes the Boltzmann peak into the mm and sub-mm regimes, resulting in low spectral intensities in the region of PRIMOS coverage. Our searches at higher frequencies suffered greatly from more heavily line-confusion-limited spectra; ALMA observations of the region are likely the only method for spatially filtering this confusion.
The first detection of methyl isocyanate (CH3NCO), a structural isomer of HOCH2CN, has recently been reported in Sgr B2(N) by Halfen et al. (2015) in a cold, extended population (Tr~26 K, NT~2×1013 cm-2) in two velocity components. A second warmer population (Tr~200 K, NT~4×1017 cm-2) was subsequently identified by Cernicharo et al. (2016) in two velocity components. For comparison, we have calculated upper limits for HOCH2CN under these excitation conditions, as well.
Rotational and vibrational partition functions at various temperatures.
For the warm population of Cernicharo et al. (2016), the CH3NCO:HOCH2CN ratio is >1000:1. The ratio for the colder population of Halfen et al. (2015) is, conversely, >1:1. If the same formation pathways are in effect for both the warm and cold populations, the upper limit established for the cold population is likely closer to ~1010 cm-2. In this case, the best-case scenario for detection is likely through masing action in low-frequency transitions. Such emission enhancement has been seen in this source (Faure et al. 2014), and has indeed been used to detect extremely low-abundance species (e.g., carbodiimide, HNCNH; McGuire et al. 2012).
Upper limits to HOCH2CN column density in PRIMOS, IRAM, and Herschel HEXOS observations toward Sgr B2(N) for several common conditions.
The spectra of 2-hydroxyacetonitrile, a molecule expected to have a role in prebiotic chemistry, were measured and analyzed up to 600 GHz. The assignment of the spectra was not obvious owing to large amplitude motion of the OH group, but was aided by high-level ab initio calculations. With the newly constrained constants, the fit and predictions are now accurate throughout the millimeter and submillimeter regions. These results enabled its search in the GBT PRIMOS survey toward Sgr B2(N). No cold population was detectable; however, the possibility of a warmer population remains, and upper limits were established.
This work was supported by the Programme National “Physique et Chimie du Milieu Interstellaire” and the Centre National d’Études Spatiales (CNES). This work was also done under ANR-13-BS05-0008-02 IMOLABS. B.A.M. is a Jansky Fellow of the National Radio Astronomy Observatory. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.
- Arrhenius, T., Arrhenius, G., & Paplawsky, W. 1994, Origins of life and evolution of the biosphere, 24, 1 [NASA ADS] [CrossRef] [Google Scholar]
- Belloche, A., Garrod, R. T., Müller, H. S. P., & Menten, K. M. 2014, Science, 345, 1584 [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
- Cazzoli, G., Lister, D. G., & Mirri, A. M. 1973, J. Chem. Soc., Faraday Trans. 2, 69, 569 [CrossRef] [Google Scholar]
- Christen, D., & Müller, H. S. P. 2003, Phys. Chem. Chem. Phys., 5, 3600 [NASA ADS] [CrossRef] [Google Scholar]
- Christen, D., Müller, H. S. P., & Coudert, L. H. 2002, 57th International Symposium on Molecular Spectroscopy, TB03 [Google Scholar]
- Chrostowska, A., Darrigan, C., Dargelos, A., Benidar, A., & Guillemin, J.-C. 2015, Chem. Phys. Chem., 16, 3660 [CrossRef] [Google Scholar]
- Coudert, L. H., & Hougen, J. T. 1988, J. Mol. Spectr., 130, 86 [NASA ADS] [CrossRef] [Google Scholar]
