A. Belloche¹ - K. M. Menten¹ - C. Comito¹ - H. S. P. Müller^1,2 - P. Schilke¹ - J. Ott^3,4,5 - S. Thorwirth¹ - C. Hieret¹

1 - Max-Planck Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
2 - I. Physikalisches Institut, Universität zu Köln, Zülpicher Str. 77, 50937 Köln, Germany
3 - National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903-2475, USA
4 - California Institute of Technology, 1200 E. California Blvd., Caltech Astronomy, 105-24, Pasadena, CA 91125-2400, USA
5 - CSIRO Australia Telescope National Facility, Cnr Vimiera & Pembroke Roads, Marsfield NSW 2122, Australia

Abstract
Context. Amino acids are building blocks of proteins and therefore key ingredients for the origin of life. The simplest amino acid, glycine (NH₂CH₂COOH), has long been searched for in the interstellar medium but has not been unambiguously detected so far. At the same time, more and more complex molecules have been newly found toward the prolific Galactic center source Sagittarius B2.
Aims. Since the search for glycine has turned out to be extremely difficult, we aimed at detecting a chemically related species (possibly a direct precursor), amino acetonitrile (NH₂CH₂CN).
Methods. With the IRAM 30 m telescope we carried out a complete line survey of the hot core regions Sgr B2(N) and (M) in the 3 mm range, plus partial surveys at 2 and 1.3 mm. We analyzed our 30 m line survey in the LTE approximation and modeled the emission of all known molecules simultaneously. We identified spectral features at the frequencies predicted for amino acetonitrile lines having intensities compatible with a unique rotation temperature. We also used the Very Large Array to look for cold, extended emission from amino acetonitrile.
Results. We detected amino acetonitrile in Sgr B2(N) in our 30 m telescope line survey and conducted confirmatory observations of selected lines with the IRAM Plateau de Bure and the Australia Telescope Compact Array interferometers. The emission arises from a known hot core, the Large Molecule Heimat, and is compact with a source diameter of 2 $\hbox{$^{\prime\prime}$ }$ (0.08 pc). We derived a column density of 2.8 $\times$ 10¹⁶ cm^-2, a temperature of 100 K, and a linewidth of 7 km s^-1. Based on the simultaneously observed continuum emission, we calculated a density of 1.7 $\times$ 10⁸ cm^-3, a mass of 2340 $M_\odot$ , and an amino acetonitrile fractional abundance of 2.2 $\times$ 10^-9. The high abundance and temperature may indicate that amino acetonitrile is formed by grain surface chemistry. We did not detect any hot, compact amino acetonitrile emission toward Sgr B2(M) or any cold, extended emission toward Sgr B2, with column-density upper limits of 6 $\times$ 10¹⁵ and 3 $\times$ 10^12-14 cm^-2, respectively.
Conclusions. Based on our amino acetonitrile detection toward Sgr B2(N) and a comparison to the pair methylcyanide/acetic acid both detected in this source, we suggest that the column density of both glycine conformers in Sgr B2(N) is well below the best upper limits published recently by other authors, and probably below the confusion limit in the 1-3 mm range.

Key words: astrobiology - astrochemistry - line: identification - stars: formation - ISM: individual objects: Sagittarius B2 - ISM: molecules

1 Introduction - general methods

Among the still growing list of complex molecules found in the interstellar medium, so-called ``bio''molecules garner special attention. In particular, the quest for interstellar amino acids, building blocks of proteins, has engaged radio and millimeter wavelength astronomers for a long time. Numerous published and unpublished searches have been made for interstellar glycine, the simplest amino acid (Ceccarelli et al. 2000; Jones et al. 2007; Combes et al. 1996; Hollis et al. 2003,1980; Brown et al. 1979; Berulis et al. 1985; Cunningham et al. 2007). Its recent ``detection'' claimed by Kuan et al. (2003) has been persuasively rebutted by Snyder et al. (2005). Since the early days of molecular radio astronomy, Sagittarius B2 has been a favorite target in searches for complex molecules in space.

1.1 The target: Sagittarius B2

Sagittarius B2 (hereafter Sgr B2 for short) is a very massive (several million solar masses) and extremely active region of high-mass star formation at a projected distance of $\sim$ 100 pc from the Galactic center. Its distance from the Sun is assumed to be the same as the Galactic center distance, R₀. Reid (1993), reviewing various methods to determine R₀, arrived at a ``best estimate'' of 8.0 $\pm$ 0.5 kpc, a value that we adopt in this article. It is supported by recent modeling of trajectories of stars orbiting the central black hole, which yields 7.94 $\pm$ 0.42 kpc (Eisenhauer et al. 2003).

There are two major centers of activity, Sgr B2(M) and Sgr B2(N) separated by $\sim$ 2 pc. In each of them, recent star formation manifests itself in a multitude of H II regions of many sizes, from hypercompact to compact (Gaume et al. 1995), and there is abundant material to form new stars evident by massive sources of molecular line emission and submillimeter continuum emission from dust (Lis et al. 1991,1993).

