© ESO, 2016

1. Introduction

Ice mantles on dust grains are formed from the accretion of gas phase molecules during the gravitational collapse of clouds that form new stars and their planetary systems. Surface reactions increase the chemical complexity of the mantles which, when their temperature rises, release their frozen molecular material to the gas phase (Herbst & van Dishoeck 2009). The gas and dust material of the cloud cores formed during the gravitational collapse participate in the formation of giant gas planets and rocky bodies such as comets, asteroids, and the Earth. Comets are considered to be a repository of information on their frozen surface of the gas and dust composition of the primitive solar nebula. As they approach the Sun in their journey their surface ices start to evaporate and to create a coma of gas and dust that has been extensively studied at all wavelengths (Bockelée-Morvan et al. 2000; Crovisier 2006; Mumma & Charnley 2011). A good correlation has been found between the type of molecules detected in the coma of comets and those of warm molecular clouds (Bockelée-Morvan et al. 2000; Biver et al. 2015).

The Cometary Sampling and Composition (COSAC) experiment aboard Rosetta’s Philae lander has measured in situ abundances of the main components of the surface of comet 67P/Churyumov-Gerasimenko (Goesmann et al. 2015). Most of the “gas on the rocks” species assigned by COSAC are well-known interstellar molecules. However, many of the species detected by COSAC have not been observed in the coma of previously observed comets (Goesmann et al. 2015), while some organic molecules observed in the coma of comets are not detected by COSAC. Nevertheless, the coma organics can be formed by UV irradiation of the ices forming alcohols and carbonyls from CO and H₂O ices (Muñoz Caro & Dartois 2013). Among the organic molecules detected on the surface of the comet, only one was not detected in space, methyl isocyanate (CH₃NCO), which could potentially contribute to the mass peak m/z = 57 of COSAC. Although it is a molecule of potential relevance, to our knowledge it has not been included in any astrochemical network, and to date its detection has only been reported towards Sgr B2 based on a reduced number of lines (Halfen et al. 2015), whereas on 67P/Churymov-Gerasimenko it was identified as one of the most abundant molecules after water (Goesmann et al. 2015). The most promising interstellar clouds in which to search for CH₃NCO are hot cores and hot corinos.

In 2006 we started a sensitive line survey in the millimetre domain (80−280 GHz) of Orion using the IRAM 30 m radio telescope (Tercero et al. 2010) with the goal of fully characterizing its chemical composition (see Appendix B). However, owing to the high kinetic temperature of the gas, T_K ≃ 100−300 K, many rotational and vibrational levels of abundant species are populated, producing a forest of spectral lines. Around 15000 spectral features were initially detected of which 8000 were unassigned. A systematic work programme in spectroscopic laboratories was started, which allowed us to significantly reduce the number of unidentified lines and to detect new molecules such as NH₃D⁺ (Cernicharo et al. 2013; Domenech et al. 2013), CH₃COOCH₃ (Tercero et al. 2013), and CH₃OCH₂CH₃ (Tercero et al. 2015).

In this Letter we report on the complete laboratory spectroscopic characterization of CH₃NCO and the detection of 399 of its lines in Orion. We also discuss the spectroscopic data used by Halfen et al. (2015) and comment on their column density estimations towards Sgr B2.

2. Spectroscopy

CH₃NCO was initially searched for in our Orion data by using the available sub-40 GHz microwave spectroscopic work for this species (Koput 1986). However, irrespective of the usual problems of extrapolation outside the data region, we faced additional problems affecting the rotational spectrum of this molecule (see Fig. 1). Koput complemented his thorough experimental investigation of the 8−40 GHz region by Stark spectroscopy with analysis based on fitting a five-dimensional, quasi-symmetric top Hamiltonian. He was unable to fit the measured transition frequencies to experimental accuracy, but he reached unambiguous assignment of rotational transitions in many internal rotation substates of the ground and the two lowest excited states of the CNC bending vibrational mode (Koput 1986). The very low effective barrier to internal rotation, V₃ = 21 cm^-1, was found to result in a series of low-lying internal rotor states, with the distribution close to that of the free rotor pattern, and estimated energies relative to the m = 0 ground state of 8.4 cm^-1 for the m = 1 substate, 36.8 cm^-1 for m = 2, 79.7 and 80.3 cm^-1 for the two nearly degenerate m = 3 substates, and 140.6 cm^-1 for m = 4. The next higher vibrational state was found to be the first excited state of the CNC bending mode, ν_b = 1 at 182.2 cm^-1. That state, in turn, carries its own stack of internal rotor states, and so on. The resulting high state density and considerable vibration-rotation coupling are the main reason why the laboratory understanding of the rotational spectrum of this molecule has been considerably delayed.

In order to derive a reliable CH₃NCO line list for astrophysical applications we decided to cut through these problems by using the procedure described in detail in Appendix A and already discussed in two progress reports (Kisiel et al. 2010, 2015). The complete experimental spectrum in the region 40−363 GHz was measured and was first broken down into over 220 correlated line sequences. The assignment reached by Koput (1986), was then transferred to the sequences extending over the entire covered frequency range. As a result, we derived a line list for the most relevant rotational transition sequences for CH₃NCO consisting almost entirely of experimental frequencies and free from any extrapolation (see Appendix A).

Our extensive spectroscopic coverage allows line-by-line comparison (see Table A.7) with the newly reported spectroscopic data in Halfen et al. (2015), where the Koput sub-40 GHz data is extended by 21 laboratory measurements in the region 60−87.5 GHz. We find good agreement between our data and the Halfen et al. (2015) data for 13 out of the 21 reported lines, and this includes the K_a = 0,1, m = 0 lines crucial to the Sgr B2(N) detection in their work. At the same time we note that seven lines show intriguing frequency differences of very close to 1 MHz between these two data sets. There is also a discrepancy in assignment concerning the line at 77 107.478 MHz, which according to our analysis is K = −2, m = 2 and not K = +1m = 1, and its upper level is 91 K and not 44 K. This limited agreement suggests that extrapolations made in Halfen et al. (2015) may be rather uncertain. The worst case is provided by the predicted line at 102 829.29 MHz searched for in Halfen et al. (2015), since it turns out that there is no line in the laboratory spectrum within ±10 MHz. Our comprehensive study of the laboratory spectrum of CH₃NCO showed that, even with self-consistent data, multi m-state fitting of this spectrum to experimental accuracy, and thus reliable predictions, are not yet possible. We emphasize again that the line list of 1269 lines up to 362 GHz that we report in Appendix A covers the frequency region relevant to astrophysical searches by means of actual laboratory measured frequencies.