- Császár, A. G., Szalay, V., & Senent, M. L. 2004, J. Chem. Phys., 120, 1203 [NASA ADS] [CrossRef] [Google Scholar]
- Danger, G., Duvernay, F., Theulé, P., Borget, F., & Chiavassa, T. 2012, ApJ, 756, 11 [NASA ADS] [CrossRef] [Google Scholar]
- Danger, G., Duvernay, F., Theulé, P., et al. 2013, A&A, 549, A7 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
- Danger, G., Rimola, A., Abou Mrad, N., et al. 2014, Phys. Chem. Chem. Phys., 16, 3360 [CrossRef] [Google Scholar]
- Faure, A., Remijan, A. J., Szalewicz, K., & Wiesenfeld, L. 2014, ApJ, 783, 72 [NASA ADS] [CrossRef] [Google Scholar]
- Frisch, M., Trucks, G., Schlegel, H., et al. 2009, Gaussian software, version 09, revision D01 [Google Scholar]
- Gaudry, R. 1947, Organic Syntheses, 27, 41 [CrossRef] [Google Scholar]
- Gaudry, R. 1955, Organic Syntheses, Coll, 3, 436 [Google Scholar]
- Gordy, W., & Cook, R. L. 1984, Microwave molecular spectra, 3rd edn. (New York: Wiley) [Google Scholar]
- Hollis, J., Jewell, P., Remijan, A. J., & Lovas, F. 2007, ApJ, 660, L125 [NASA ADS] [CrossRef] [Google Scholar]
- Hollis, J. M., Jewell, P., Lovas, F., & Remijan, A. 2004, ApJ, 613, L45 [NASA ADS] [CrossRef] [Google Scholar]
- Hougen, J. T. 1985, J. Mol. Spectr., 114, 395 [NASA ADS] [CrossRef] [Google Scholar]
- Kendall, R. A., Dunning Jr, T. H., & Harrison, R. J. 1992, J. Chem. Phys., 96, 6796 [NASA ADS] [CrossRef] [Google Scholar]
- Knizia, G., Adler, T. B., & Werner, H.-J. 2009, J. Chem. Phys., 130, 054104 [NASA ADS] [CrossRef] [Google Scholar]
- Loomis, R. A., Zaleski, D. P., Steber, A. L., et al. 2013, ApJ, 765, L9 [NASA ADS] [CrossRef] [Google Scholar]
- McGuire, B. A., Loomis, R. A., Charness, C. M., et al. 2012, ApJ, 758, L33 [NASA ADS] [CrossRef] [Google Scholar]
- McGuire, B. A., Carroll, P. B., Loomis, R. A., et al. 2016, Science, 352, 1449 [NASA ADS] [CrossRef] [Google Scholar]
- Mehringer, D. M., Palmer, P., Goss, W., & Yusef-Zadeh, F. 1993, ApJ, 412, 684 [NASA ADS] [CrossRef] [Google Scholar]
- Motiyenko, R. A., Margulès, L., Goubet, M., et al. 2010, J. Phys. Chem. A, 114, 2794 [CrossRef] [Google Scholar]
- Motiyenko, R. A., Margulès, L., Alekseev, E. A., & Guillemin, J.-C. 2015, J. Phys. Chem. A, 119, 1048 [CrossRef] [Google Scholar]
- Neill, J. L., Muckle, M. T., Zaleski, D. P., et al. 2012, ApJ, 755, 153 [NASA ADS] [CrossRef] [Google Scholar]
- Pickett, H. M. 1972, J. Chem. Phys., 56, 1715 [NASA ADS] [CrossRef] [Google Scholar]
- Senent, M. 1998a, J. Mol. Spectr., 191, 265 [NASA ADS] [CrossRef] [Google Scholar]
- Senent, M. 1998b, Chem. Phys. Lett., 296, 299 [NASA ADS] [CrossRef] [Google Scholar]
- Senent, M. 2004, J. Phys. Chem. A, 108, 6286 [CrossRef] [Google Scholar]
- Senent, M., Dalbouha, S., Cuisset, A., & Sadovskii, D. 2015, J. Phys. Chem. A, 119, 9644 [CrossRef] [Google Scholar]
- Smirnov, I., Alekseev, E., Piddyachiy, V., Ilyushin, V., & Motiyenko, R. 2013, J. Mol. Spectr., 293, 33 [NASA ADS] [CrossRef] [Google Scholar]
- Werner, H., Knowles, P., Knizia, G., et al. 2012, MOLPRO, version 2012.1, a package of ab initio programs, see http://www.molpro.net [Google Scholar]
- Werner, H.-J., Adler, T. B., & Manby, F. R. 2007, J. Chem. Phys., 126, 164102 [NASA ADS] [CrossRef] [Google Scholar]
- Woon, D. E. 2001, Icarus, 149, 277 [NASA ADS] [CrossRef] [Google Scholar]
- Zakharenko, O., Motiyenko, R. A., Margulès, L., & Huet, T. R. 2015, J. Mol. Spectr., 317, 41 [NASA ADS] [CrossRef] [Google Scholar]