1.1.1 Sgr B2 as part of the Central Molecular Zone

Some of the first detections of interstellar organic molecules (at cm-wavelengths!) were made toward Sgr B2 (see Menten 2004, for a historical perspective). The low intrinsic line strengths make these cm lines unlikely candidates for detection. However, the situation is helped, first, by the fact that many of the transitions in question may have inverted levels (Menten 2004) and amplify background radio continuum emission which is very intense at cm wavelengths (Hollis et al. 2007). Second, the spatial distributions of many species are characterized by spatially extended emission covering areas beyond Sgr B2 itself, filling single dish telescope beams, thus producing appreciable intensity even when observed with low spatial resolution (Jones et al. 2008; Cummins et al. 1986). This emission is characterized by low rotation temperatures, favoring lower frequency lines. Recent identifications of ``new'' species include glycolaldehyde CH₂OHCHO (Hollis et al. 2001,2000), ethylene glycol HOCH₂CH₂OH (Hollis et al. 2002), and vinyl alcohol CH₂CHOH (Turner & Apponi 2001).

Sgr B2 and its surroundings are part of the Central Molecular Zone (CMZ) of our Galaxy, a $\sim$ $\pm 0\hbox{$.\!\!^\circ$ }3$ latitude wide band stretching around the Galactic center from longitude $l \sim +1\hbox{$.\!\!^\circ$ }6$ to $-1\hbox{$.\!\!^\circ$ }1$ (see, e.g., Morris & Serabyn 1996). The CMZ contains spatially extended emission of many complex organic molecules (Dahmen et al. 1997; Menten 2004; Requena-Torres et al. 2006; Minh et al. 1992).

1.1.2 The Large Molecule Heimat

Near Sgr B2(N), there is a hot, dense compact source that has a mm-wavelength line density second to no other known object. This source, for which Snyder et al. (1994) coined the name ``Large Molecule Heimat'' (LMH), is characterized by very high densities (>10⁷ cm^-3) and gas temperatures (>100 K). In recent years arcsecond resolution interferometry with the BIMA array has resulted in the detection and imaging of increasingly complex organic species toward the LMH, such as vinyl cyanide CH₂CHCN, methyl formate HCOOCH₃, and ethyl cyanide CH₃CH₂CN (Miao et al. 1995; Miao & Snyder 1997), formamide NH₂CHO, isocyanic acid HNCO, and methyl formate HCOOCH₃ (Kuan & Snyder 1996), acetic acid CH₃COOH (Mehringer et al. 1997; Remijan et al. 2002), formic acid HCOOH (Liu et al. 2001), and acetone (CH₃)₂CO (Snyder et al. 2002). All the interferometric observations are consistent with a compact (<few arcsec diameter) source that had already been identified as the source of high-density-tracing non-metastable ammonia line emission by Vogel et al. (1987) and thermal methanol emission by Mehringer & Menten (1997, their source ``i''). The LMH also hosts a powerful H₂O maser region (Reid et al. 1988), which provides evidence that it is very young (see Sect. 4.1).

1.2 The complex spectra of complex molecules

Complex molecules in general have large partition functions, in particular for the elevated temperatures (>100 K) in molecular hot cores, dense and compact cloud condensations internally heated by a deeply embedded, young high-mass (proto)stellar object. Therefore, most individual spectral lines are weak and might easily get hidden in the ``line forest'' found toward these frequently extremely line-rich sources. To a large part, this forest consists of rotational lines, many of them presently unidentifiable, from within relatively low-lying vibrational states of molecules. Most of the candidate molecules from which these lines originate are known to exist in these sources, but laboratory spectroscopy is presently lacking for lines from the states in question. At this point in the game, unequivocally identifying a species in a spectrum of a hot core covering a wide spectral range requires the following steps: as described in detail in Sect. 3.2, assuming Local Thermodynamic Equilibrium (LTE) (which applies at the high densities in hot cores) a model spectrum is calculated for an assumed rotation temperature, column density, line width and other parameters. This predicts lines of a given intensity at all the known frequencies. Then at least two conditions have to be fulfilled: (i) all predicted lines should have a counterpart in the observed spectrum with the right intensity and width - no single line should be missing; (ii) follow-up observations with interferometers have to prove whether all lines from the candidate species are emitted from the same spot. Given the chemical variety in hot core regions, this is a powerful constraint. Moreover, interferometer images tend to have less line confusion, since many lines that are blended in larger beam single-dish spectra arise from different locations or are emitted by an extended region that is spatially filtered out. Using an interferometer for aiding molecule identifications was pioneered by Snyder and collaborators who (mostly) used the Berkeley-Illinois-Maryland-Array (BIMA) to clearly identify a number of species in the Sgr B2(N) Large Molecule Heimat (see Sect. 1.1.2).

We carried out a complete line survey of the hot core regions Sgr B2(N) and (M) with the IRAM 30 m telescope at 3 mm, along with partial surveys at 2 and 1.3 mm. One of the overall goals of our survey was to better characterize the molecular content of both regions. It also allows searches for ``new'' species once we have identified the lines emitted by known molecules (including vibrationally and torsionally excited states). In particular, many complex molecules have enough lines in the covered frequency ranges to apply criterion (i) above. Once a species fulfils this criterion, interferometer measurements of selected lines can be made to check criterion (ii).