3. Results

Once the key transitions of CH₃NCO were assigned in the laboratory, we used the resulting line list to search for this species in Orion KL. Of the 521 lines of CH₃NCO expected in the frequency coverage of the IRAM 30 m data (80−280 GHz), 236 are found to be unblended with other features, and 163 are partially blended but still identifiable in the combined line profiles with the contaminating lines. The remaining 122 lines are completely blended with strong features of one or more of the most abundant species, most of them above 200 GHz where the line density in Orion becomes very considerable. None of the unblended or partially blended features is missing. Figure 2 shows a selected number of lines from our line survey and from the ALMA Science Verification (SV) data of Orion (see also Fig. B.1 which shows all 399 detected lines). The ALMA data between 213.7 and 246.7 GHz mitigates line blending owing to the high spatial resolution that distinguishes between the contributions from the different hot cores of Orion. Hence, many features that are fully contaminated in the IRAM data do appear clearly in the ALMA observations. The dipole moment of CH₃NCO is high, 2.88 D (Kasten & Dreizler 1986), but its partition function at 150 K is also very high (≃3.0 × 10⁴, see Appendix A). As a result the lines appear weaker than for species with similar abundances and dipole moments. Nevertheless, the current detection of CH₃NCO in space is based on the largest data set of convincingly assigned lines for a previously unobserved molecule. Moreover, the assigned lines correspond to the strongest unidentified features we have in our data. In the 30 m and ALMA SV data the lines are strikingly prominent (see Figs. 2 and B.1).

In order to derive the relative abundance of CH₃NCO with respect to other organic molecules found in the comet, we used the MADEX code (Cernicharo 2012) to model all the lines that we detected with the IRAM 30 m radio telescope and the ALMA interferometer. The spatial distribution of these species, as observed with the ALMA data, together with that of CH₃NCO, is shown in Figs. 3 and B.3. At first glance the spatial distribution of CH₃NCO looks similar to that of the other species with two emission peaks, A and B symmetrically placed with respect to source n. However, a detailed inspection of these figures shows significant differences in the extent of their spatial distribution and in the position of the maxima. The most similar spatial distributions are those of CH₃NCO, HNCO, NH₂CHO, and CH₃CN. The species HDO, which traces water vapour, also shows A and B cores and is used to derive relative abundances with respect to H₂O (see Appendix B). The remaining molecules have their emission peaks at different positions and with a different spatial extent (see Appendix B). Similar discrepancies in the spatial distribution of other molecular species in Orion have been reported (see Appendix B). Spatial correlation between species with similar chemical routes could help in assessing this new detection. In Appendix B we discuss various aspects regarding the complexity of this source and, in addition, very little is known about the formation routes of CH₃NCO so the suspicion that this species could be related to HNCO or NH₂CHO is still very tentative. The main support that the maps provide for the detection is the correlation between the two CH₃NCO maps that depict the distribution of the molecule for two different internal rotation states (m = 0 and m = 1; see Figs. 3 and B.3), which shows that the lines arise cospatially removing the possibility of their origin from different species. In Appendix B we discuss the spatial distribution of the species shown in Figs. 3 and B.3.

Because of the different angular resolution of our data, two sets of models (the methods are described in Appendix B) have been run in order to fit the molecular emission of all these species. We used one model for the single-dish data, which provides a convolved view of the source structure with the angular resolution of the 30m telescope that ranges from 10′′ to 27′′, and a different model for the ALMA data, which have an angular resolution of 1$\mathrm{.}{\mathrm{\prime \prime }}^{}$8 × 1$\mathrm{.}{\mathrm{\prime \prime }}^{}$8 (see Appendix B). The methyl isocyanate lines are reasonably well fitted with a kinetic temperature of 150 K. The column densities obtained from the ALMA data are given in Table B.1, which also reproduces the relative abundances of the species found by COSAC. The CO abundance in the comet was poorly determined by COSAC (Goesmann et al. 2015) and we have referred all abundances to H₂O as in the COSAC results. Table B.1 shows that CH₃NCO in the Orion clouds is 400, 15, 10, and 5 times less abundant than CH₃OH, HNCO, CH₃CN, and CH₃CH₂CN, respectively. However, CH₃NCO is more abundant than amines, aldehydes, and acetones.

4. Discussion

We attempted to detect two isomers of CH₃NCO: CH₃CNO (acetonitrile oxide) and CH₃OCN (methyl cyanate). The energy difference of these isomers relative to the most stable one (CH₃NCO) is 8785 cm^-1 (25.1 kcal mol^-1) for CH₃OCN and 20 090 cm^-1 (57.4 kcal mol^-1) for CH₃CNO (Pasinazki & Westwood 2001). CH₃CNO has been characterized in the laboratory at millimetre wavelengths (Winnewisser et al. 1982). We derive an upper limit for its column density in Orion of ≤6 × 10¹³ cm^-2, i.e., a factor of ≃100 below that for CH₃NCO (see Table B.1). The experimental rotational spectra of methyl cyanate were recorded recently in the millimetre wave domain from 130 to 350 GHz and more than 300 internal rotation A-E doublets were successfully analysed (Kolesniková et al., in prep.). From the predicted spectrum we also derive an upper limit to its column density of ≤1 × 10¹⁴ cm^-2. Hence, only the most stable isomer, CH₃NCO, seems to be present in Orion. Tercero et al. (2013) found a negative detection of three isomers of methyl acetate and ethyl formate in Orion KL. López et al. (2014) also found upper limits to the abundance of the isocyanide isomers of ethyl and vinyl cyanide. A large discussion of the importance of bonding energy differences in the abundances of interstellar isomers was provided by Remijan et al. (2005). López et al. (in prep.) address the detection and abundances of two isomers of methyl formate in Orion KL.

We have searched for CH₃NCO toward a cold prestellar core, B1-b, in which a large variety of complex organic molecules, some of them typical of hot cores, have been found: CH₃O, CH₃OCOH, CH₃OCH₃, CH₃CHO, and CH₃SH. The abundances of these species relative to H₂, X≃ 10^-11 (Cernicharo et al. 2012), are much lower in B1-b than in Orion. Using the data from Cernicharo et al. (2012) we have obtained an upper limit for the column density of CH₃NCO in B1-b of ≤2 × 10¹¹ cm^-2 (X≤ 2 × 10^-12).

CH₃NCO has also been found towards Sgr B2 by Halfen et al. (2015). They derive a rather low rotational temperature, T_rot ≃ 24 K. However, we have searched for CH₃NCO in other public line surveys of Sgr B2 (Belloche et al. 2013) and found clear identification of methyl isocyanate in the warm gas of Sgr B2, with T_K ≃ 200 K and N(CH₃NCO) = 5 × 10¹⁷ cm^-2 for the 63 km s^-1 component and 3 × 10¹⁷ cm^-2 for the second component at 73 km s^-1 (see Appendix B). Using our laboratory frequencies, partition function and energy of the levels we have been able to identify 76 lines (43 unblended and 33 partially blended) in the 3 mm Sgr B2(N) data of Belloche et al. (2013). Several lines claimed missing in Halfen et al. (2015) are in fact well detected, but at ≃10 MHz from the predictions they use (see Fig. B.2). Most of the lines shown in Fig. B.2 have not been detected by Halfen et al. (2015) since they correspond to CH₃NCO transitions with K ≥ 2 and/or m ≥ 2. Assuming the column densities derived by Belloche et al. (2013) for the hot component at 200 K and v_LSR ≃ 63 km s^-1, we derive N(CH₃OH)/N(CH₃NCO) ≃ 40, N(HNCO)/N(CH₃NCO) ≃ 40, and N(CH₃CN)/N(CH₃NCO) ≃ 50, which are different from the values we derived for Orion KL (see Table B.1): 400, 15, 10, and 225, 15, 25, in Positions A and B, respectively. The measured difference in the abundance ratio between the two regions again points to a combination of physical and chemical differences based on the evolutionary states of the clouds found in the Galactic centre and outer disk of the Galaxy.

Nothing is known about the formation of methyl isocyanate in the gas phase or on the surface of the grains in molecular clouds. Some gas phase reactions that could form CH₃NCO have been discussed by Halfen et al. (2015). However, the high abundance for this species presently found in Orion and Sgr B2, together with the fact that it is detected in hot cores and not in cold dark clouds, also point towards a chemistry conducted in the grain mantles. Dedicated laboratory experiments on ices have to be performed in order to learn about the formation processes of CH₃NCO. In the ices CH₃NCO could be formed by methylation of HNCO (Goesmann et al. 2015). It is well known that at ambient temperature methyl isocyanate reacts with many substances that contain N−H or O−H groups and with water, which are common in the gas phase of Orion. Hence, the observed differences in molecular abundances between the gas phase and in the comet ice, if the comet surface still keeps pristine dust grains, imply a very rich chemistry in the gas phase for the molecules escaping the ices. Molecular abundances in Orion could result from the chemical history of the ejected material from the grains. It will be of great interest to observe the coma of the comet in order to obtain the abundances of the gas phase species and information on how the COSAC identified molecules survive ejection from the comet surface.

---

Acknowledgments

The Spanish authors thank MINECO for funding support from the CONSOLIDER-Ingenio program “ASTROMOL” CSD 2009-00038, AYA2012-32032, CTQ 2013-40717 P, CTQ 2010-19008, and the ERC synergy grant ERC-2013-Syg-610256-NANOCOSMOS. They also thank Junta de Castilla y León under grants VA070A08 and VA175U13. Z.K. acknowledges a grant from the Polish National Science Centre, decision number DEC/2011/02/A/ST2/00298. J.-C.G. and J.C. thank the ANR-13-BS05-0008-02 IMOLABS. J.-C.G. thanks the Program PCMI (INSU-CNRS) and the Centre National d’Études Spatiales (CNES) for funding support. The OSU authors acknowledge funding from NASA, NSF, and the Army Research Office.

References

Appendix A: Spectroscopy of CH₃NCO

Thorough analysis of the rotational spectrum of a target molecule is a crucial step that enables astrophysical detection. In the case of CH₃NCO the understanding of the rotational spectrum, and thus delivery of suitable line lists for astrophysical use has been considerably delayed by the complications arising from the presence of two large amplitude internal motions in this molecule. The first laboratory studies were carried out around half a century ago (Curl et al. 1963; Lett et al. 1967) when it was found that the molecule has a very low barrier to internal rotation of the methyl group and also a low-frequency bending mode. In the 1980s those studies were greatly extended by Koput (1984, 1986, 1988). His use of Stark spectroscopy allowed the unambiguous assignment of transitions for different values of K. The use of a semi-rigid rotor Hamiltonian and relative intensity considerations allowed the assignment of the vibrational quantum numbers m and ν_b. In this way a very self-consistent picture of the complex vibrational energy structure was reached, as already described in connection with Fig. 1. In an investigation contemporary with that of Koput, the electric dipole moment of CH₃NCO was also precisely determined: μ_a = 2.882 (8) D (Kasten & Dreizler 1986).