1.3 Amino acetonitrile

One of our target molecules was amino acetonitrile (NH₂CH₂CN), a molecule chemically related to glycine. Whether it is a precursor to the latter is under debate (see Sect. 4.3). Not many astronomical searches for amino acetonitrile have been reported in the literature. In his dissertation, Storey (1976) reported searches for the J_{K_a,K_c} = 2₁₁-2₁₂ and 1₀₁-0₀₀ transitions at 1350.5 and 9071.7 MHz, respectively with the Parkes 64 m telescope. On afterthought, the only chance of success for their observations would have been if amino acetonitrile existed on large spatial scales, similar to the molecules described in Sect. 1.1.1 (see Sect. 3.7 for further limits on extended amino acetonitrile emission). Recently, Wirström et al. (2007) reported unsuccessful searches of a number of mm-wavelength transitions of amino acetonitrile toward a number of hot cores.

Here, we report our detection of warm compact emission from amino acetonitrile in Sgr B2(N) with the IRAM 30 m telescope, the Plateau de Bure Interferometer (PdBI) and the Australia Telescope Compact Array (ATCA), and upper limits on cold, spatially extended emission from amino acetonitrile that we obtained with the NRAO Very Large Array (VLA). Section 2 summarizes the observational details. We present our results in Sect. 3. Implications in terms of interstellar chemistry are discussed in Sect. 4. Our conclusions are summarized in Sect. 5.

2 Observations and data reduction

2.1 Single-dish observations and data reduction

We carried out millimeter line observations with the IRAM 30 m telescope on Pico Veleta, Spain, in January 2004, September 2004 and January 2005. We used four SIS heterodyne receivers simultaneously, two in the 3 mm window connected to the autocorrelation spectrometer VESPA and two in the 1.3 mm window with filter banks as backends. A few selected frequency ranges were also observed with one SIS receiver at 2 mm in January 2004. The channel spacing and bandwidth were 0.313 and 420 MHz for each receiver at 3 and 2 mm, and 1 and 512 MHz for each receiver at 1.3 mm, respectively. The observations were done in single-sideband mode with sideband rejections of $\sim$ 1-3 $\%$ at 3 mm, $\sim$ 5-7 $\%$ at 2 mm, and $\sim$ 5-8 $\%$ at 1.3 mm. The half-power beamwidths can be computed with the equation HPBW ( $\hbox{$^{\prime\prime}$ }$ ) = $\frac{2460}{\nu({\rm GHz})}$ . The forward efficiencies $F_{{\rm eff}}$ were 0.95 at 3 mm, 0.93 at 2 mm, and 0.91 at 1.3 mm, respectively. The main-beam efficiencies were computed using the Ruze function $B_{{\rm eff}} = 1.2 \epsilon ~~ {\rm e}^{-(4 \pi R \sigma / \lambda)^2}$ , with $\epsilon = 0.69$ , $R \sigma = 0.07$ , and $\lambda$ the wavelength in mm (see the IRAM 30 m telescope system summary on http://www.iram.es). The system temperatures ranged from 96 to 600 K at 3 mm, from 220 to 720 K at 2 mm (except at 176 GHz where they ranged from 2400 to 3000 K), and from 280 to 1200 K at 1.3 mm. The telescope pointing was checked every $\sim$ 1.5 h on Mercury, Mars, 1757-240 or G10.62, and found to be accurate to 2-3 $\hbox{$^{\prime\prime}$ }$ (rms). The telescope focus was optimized on Mercury, Jupiter, Mars or G34.3+0.2 every $\sim$ 1.5-3 h. The observations were performed toward both sources Sgr B2(N) ( $\alpha_{{\rm J2000}}$ = 17 $^{\rm h}$ 47 $^{\rm m}$ 20 $\hbox{$.\!\!^{\rm s}$ }$ 0, $\delta_{{\rm J2000}}$ = $-28^\circ$ 22 $\hbox{$^\prime$ }$ 19.0 $\hbox{$^{\prime\prime}$ }$ , $V_{{\rm lsr}}$ = 64 km s^-1) and Sgr B2(M) ( $\alpha_{{\rm J2000}}$ = 17 $^{\rm h}$ 47 $^{\rm m}$ 20 $\hbox{$.\!\!^{\rm s}$ }$ 4, $\delta_{{\rm J2000}}$ = $-28^\circ$ 23 $\hbox{$^\prime$ }$ 07.0 $\hbox{$^{\prime\prime}$ }$ , $V_{{\rm lsr}}$ = 62 km s^-1) in position-switching mode with a reference position offset by ( $\Delta\alpha$ , $\Delta\delta$ ) = ( $-752\hbox{$^{\prime\prime}$ }$ , $+342\hbox{$^{\prime\prime}$ }$ ) with respect to the former. The emission toward this reference position was found to be weak: $T_{{\rm a}}^\star$ (¹²CO 1-0) $\sim$ 2 K, $T_{{\rm a}}^\star$ (CS 2-1) $\la$ 0.05 K, $T_{{\rm a}}^\star$ (¹²CO 2-1) $\sim$ 1.5 K, $T_{{\rm a}}^\star$ (¹³CO 2-1) $\la$ 0.1 K, and it is negligible for higher excitation lines and/or complex species.