Further progress in the investigation of the rotational spectrum of methyl isocyanate was achieved as part of this work. In a first stage a practically continuous 117−364 GHz of the molecule was recorded by means of the Fast Scanning Submillimeter Spectroscopic Technique (FASSST) developed at The Ohio State University (Petkie et al. 1997; Medvedev et al. 2004). The spectrum was recorded at room-temperature using a commercially available sample. Although the Koput (1986) analysis was very successful in accounting for the consequences of the large amplitude motions on the rotational spectrum of this molecule, the deviation of the fit to 210 transitions up to J = 4 ← 3 was 3 MHz, which was well in excess of the 0.05 MHz frequency measurement precision of his spectrometer. Furthermore, even a cursory inspection of the FASSST spectrum (see Fig. 1) revealed that it was devoid of easily discernible spectral patterns and that progress in its traditional interpretation would be difficult. We have, therefore, decided to pursue a pragmatic approach to the characterization of this spectrum based on the generic property of rotational spectra, in that evolution of the frequencies of rotational transitions of a given set of quantum numbers can be described by a smooth function in the J quantum number. It was eventually possible to identify over 220 such line sequences and to fit each sequence to within experimental accuracy with a linear rotor-type model using a power series expansion in J(J + 1), as discussed in a preliminary report in 2010 (Kisiel et al. 2010). The separate line sequences were still without a confident K, m, and ν_b quantum number assignment. An initial attempt to transfer such assignments from the Koput work by linking sequences in the FASSST region with those below 40 GHz revealed many ambiguities, due to a still considerable region of missing experimental coverage of the rotational spectrum of CH₃NCO.

For this reason, in a second stage of the laboratory part of the present work, the missing 40−117 GHz region of the spectrum was covered in Valladolid by using two complementary spectrometers. The 50−117 GHz segment of the spectrum was recorded with a harmonic generation spectrometer based on the multiplication of the synthesizer frequency of up to 20 GHz and employing source modulation detection (Daly et al. 2014). The J = 5 ← 4 transition at 41−45 GHz was measured with a Stark modulation spectrometer (Daly et al. 2015) at a modulation voltage of 430 V. The sample was synthesized using the method described by Brändström et al. (1974) and in both spectrometers was kept at room temperature and a pressure of 20 μbar. As a result of the combined effort from the participating laboratories we were able to achieve complete room temperature experimental coverage of the rotational spectrum of methyl isocyanate from the lowest J = 1 ← 0 transition up to the J = 41 ← 40 transition centred near 358 GHz. The analysis procedure first involved creating a single spectrum by combining all available experimental spectra. The intensities for the lines measured by Koput are not available so that the <40 GHz spectral segment was simulated by assuming equal intensity transitions at the reported frequencies. Measurements on this spectrum were then performed with the AABS package for Assignment and Analysis of Broadband Spectra (Kisiel et al. 2005, 2012), which is freely available from the PROSPE website (Kisiel 2001a,b). The fits and predictions were made with Pickett’s SPCAT/SPFIT program suite (Pickett 1991) and each line sequence required its own dedicated set of analysis files. In the case of explicit asymmetric rotor fits Watson’s A-reduced asymmetric rotor Hamiltonian (Watson 1977) was used. The Loomis-Wood type mode for displaying the rotational spectrum (as pioneered for rotational spectroscopy by Winnewisser et al. 1989 and built into the AABS package) was used for graphical assignment of transitions in specific line sequences. Some preliminary results were reported in the middle of last year (Kisiel et al. 2015) and Loomis-Wood plots for several sample transition sequences are reproduced in Fig. A.1. These plots allow the transfer of the Koput assignment to the higher frequency spectrum. We also consider the Koput assignment to be reliable since it was reached using variable voltage Stark modulation spectroscopy, which provides good discrimination between transitions with different values of K according to their different line profiles. Each of these identified line sequences is reproduced by a suitable J(J + 1) power series expansion for the rotational energies $\begin{array}{ccc}{\mathit{E}}_{\mathrm{rot}}& \mathrm{=}& \mathit{BJ}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}\mathrm{)}\mathrm{-}{\mathit{D}}_{\mathit{J}}{\mathit{J}}^{\mathrm{2}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{2}}\mathrm{+}{\mathit{H}}_{\mathit{J}}{\mathit{J}}^{\mathrm{3}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{3}}\\ & & \mathrm{+}{\mathit{L}}_{\mathit{J}}{\mathit{J}}^{\mathrm{4}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{4}}\mathrm{+}{\mathit{P}}_{\mathit{J}}{\mathit{J}}^{\mathrm{5}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{5}}\mathrm{+}{\mathit{P}}_{\mathrm{12}}{\mathit{J}}^{\mathrm{6}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{6}}\\ & & \mathrm{+}{\mathit{P}}_{\mathrm{14}}{\mathit{J}}^{\mathrm{7}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{7}}\mathrm{+}{\mathit{P}}_{\mathrm{16}}{\mathit{J}}^{\mathrm{8}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{8}}\mathrm{+}{\mathit{P}}_{\mathrm{18}}{\mathit{J}}^{\mathrm{9}}\mathrm{(}\mathit{J}\mathrm{+}\mathrm{1}{\mathrm{)}}^{\mathrm{9}}\end{array}$(A.1) where the length of the expansion and the magnitudes of the expansion parameters B, D_J, H_J, etc., are determined by the interactions and also by the asymmetry contributions affecting the energy levels in question. The required length of the expansion ranges from only three terms to being insufficient even if all nine terms in Eq. (A.1) are used. In several rare cases two power expansions were needed to model a line sequence with a particularly complex J dependence, by splitting the sequence into two parts. The results of such fits for the key assigned transition sequences used in the present astrophysical identifications are listed in Table A.1. In labelling the sequences we conform, as closely as possible, to the notation of Koput (1986) in order to ensure traceability. There are only two differences where the original notation was ambiguous and was modified. In the case of the ground state, ν_b = 0, m = 0, for which asymmetric rotor notation would normally be more appropriate, we modified Koput’s notation to K = 1L, 1U, etc., to distinguish between lower and upper frequency sequences for the same value of K_a. Similarly, we used the notation m = 3L and 3U in order to distinguish between those of the two nearly degenerate m = 3 substates which give rise to lower and upper frequency rotational transition sequences.

The series fits and the Loomis-Wood displays were an invaluable aid to the analysis that allowed the identification of sequences of lines, extending up from the lowest J rotational transitions. It should be stated, however, that the transition frequencies for CH₃NCO reported in the line lists below are actual experimental frequencies. The power expansion fits were used to interpolate between the measured lines only in several cases of transitions missing in the spectrum for technical reasons. It is anticipated that these expansions may have useful predictive properties outside the present data region, but only for those series for which the expansion containing a moderate number of expansion terms is sufficient.

Appendix A.1: Spectroscopic constants and the partition function

The line lists for astrophysical detection require not only reliable line frequencies, but also line intensities that ultimately have a bearing on the derived column densities. Calculation of line intensities is crucially dependent on the rotational partition function, Q_rot, and CH₃NCO also poses a considerable inconvenience in this respect. For asymmetric rotor molecules with vibrational modes at relatively high frequencies the ground vibrational state rotational partition function can be usefully approximated with $\begin{array}{ccc}{\mathit{Q}}_{\mathrm{rot}}\mathrm{=}\mathrm{5.3311}\mathrm{\times }{\mathrm{10}}^{\mathrm{6}}\mathrm{(}{\mathit{T}}^{\mathrm{3}}\mathit{/}\mathrm{(}\mathit{ABC}\mathrm{)}{\mathrm{)}}^{\mathrm{1}\mathit{/}\mathrm{2}}\mathit{,}& & \end{array}$(A.2)where A, B, and C are the rotational constants for the molecule (in MHz) and T is the temperature (in K). Assumption of the rotational partition function for just the ground state of the molecule as the total partition function and neglect of population in excited vibrational states is normally associated with a relatively small underestimate, especially for low-temperature environments. For CH₃NCO this is not the case owing to the plethora of low-lying internal rotation substates, and it is necessary to consider the vibration-rotation partition function, Q_vr, explicitly. Furthermore, the difficulties in fitting more than a relatively small subset of measured transitions affect even the determination of the rotational constants. We summarize the situation in Table A.2, which demonstrates that even with a considerable reduction in the range of K_a values for the fitted transitions, the fit is less satisfactory than is usually the case with rotational spectra. Inclusion of measured transitions for all measured values of J requires the use of a rather large number of centrifugal distortion constants and values of some of these constants are also of considerable magnitude. This points to the very effective nature of such constants and to the considerable vibration-rotation interactions in the CH₃NCO molecule. The deviation of fit is also somewhat greater than the experimental accuracy. With these reservations in mind, we find rather good agreement between rotational constants resulting from the two fits based on very different subsets of the current data (the two central columns of Table A.2). We therefore use the combined fit of the indicated limited data (the rightmost column of Table A.2) as the source of rotational constants for evaluation of the partition function. The complete results of that fit are reported in Table A.3.

The second issue affecting the evaluation of the partition function is the summation over the energy levels. We used the SPCAT program to carry out this summation numerically, and the results are summarized in Table A.4. The nearly free internal rotor levels m = 1,2,4,5 are closer in character to the symmetric top case so that we used rotational constants A and (B + C)/2 in the partition function calculation, while for the remaining sublevels (m = 0 and m being a multiple of 3) we used separate values of A, B, and C (all from the final column of Table A.2). It can be seen in Table A.4a that at room-temperature and at a specimen astrophysical temperature of 120 K the partition function Q_vr increases significantly on addition of successive vibrational substates. It is only upon reaching vibrational energy of 400 cm^-1 above the ground state that a moderate saturation of the value of Q_vr becomes apparent. This corresponds to the consideration of energy levels in the ground state and the 17 vibrational substates immediately above it.

Table A.4b further illustrates some technical aspects associated with the evaluation of Q_vr such as the necessity of using sufficiently broad limits on the values of the J and K quantum numbers. It turns out that the large values of centrifugal distortion constants from the effective fit for K ≤ 2 lead to unphysical compression of energy levels at the highest values of J and K, as highlighted by specific diagnostics from the SPCAT program. At the same time, the details of the summation have little effect on the estimate of the low temperature partition function. For this reason we settle on the “recommended” conditions, which pass the test that the value from the numerical summation for just the ground state is close to that from the analytical formula in Eq. (A.2).

The final result of these considerations is that at room-temperature it is necessary to use a partition function value that is almost 8 times greater than that for just the ground state, while at 120 K this ratio is closer to 4. It is clear that a reliable estimate of the partition function to use for the calculation of transition intensities will have a direct bearing on the derived astrophysical abundances of CH₃NCO.

Appendix A.2: Line lists

The final results of the analysis of the laboratory rotational spectrum of CH₃NCO for use in astrophysical applications are presented in Tables A.5 and A.6. The experimental frequencies for 27 rotational transition sequences with the lowest vibrational energy are listed in Table A.5. Five more specimen transition sequences corresponding to higher vibrational energies and presenting a challenge to confident astrophysical detection are also included. This line list is also given in Table A.6 in the standard format of the JPL catalog (Pickett et al. 1998). Table A.6 contains line intensities and lower state energies so that it can be easily converted to a different temperature. Table A.6 constitutes the main tool for astrophysical detection and characterization of CH₃CNO, and we summarize the key parameters used in its evaluation: μ = μ_a = 2.882 D, T = 300 K, and Q_vr = 138 369 calculated by summing over rotational levels limited by K< 11 and J< 100 for all vibrational states up to 402.5 cm^-1 above the ground state.