We observed the full 3 mm window between 80 and 116 GHz toward both sources. The step between two adjacent tuning frequencies was 395 MHz, which yielded an overlap of 50 MHz. The autocorrelator VESPA produces artificial spikes with a width of 3-5 channels at the junction between subbands (typically 2 or 3 spikes per spectrum). To get rid of these artefacts, half of the integration time at each tuning frequency was spent with the backend shifted by 50 MHz, so that we could, without any loss of information, systematically remove in each spectrum 5 channels at each of the 6 junctions between subbands that were possibly affected by this phenomenon. At 2 mm, we observed at only 8 selected frequencies, and removed the artificial spikes in the same way as at 3 mm. At 1 mm, we covered the frequency ranges 201.8 to 204.6 GHz and 205.0 to 217.7 GHz, plus a number of selected spots at higher frequency. For each individual spectrum, we removed a 0 ${{\rm th}}$ -order (constant) baseline by selecting a group of channels which seemed to be free of emission or absorption. However, many spectra are full of lines, especially at 1.3 mm where we reached the confusion limit, and we may have overestimated the level of the baseline for some of them.

In $T_{{\rm a}}^\star$ scale, the rms noise level achieved towards Sgr B2(N) is about 15-20 mK below 100 GHz, 20-30 mK between 100 and 114.5 GHz, and about 50 mK between 114.5 and 116 GHz. At 1.3 mm, we reached the confusion limit for most of the spectra. The data were reduced with the CLASS software, which is part of the GILDAS software package (see http://www.iram.fr/IRAMFR/GILDAS).

2.2 Interferometric observations with the PdBI

We observed Sgr B2(N) with the PdBI for 4.7 h on February 7 ${{\rm th}}$ , 2006 with 6 antennas in the high-resolution A configuration (E24E68E04N46W27N29). The coordinates of the phase center were $\alpha_{{\rm J2000}} = 17^{{\rm h}}47^{{\rm m}}20\hbox{$.\!\!^{\rm s}$ }00$ , $\delta _{{\rm J2000}} = -28^\circ 22\hbox {$^\prime $ }19.0\hbox {$^{\prime \prime }$ }$ . The 3 and 1.2 mm receivers were tuned to 81.982 and 245.380 GHz, respectively, in single side band mode. At 3 mm, there were two spectral windows centered at 81.736 and 82.228 GHz with a bandwidth of 80 MHz and a channel separation of 0.313 MHz, and two continuum windows of 320 MHz bandwidth centered at 81.852 and 82.112 GHz. The atmospheric phase stability was good for the 3 mm band but bad for the 1.2 mm band. Therefore we do not analyze the 1.2 mm data. The system temperatures were typically 150-220 K at 3 mm in the lower sideband. The (naturally-weighted) synthesized half-power beam width was $3.4\hbox{$^{\prime\prime}$ }$ $\times$ $0.8\hbox{$^{\prime\prime}$ }$ with PA 10 $^\circ$ , and the primary beam was $\sim$ $61\hbox{$^{\prime\prime}$ }$ FWHM. The correlator bandpass was calibrated on the quasar 3C 273. Phase calibration was determined on the nearby sources NRAO 530 and 1622-297. The time-dependent amplitude calibration was done on 1622-297, NRAO 530, 1334-127, and 1749+096, while the absolute flux density scale was derived from MWC 349. The absolute calibration uncertainty is estimated to be $\sim$ 15 $\%$ . The data were calibrated and imaged using the GILDAS software. The continuum emission was estimated on line-free portions of the bands and removed in the uv plane. The deconvolution was performed with the CLEAN method (Clark 1980).