We have already commented above on the key discrepancies between the experimental frequencies reported in Halfen et al. (2015) and those in this work as summarized in Table A.7. We further note that if we replace our frequencies in the fit for the ground state reported in Table A.3 by the three problematic m = 0 lines at 60 160.932, 68 753.890, and 70 009.179 MHz (see Table A.7), then those lines are incompatible with this fit at obs.-calc. frequencies of near −1.1 MHz. Our frequencies for these transitions have obs.-calc. values of less than 0.03 MHz, in a fit encompassing 201 pure rotational transition frequencies. In fact, the discrepancy of 1 MHz from a line sequence comprising around 40 lines and consisting of both Koput (1986) and our mm-wave lines would be immediately visible. This is clearly apparent in Fig. A.1 (second column from the left), where the 70010.208 MHz line is in good agreement with the sequence defined by a low order J(J + 1)-type fit, whereas a frequency of 70 009.179 MHz is clearly incompatible.

Finally, we note that the quantization used in Halfen et al. (2015) for the m = 0, K_a = 1 transitions deviates from standard practice. While the quantization for the m> 0 lines is somewhat arbitrary and largely dependent on the fitting program, the quantization for the ground state is well established. In the case when the splitting between the two a-dipole ΔJ = 1, K_a = 1 transitions for a given value of J′′ is dominated by the asymmetry term in the rotational Hamiltonian (as is the case for CH₃NCO) it is the K_a = 1 transition with the higher value of K_c that is at lower frequency, and the one with lower value of K_c that is at higher frequency. This quantization leads to the moderately successful effective fit for the ground (m = 0) state reported by us in Table A.3. On the other hand, in Halfen et al. (2015) the K_c quantization for m = 0, K_a = 1 transitions is systematically reversed.

Appendix A.3: Nuclear quadrupole hyperfine splitting

It should be noted that the lowest-J rotational transitions of CH₃NCO are subject to nuclear quadrupole hyperfine splitting, and that the line lists discussed above provide in such cases only the average, hyperfine-free frequencies. The hyperfine splitting arising from the presence of the ¹⁴N nucleus in the CH₃NCO molecule was studied in previous works (Kasten & Dreizler 1986; Curl et al. 1963; Lett et al. 1967). We have now, for the first time, observed such lines at the very low temperature of supersonic expansion using two spectroscopic techniques. CH₃NCO was first investigated by broadband chirped-pulse Fourier transform microwave (CP-FTMW) spectroscopy (Brown et al. 2008) in two different frequency ranges: 6−18 and 25−26 GHz (Mata et al. 2012). A gas mixture containing 0.3% of CH₃NCO in neon at the backing pressure of 3 bar was used for the pulsed jet expansion. These scans allowed explicit confirmation of the assignment of the ground state (ν_b = 0, m = 0) lines since these lines (and some of the lowest ν_b = 0, m = 1 substate lines) are the only lines expected to be visible at these conditions. In order to resolve the nuclear quadrupole hyperfine structure, further measurements were performed by supersonic expansion, Fabry-Perot cavity, and Fourier transform microwave (FP-FTMW) spectroscopy (Grabow et al. 1996) using two different configurations covering the frequency range from 4 to 18 GHz (Alonso et al. 1997; Bermúdez et al. 2014). Figure A.2 shows the experimentally recorded hyperfine patterns for the first two lowest-J transitions of CH₃NCO and it is clear that at sub-MHz resolution consideration of such splitting is important. Table A.8 lists hyperfine component frequencies for all K ≤ 2 rotational transitions up to 80 GHz. This prediction has an estimated accuracy of around 5 kHz and was made by using the rightmost set of spectroscopic constants from Table A.2. It is apparent that the central ΔF = + 1 components quickly coalesce, while the outer ΔF = 0 components rapidly lose intensity.

Appendix B: Astrophysical methods

Appendix B.1: The Orion BN/KL region

Orion BN/KL (Becklin-Neugebauer/Kleinmann-Low; Becklin & Neugebauer 1967; Kleinmann & Low 1967) is the closest (about 414 pc; Menten et al. 2007) region exhibiting several processes related to young stellar objects (YSOs) and high-mass star formation (Genzel & Stutzki 1989). This region is located at the core of the Orion molecular cloud 1 (OMC1) which lies behind the Orion Nebula cluster (O’Dell 2001). Infrared facilities and radio telescopes can observe inside dusty regions where the formation of stars or their associated violent phenomena take place, providing very useful information about these processes. Near- and mid-IR subarcsecond resolution imaging and (sub)millimetre interferometric observations have identified the main sources of luminosity, heating, and dynamics in the region. At first, IRc2 was proposed as the main source of luminosity, heating, and dynamics within the region (Wynn-Williams et al. 1984). However, the detection of two radio continuum sources, B (coincident with the BN object which was resolved with high spatial resolution at 7 mm; Rodríguez et al. 2009) and I (centroid of the SiO masers and located 0.5′′ south of IRc2; Menten & Reid 1995) revealed that the intrinsic IR luminosity of IRc2 (L ~ 1000 L_⊙) is only a fraction of the luminosity of the entire system (1 × 10⁵L_⊙; Gezari et al. 1998). The radio continuum emission of the IR source n was also identified by Menten & Reid (1995) (3′′ south-west from source I), suggesting this source as another precursor of the large-scale phenomena. In addition, several IR and millimetre continuum sources have been identified (Shuping et al. 2004; Wu et al. 2014). Thus, the core of Orion KL contains at least three self-luminous objects (protostars), the compact HII regions I, n (see Figs. 3 and B.3), and BN; these sources are within a region of ~10′′ (~0.02 pc). Their proper motions reveal that they run away from a common region (Gómez et al. 2005). Different scenarios, as well as the formation of high-mass stars, have been proposed to explain the complexity of this source. An explosion of a multistar system (sources I, n, and BN) that took place ~500 years ago has been proposed as the main factor responsible for most of the Orion KL gas components (Zapata et al. 2011).

The different gas components show distinct physical and chemical properties, which can be identified even with single-dish telescopes by a characteristic systemic velocity (Blake et al. 1987; Schilke et al. 2001; Tercero et al. 2010). Classically, these components have been called Hot Core, a hot, ~200−300 K, and dense clump rich in complex organic saturated N-bearing species such as CH₃CH₂CN, characterized by Δv_FWHM ~ 5−15 km s^-1 and v_LSR ~ 5−7 km s^-1; Extended Ridge, the host, quiescent, and relatively cold, ~60 K ambient cloud rich in simple species such as CS or HCN, emitting lines with Δv_FWHM ~ 3−4 km s^-1 and v_LSR ~ 8−9 km s^-1; Compact Ridge, a warm, ~150 K, and compact clump rich in organic saturated O-rich species such as CH₃OCOH or CH₃OCH₃, whose spectral features are characterized by Δv_FWHM ~ 2−3 km s^-1 and v_LSR ~ 7−8 km s^-1; and Plateau, molecular outflows presenting typical shock chemistry with molecules such as SO or SiO (Goicoechea et al. 2015); the low velocity flow (oriented along the SW-NE direction and probably driven by source I) is characterized by lines with Δv_FWHM ~ 20 km s^-1 and v_LSR ~ 5−6 km s^-1, whereas the high-velocity flow (oriented along the SE-NW direction and whose origin could be the explosive event caused by the dynamical decay of a multi-star system briefly described above) presents lines with Δv_FWHM as wide as 150 km s^-1 and v_LSR ~ 10 km s^-1. Interferometric observations have shown other cloud components inside Orion BN/KL such as the millimetre continuum sources MM2, MM4 (which corresponds approximately to Position B, see Figs. 3 and B.3), MM5, and MM6 (Wu et al. 2014). In addition, these latter observations reveal the hot core to be a collection of very dense and hot clumps presenting different chemistry and radial velocities (Favre et al. 2011; Tercero et al. 2015). Position A of Figs. 3 and B.3 is located in the middle of the hot core clumpy structure.

Appendix B.2: Observations

Appendix B.2.1: IRAM 30 m

We performed a sensitive, fully single side band reduced, line survey towards Orion IRc2 with the IRAM 30 m telescope over broad frequency ranges (total frequency coverage of ~168 GHz; Tercero et al. 2010). Our main goal was to obtain a deep insight into the molecular content and chemistry of Orion BN/KL and to improve our knowledge of its prevailing physical conditions. It also allows us to search for new molecular species and new isotopologues, as well as the rotational emission of vibrationally excited states of molecules already known to exist in this source (see Sect. B.3 and main text).

The line survey was performed over three millimetre-wave windows (3, 2, and 1.3 mm, covering frequency ranges 80−115.5 GHz, 130−178 GHz, and 197−281 GHz, respectively) with the IRAM 30 m telescope. The observations were carried out between September 2004 and January 2007 pointing toward the IRc2 source at α_2000.0 = 5^h35^m14$\mathit{.}{}^{\mathrm{s}}$5, δ_2000.0 = −5°22′30$\mathrm{.}{\mathrm{\prime \prime }}^{}$0. Four SiS receivers were used simultaneously, with image sideband rejections within ~13 dB (1.3 mm receivers), 12−16 dB (2 mm receivers), and 20−27 dB (3 mm receivers). The observations were performed using the wobbler switching mode with a beam throw in azimuth of ±120′′. No contamination from the off position affected our observations, except for a marginal amount at the lowest elevations (25°) for molecules showing low-J emission along the ambient molecular cloud. System temperatures were in the range of 100−800 K from the lowest to the highest frequencies. For the spectra between 172−178 GHz, the system temperature was significantly higher, 1000−4000 K, owing to proximity of the atmospheric water line at 183.31 GHz. The intensity scale was calibrated using the atmospheric transmission model (ATM; Pardo et al. 2001; Cernicharo 1985). Focus and pointing were checked every 1−2 hours on planets or nearby quasars. Backends provided a spectral resolution of 1−1.25 MHz. The half power beam width (HPBW) of the 30m telescope ranged from 31′′ to 9′′ from 80 to 280 GHz (HPBW[arcsec] = 2460/Freq.[GHz]). The data were reduced using the GILDAS package¹ checking for image sideband contamination and fitting and removing 0−1 order baselines. Figures using these data are shown in main beam temperature (T_MB).

Appendix B.2.2: ALMA SV

The ALMA Science Verification (SV) data were taken in January 2012 towards the IRc2 region in Orion. The observations were carried out with 16 antennas of 12 m in the frequency range from 213.715 to 246.627 GHz (Band 6). The primary beam was ~27′′. Spectral resolution was 0.488 MHz (~0.64 km s^-1 in the observed frequency range). The observations were centred on coordinates: α_J2000 = 05^h35^m14$\mathit{.}{}^{\mathrm{s}}$35, δ_J2000 = −05°22′35$\mathrm{.}{\mathrm{\prime \prime }}^{}$00. The CASA software² was used for initial processing, and then the visibilities were exported to the GILDAS package for further analysis. The synthesized beam ranged from 2$\mathrm{.}{\mathrm{\prime \prime }}^{}$00 × 1$\mathrm{.}{\mathrm{\prime \prime }}^{}$48 with a PA of 176° at 214.0 GHz to 1$\mathrm{.}{\mathrm{\prime \prime }}^{}$75 × 1$\mathrm{.}{\mathrm{\prime \prime }}^{}$29 with a PA of 164° at 246.4 GHz. The brightness temperature to flux density conversion factor is 9 K for 1 Jy per beam. The continuum emission was subtracted in the maps by carefully selecting line-free channels.

Appendix B.3: Astronomical model

Appendix B.3.1: Orion KL

Frequency predictions and intensities reported in this work were implemented in MADEX (Cernicharo 2012) to model the emission of CH₃NCO and to search for it toward Orion KL. MADEX is an excitation and radiative transfer code which includes more than 5200 molecular species with accurate spectroscopy. To derive the physical parameters shown in Table B.1, we performed a model that reproduces more line profiles more accurately from transitions covering a wide energy range within a ~30% of the uncertainty in line intensity (red lines in Figs. 2 and B.1).