2.3 Interferometric observations with the ATCA

We observed Sgr B2(N) with the ATCA on May 17 ${{\rm th}}$ , 2006 in the hybrid H 214 configuration for 6 h, for 7 h on July 30 ${{\rm th}}$ , 2006 in the H 168 configuration, and for 6 hours in the compact H 75 configuration on September 25 ${{\rm th}}$ , 2006. The coordinates of the phase center were $\alpha_{{\rm J2000}} = 17^{{\rm h}}47^{{\rm m}}20\hbox{$.\!\!^{\rm s}$ }00$ , $\delta _{{\rm J2000}} = -28^\circ 22\hbox {$^\prime $ }19.0\hbox {$^{\prime \prime }$ }$ . The 3 mm receiver was alternately tuned to three frequency pairs of 90.550 and 93.200 GHz, 90.779 and 93.200 GHz, and 99.978 and 97.378 GHz in single side band mode, where only the first frequency of each pair was in spectral line mode with 32 MHz bandwidth and 128 channels. The second frequency of each pair was configured for continuum observations with 128 MHz bandwidth each. The system temperatures were typically 60 K. The (naturally-weighted) synthesized half-power beam width was $2.8\hbox{$^{\prime\prime}$ }$ $\times$ $1.9\hbox{$^{\prime\prime}$ }$ with PA $72^\circ$ for H 214, $3.9\hbox{$^{\prime\prime}$ }$ $\times$ $1.9\hbox{$^{\prime\prime}$ }$ with PA $-86^\circ$ for H 168, and $6.9\hbox{$^{\prime\prime}$ }$ $\times$ $5.4\hbox{$^{\prime\prime}$ }$ with PA $-73^\circ$ for H 75. The combination of H 214 and H 168 yields a synthesized half-power beam width of $3.0\hbox{$^{\prime\prime}$ }$ $\times$ $1.9\hbox{$^{\prime\prime}$ }$ with PA $82^\circ$ , and the combination of all three configurations a synthesized half-power beam width of $3.4\hbox{$^{\prime\prime}$ }$ $\times$ $2.3\hbox{$^{\prime\prime}$ }$ with PA $83^\circ$ . The primary beam was $\sim$ $2.4\hbox{$^\prime$ }$ FWHM. The correlator bandpass was calibrated on PKS 1253-055. The phase and gain calibration was determined on the nearby source PKS 1759-39. The absolute flux density scale was derived from Uranus. The absolute calibration uncertainty is estimated to be $\sim$ 20 $\%$ . The data were calibrated, continuum subtracted, imaged, and deconvolved using the software package MIRIAD (Sault et al. 1995).

Our ATCA data are affected by two problems. First, the tuning frequency used in May was not updated for the new observatory velocity in July and September. As a result, the observed bands were shifted by +12 and +16 MHz in rest frequency in the H 168 and H 75 configurations with respect to the H 214 configuration. Second, we suspect a technical problem with the tuning at 99 GHz in May and July because we do not detect any line in the H 214 and H 168 configurations while we easily detect two lines in the H 75 configuration: one unidentified line and one line from CH₃CH₃CO, $v_{{\rm t}}=1$ according to our line survey with the IRAM 30 m telescope. Comparing this band to the two other bands where we detect every line in each configuration (albeit with different intensities due to variable spatial filtering), we consider it to be very unlikely that the two lines detected at 99 GHz in the H 75 configuration are completely filtered out in the H 214 and H 168 configurations. Since the amino acetonitrile transition is shifted out of the H 75 band at 99 GHz (due to the variation of the observatory velocity), we do not analyze this dataset in the present article.

2.4 Interferometric observations with the VLA

We used the NRAO Very Large Array to search for the 1₀₁-0₀₀ multiplet of amino acetonitrile at 9071.208 MHz and examine the possibility of cold extended emission from this molecule. The VLA data were taken over a 1.5 h interval on February 13 ${{\rm th}}$ , 2003 when the array was in its lowest-resolution (D) configuration. Three $\sim$ 20 min long scans of the following position in Sgr B2 were alternated with scans of the phase calibrator NRAO 530. For absolute flux density calibration, 3C 286 was observed. Our phase center in Sgr B2 was at $\alpha_{{\rm J2000}}$ = 17 $^{\rm h}$ 47 $^{\rm m}$ 20 $\hbox{$.\!\!^{\rm s}$ }$ 00, $\delta_{{\rm J2000}}$ = $-28^\circ$ 22 $\hbox{$^\prime$ }$ 51.0 $\hbox{$^{\prime\prime}$ }$ . This is $32\hbox{$^{\prime\prime}$ }$ South and $16\hbox{$^{\prime\prime}$ }$ North of our 30 m telescope pointing positions for Sgr B2(N) and (M), respectively.

Our observations were done in spectral line mode with one intermediate frequency (IF) band split into 32 channels, each of which had a width of 0.1953 MHz, corresponding to 6.46 km s^-1. The usable central 72 $\%$ frequency range of the IF bandwidth, 4.49 MHz, covered all the multiplet's 7 hyperfine structure (hfs) components. This frequency range corresponds to a total velocity coverage of 148 km s^-1. The center velocity was set to $V_{{\rm lsr}}$ = 65 km s^-1. A 4.49 MHz bandwidth ``pseudo continuum'' database (the so-called ``channel 0'') was created by averaging the central 23 channels. The uv-data were calibrated using the NRAO's Astronomical Imaging Processing System (AIPS). Several iterations of self calibration delivered a high quality continuum image. Using UVLIN, the average of selected regions of the line uv-database were subtracted channel by channel from the latter to remove the continuum level. To calibrate the spectral line data, the phase and amplitude corrections determined by the initial calibration, as well as by the self calibration were transferred to the line database and applied to them successively, producing a 23 channel database which was imaged channel by channel using natural weighting. The synthesized beam width of the images is $20.5\hbox{$^{\prime\prime}$ }$ $\times$ $7.0\hbox{$^{\prime\prime}$ }$ FWHM with a position angle of $-6.3^\circ$ East of North.