Owing to the lack of collisional rates for this molecule, we used LTE approximation. Nevertheless, taking into account the physical conditions of the considered components of the cloud (see Table B.2), we expect that this approximation works reasonably well (see e.g. Fortman et al. 2012). For each spectral component, we assume uniform physical conditions of the cloud (line width, radial velocity, and source size with uniform brightness temperature and optical depth over this size), which we choose by taking into account the parameters obtained from Gaussian fits of the line profiles and the emission shown by the ALMA maps. To fit the IRAM 30 m lines, we also considered the beam coupling effects adding the offset position with respect to IRc2 (the pointing position). Corrections for beam dilution are also applied for each line depending on the different beam sizes at different frequencies. Therefore, we fixed all of the above parameters leaving only the rotational temperature and the column density as free parameters for each component. In order to determine the uncertainty of the values of column density and of temperature, we ran several models varying only the values for these parameters and fixing the rest. To fit the ALMA data we extracted the averaged spectrum over 5 × 5 pixels (~1′′× 1′′) around Position A and Position B (see Figs. 3 and B.3). To derive the column densities and relative abundances of the species shown in Table B.1, we adopted the same model of CH₃NCO for each position varying the column density for each one of the species of Table B.1. In Fig. B.1, a model containing all species already studied in the IRAM 30 m survey is included (cyan line). The molecules considered in the models and references to their modelling and analysis are given in Table B.3. This table also summarizes the works performed to date based on the 30 m data and devoted to mitigating the number of unidentified lines in the IRAM 30 m survey of Orion (especially those in collaboration with spectroscopy laboratories), to searching for new molecular species in interestelar clouds, and to improving the knowledge of the physical-chemical properties of Orion KL.

Appendix B.3.2: Sgr B2(N)

We have searched in published line surveys of Sgr B2 and W51 for CH₃NCO. We find that 21 unidentified features above 13σ found in Sgr B2 by Belloche et al. (2013) can be unambiguously assigned to CH₃NCO. However, in the PRIMOS data (Neill et al. 2012) of the same source the assignment is not obvious. CH₃NCO was also reported in Sgr B2 during the submission process of this work (Halfen et al. 2015). We used MADEX and the public data of Sgr B2(N) provided by Belloche et al. (2013) to derive the rotational temperature and column density of methyl isocyanate in this source (see Fig. B.2, red line). We find that the two hot (T_rot = 200 K) and compact (d_sou ≃ 2′′) components derived by Belloche et al. (2013) for HNCO are enough to reproduce all lines that appear at 3 mm. We derive column densities of N = 5 × 10¹⁷ cm^-2 for the 63 km s^-1 component and 3 × 10¹⁷ cm^-2 for the second component at 73 km s^-1 (assuming Δv_FWHM = 7 km s^-1, see Belloche et al. 2013).

In W51 (Pulliam et al. 2012), another well-known hot core with similar molecular content than Orion, we do not find an obvious detection in the data, because the lines are weaker owing to the higher distance of W51 compared to Orion.

Methyl isocyanate turns out to be one of the prominent molecules in hot cores and it is surprising to realize that it is present in many line surveys. It was only the lack of suitable laboratory data delayed its discovery. In the present work it is the close collaboration between astronomers and molecular physicists that has made such detailed characterization of this new molecular species possible.

Appendix B.4: Spatial distribution in Orion

The spatial distribution of several molecules detected in Orion KL is shown in Figs. 3 and B.3. Despite the lack of short spacing data required for completion of the possible extended emission of some of these species (see e.g. CH₃OH or DCN), we can obtain an overview of the main cores that host these species. Moreover, the discussion has to address the large differences in energies of the upper level for the transitions involved in those maps (it is not possible to find transitions with similar energies of, for instance, DCN and HC₃N ν₇ = 1 between 213.7−247.5 GHz, the coverage of the ALMA SV data). In addition, owing to the extreme complexity of Orion KL, it is not straightforward to find lines free of blending with other species in the whole region: many lines of a large variety of molecules that appear unblended in some cores, overlap with lines of other species in the average spectrum of the region. This makes it more difficult to derive maps of lines involving transitions with uniform upper level energies. Nevertheless, by taking these restrictions into account, the discussion can be focused on the main differences/similarities found for the different “families” of molecules.

As we have pointed out in the main text, we note a similar spatial distribution for CH₃NCO, NH₂CHO, and HNCO. The main differences of these maps are a larger extension of HNCO due to its higher abundance and the lack of the clump in the NW of Position B for NH₂CHO; in addition, CH₃NCO does not emit in this region, whereas HNCO and NH₂CHO show some emission at the position of the compact ridge. It is worth noting that these species are the only three detected in Orion KL containing nitrogen, oxygen, carbon, and hydrogen together.

Following with the compact ridge, this component is the main host of CH₃OCH₃, CH₃OCOH, and CH₂OCH₂, all of them organic compounds containing oxygen. Molecules that we could naively think that are related with them, such as CH₃CH₂OH, CH₃COCH₃, CH₃COOH, OHCH₂CH₂OH, and CH₂OHCHO, present a completely different spatial distribution (Guélin et al. 2008; Peng et al. 2012, 2013; Widicus Weaver & Friedel 2012; Friedel & Widicus Weaver 2012; Brouillet et al. 2015; Feng et al. 2015; Tercero et al. 2015). Although there are some remarkable differences in the emission of the latter species, we observe the main core located at the SE of source n (at the middle-south of the hot core). Abundant species in the region such as DCN, HDO, and CH₃CN do not show significant emission at the compact ridge.

On the other hand, complex species with nitrogen such as CH₃CH₂CN, CH₂CHCN, and HC₃N, present their emission at the north of the hot core. In addition, emission at the NE of source I is more prominent for these species (Guélin et al. 2008; Widicus Weaver & Friedel 2012; Friedel & Widicus Weaver 2012).

Finally, although we do not discuss the emission peaks of CH₃OH (methanol, one of the most abundant species), because they could be affected by opacity effects, we note that methanol is ubiquitous in the region.

All Tables

All Figures