3 Identification of amino acetonitrile

3.1 Amino acetonitrile frequencies

In the course of the present investigation an amino acetonitrile entry (tag: 56507) has been prepared for the catalog of the Cologne Database for Molecular Spectroscopy (CDMS, see Müller et al. 2005,2001). The laboratory transition frequencies were summarized by Bogey et al. (1990). Their work included microwave transitions reported without ¹⁴N quadrupole splitting by MacDonald & Tyler (1972), Pickett (1973), as well as microwave transitions reported with quadrupole splitting by Brown et al. (1977); the latter data were used with the reported splitting. Line fitting and prediction of transition frequencies was done with the SPFIT/SPCAT suite of programs (Pickett 1991) using a Watson type Hamiltonian in the S reduction in the representation I^r (see, e.g., Gordy & Cook 1984).

The set of spectroscopic parameters reported by Bogey et al. (1990) included terms of up to decic order (S_K), rather unusual for an apparently rigid and fairly heavy molecule, and we found the higher order parameters to be surprisingly large. Moreover, the off-diagonal sextic distortion parameter h₃ was larger in magnitude than h₂, and h₁ was not even used in the fit; the importance of these parameters is reversed to what is more commonly found. Therefore, we performed a trial fit with the octic and decic parameters as well as h₂ and h₃ omitted and the sextic term H_K fixed to a value that was estimated from $A/D_K \approx D_K/H_K$ . Subsequently, we found that inclusion of h₁ improved the quality of the fit. All transitions but five having $\Delta K_a \ge 1$ and $K''_a \ge 3$ could be reproduced well. Two of these had K_a = 3-2 and deviated $\sim$ 3.5 MHz from the predicted frequencies which was only twice the predicted uncertainty. The inclusion of these transitions in the fit caused relatively small changes in A and D_K; changes in the remaining parameters were within the uncertainties. Therefore, it is likely that the assignments of these two transitions are correct. Effects on the predicted $\Delta K_a = 0$ transition frequencies are negligible. The remaining three transitions are considered to be mis-assignments as their inclusion would require many more higher order distortion parameters with apparently unphysical values and a poorer quality of the fit. Hence, those transitions were omitted from our fit. The resulting spectroscopic parameters are given in Table 1. The rotational partition function at 75 and 150 K is 4403 and 12 460, respectively.

3.2 Modeling of the 30 m line survey

The average line density above 3 $\sigma$ in our 30 m survey is about 100 and 25 features per GHz for Sgr B2(N) and (M), respectively, translating into about 3700 and 950 lines over the whole 80-116 GHz band. To identify a new molecule in such a line forest and reduce the risk of mis-assignments, it is essential to model first the emission of all known molecules, including vibrationally and torsionally excited states, and their isotopologues. We used the XCLASS software (see Comito et al. 2005, and references therein) to model the emission and absorption lines in the LTE approximation. These calculations take into account the beam dilution, the line opacity, and the line blending. The molecular spectroscopic parameters are taken from our line catalog which contains all entries from the CDMS catalog (Müller et al. 2005,2001) and from the molecular spectroscopic database of the Jet Propulsion Laboratory (JPL, see Pickett et al. 1998), plus additional ``private'' entries.

Each molecule is modeled separately with the following set of input parameters: source size, rotational temperature, column density, velocity linewidth, velocity offset with respect to the systemic velocity of the source, and a flag indicating if it is an emission or absorption component. For some of the molecules, it was necessary to include several velocity components to reproduce the observed spectra. The velocity components in emission are supposed to be non-interacting, i.e. the intensities add up linearly. The radiative transfer is computed in the following way: first the emission line spectrum is calculated, and then the absorption lines, using the full (lines + continuum) emission spectrum as background to absorb against. The vibrationally and/or torsionally excited states of some molecules were modeled separately from the ground state. The input parameters were varied until a good fit to the data was obtained for each molecule. The whole spectrum including all the identified molecules was then computed at once, and the parameters for each molecule were adjusted again when necessary. The quality of the fit was checked by eye over the whole frequency coverage of the line survey. We favored our eye-checking method against an automated fitting because the high occurence of line blending and the uncertainty in the baseline removal would in many cases make an automated fitting procedure fail.

The detailed results of this modeling will be published in a forthcoming article describing the complete survey (Belloche et al., in prep.). So far, we have identified 51 different molecules, 60 isotopologues, and 41 vibrationally/torsionally excited states in Sgr B2(N), which represent about 60 $\%$ of the lines detected above the 3 $\sigma$ level. In Sgr B2(M), the corresponding numbers are 41, 50, 20, and 50 $\%$ , respectively.

3.3 Detection of amino acetonitrile with the 30 m telescope

Table 3: Transitions of amino acetonitrile detected toward Sgr B2(N) with the IRAM 30 m telescope.

We consider it essential for claiming a detection of a new molecule that all lines of this molecule in our observed bands are consistent with this claim, i.e. are either detected or blended with lines of other species. Therefore, in the following, we inspect all transitions of amino acetonitrile in our frequency range. Our line survey at 3, 2, and 1.3 mm covers 596 transitions of our amino acetonitrile catalog (v= 0 only). Our LTE modeling shows, however, that the transitions with the line strength times the appropriate (a- or b-type) dipole moment squared, $S\mu ^2$ , smaller than 20 D² are much too weak to be detectable with the sensitivity we achieved. Therefore, we list in Table 2 (Online Material) only the 398 transitions above this threshold. To save some space, when two transitions have a frequency difference smaller than 0.1 MHz which cannot be resolved, we list only the first one. We number the transitions in Col. 1 and give their quantum numbers in Col. 2. The frequencies, the frequency uncertainties, the energies of the lower levels in temperature units, and the $S\mu ^2$ values are listed in Cols. 3-6, respectively. Since the spectra are in most cases close to the line confusion limit and it is difficult to measure the noise level, we give in Col. 7 the rms sensitivity computed from the system temperature and the integration time: $\sigma = \frac{F_{{\rm eff}}}{B_{{\rm eff}}} \times \frac{2~T_{{\rm sys}}}{\sqrt{\delta f ~ t}}$ , with $F_{{\rm eff}}$ and $B_{{\rm eff}}$ the forward and beam efficiencies, $T_{{\rm sys}}$ the system temperature, $\delta f$ the spectral resolution, and t the total integration time (on-source plus off-source).

We list in Col. 8 of Table 2 comments about the blends affecting the amino acetonitrile transitions. As can be seen in this table, most of the amino acetonitrile lines covered by our survey of Sgr B2(N) are heavily blended with lines of other molecules and therefore cannot be identified in this source. Only 88 of the 398 transitions are relatively free of contamination from other molecules, known or still unidentified according to our modeling. They are marked ``Detected'' or ``Group detected'' in Col. 8 of Table 2, and are listed with more information in Table 3. They correspond to 51 observed features which are shown in Fig. 1 (Online Material) and labeled in Col. 8 of Table 3. For reference, we show the spectrum observed toward Sgr B2(M) in this figure also. We identified the amino acetonitrile lines and the blends affecting them with the LTE model of this molecule and the LTE model including all molecules (see Sect. 3.2). The parameters of our best-fit LTE model of amino acetonitrile are listed in Table 4, and the model is overlaid in red on the spectrum observed toward Sgr B2(N) in Fig. 1. The best-fit LTE model including all molecules is shown in green in the same figure. The source size we used to model the amino acetonitrile emission was derived from our interferometric measurements (see Sect. 3.4 below).

For the frequency range corresponding to each observed amino acetonitrile feature, we list in Table 3 the integrated intensities of the observed spectrum (Col. 10), of the best-fit model of amino acetonitrile (Col. 11), and of the best-fit model including all molecules (Col. 12). In these columns, the dash symbol indicates transitions belonging to the same feature. Columns 1 to 7 are the same as in Table 2. The $1 \sigma$ uncertainty given in Col. 10 was computed using the estimated noise level of Col. 7. These measurements are plotted in the form of a population diagram in Fig. 2, which plots upper level column density divided by statistical weight, $N_{\rm u}/g_{\rm u}$ , versus the upper level energy in Kelvins (see Goldsmith & Langer 1999). The data are shown in black and our best-fit model of amino acetonitrile in red. Out of 21 features encompassing several transitions, 10 contain transitions with different energy levels and were ignored in the population diagram (features 2, 3, 8, 33, 34, 35, 36, 42, and 49). We used Eq. (A5) of Snyder et al. (2005) to compute the ordinate values. This equation assumes optically thin emission. To estimate by how much line opacities affect this diagram, we applied the opacity correction factor $C_\tau = \frac{\tau}{1-{\rm e}^{-\tau}}$ (see Snyder et al. 2005; Goldsmith & Langer 1999) to the modeled intensities, using the opacities from our radiative transfer calculations (Col. 9 of Table 3); the result is shown in green in Fig. 2. The population diagram derived from the modeled spectrum is slightly shifted upwards but its shape, in particular its slope (the inverse of which approximately determines the rotation temperature), is not significantly changed, since ${\rm ln}~ C_{\tau}$ does not vary much (from 0.04 to 0.24). The populations derived from the observed spectrum in the optically thin approximation are therefore not significantly affected by the optical depth of the amino acetonitrile transitions. The scatter of the black crosses in Fig. 2 is therefore dominated by the blends with other molecules and uncertainties in the baseline removal (indicated by the downwards and upwards blue arrows, respectively). From this analysis, we conclude that our best-fit model for amino acetonitrile is fully consistent with our 30 m data of Sgr B2(N).

$\begin{figure} \par\includegraphics[angle=270,width=9cm,clip]{9203f2.eps}\end{figure}$

Figure 2: Population diagram of amino acetonitrile in Sgr B2(N). The red points were computed in the optically thin approximation using the integrated intensities of our best-fit model of amino acetonitrile, while the green points were corrected for the opacity. The black points were computed in the optically thin approximation using the integrated intensities of the spectrum observed with the IRAM 30 m telescope. The error bars are $1 \sigma$ uncertainties on $N_{\rm u}/g_{\rm u}$ . Blue arrows pointing downwards mark the transitions blended with transitions from other molecules, while blue arrows pointing upwards indicate that the baseline removed in the observed spectrum is uncertain. The arrow length is arbitrary. The measurement corresponding to feature 43 (at $E_{\rm u}/k_{\rm B}$ = 265 K) is not shown since the integrated intensity measured toward Sgr B2(N) is negative, due to the blend with CN absorption lines.

Open with DEXTER

Finally, as mentioned above, the 310 transitions of Table 2 not shown in Fig. 1 are all but one heavily blended with transitions of other molecules and cannot be clearly identified in Sgr B2(N). The single exception is amino acetonitrile transition 192 shown in Fig. 3. There are too many blended lines in this frequency range to properly remove the baseline. It is very uncertain and the true baseline is most likely at a lower level than computed here. The presence of several H¹³CN 2-1 velocity components in absorption also complicates the analysis. Therefore it is very likely that the (single) apparent disagreement concerning transition 192 between our best-fit model and the 30 m spectrum observed toward Sgr B2(N) is not real and does not invalidate our claim of detection of amino acetonitrile.

$\begin{figure} \par\includegraphics[angle=270,width=9cm,clip]{9203f3.eps}\end{figure}$

Figure 3: Spectrum obtained toward Sgr B2(N) ( bottom) and Sgr B2(M) ( top) with the IRAM 30 m telescope at the frequency of amino acetonitrile (AAN) transition 192 (see caption of Fig. 1 for more details about the color coding). There are too many blended lines in the spectrum of Sgr B2(N) to properly remove the baseline, which is very uncertain and most likely at a lower level than could be computed here. This is the only discrepancy concerning the amino acetonitrile lines in the whole survey. The absorption lines, particularly strong in the spectrum of Sgr B2(M), are velocity components of H¹³CN 2-1.

Open with DEXTER

3.4 Mapping amino acetonitrile with the PdBI

The two 3 mm spectral windows of the PdBI were chosen to cover the five amino acetonitrile features F2 to F6. The spectra toward Sgr B2(N) are shown for both windows toward 3 positions P1, P2, and P3 in Figs. 4a to f. Many lines are detected, the strongest one being a line from within the vibrationally excited state v₇ = 1 of cyanoacetylene (HC₃N) at 82.2 GHz. We also easily detect lines from within its vibrationally excited state v₄ = 1, from its isotopologues HC¹³CCN and HCC¹³CN in the v₇ = 1 state, from ethyl cyanide (C₂H₅CN), as well as two unidentified lines at 82.213 and 82.262 GHz. At a lower level, we find emission for all the amino acetonitrile features F2 to F6, and we also detect methylformate (CH₃OCHO).

Table 5: Measurements obtained toward Sgr B2(N) with the IRAM Plateau de Bure interferometer at 82 GHz.

The integrated intensity maps of the amino acetonitrile features F2 to F6 are presented in Figs. 5a to e, along with two maps of ethyl cyanide (Figs. 5g and h), four maps of cyanoacetylene in the vibrationally excited states v₄ = 1 and v₇ = 1 (Figs. 5j to m), one map of its isotopologue HC¹³CCN in the state v₇ = 1 (Fig. 5i), one map of methylformate (Fig. 5n), and a reference map computed on the PdBI line-free frequency range between F2 and F3 (Fig. 5f). The frequency intervals used to compute the integrated intensities are given in Cols. 3 and 4 of Table 5 and shown with dotted lines in Fig. 4. We used the fitting routine GAUSS_2D of the GILDAS software to measure the position, size, and peak flux of each integrated emission. The results are listed in Cols. 6 to 11 in Table 5. We label P1 the mean peak position of features F2 to F6, P2 the northern peak position of ethyl cyanide, and P3 the peak position of methylformate (see Table 6 and the plus symbols in Fig. 5). Finally, the PdBI velocity-integrated flux spatially integrated over the emitting region is listed in Col. 12 and the 30 m velocity-integrated intensity is given in Col. 13.

$\begin{figure} \par\includegraphics[angle=270,width=8.9cm,clip]{9203f4.eps}\end{figure}$

Figure 4: Spectra obtained with the Plateau de Bure interferometer (a) to f)) and the 30 m telescope (g) and h)) toward Sgr B2(N) (in black). The dotted lines show the frequency ranges listed in Table 5. The offset position with respect to the reference position of Fig. 5 is given in each panel, along with a label (P1 to P3, see their definition in Table 6). The lines identified in our 30 m survey are labeled in blue. The red spectra show our best-fit model for amino acetonitrile (AAN) while the green spectrum corresponds to the 30 m model including all molecules. The observed lines which have no counterpart in the green spectrum are still unidentified.

Open with DEXTER

We present in Fig. 5o the map of continuum emission at 82.0 GHz, integrated over line-free frequency ranges. The continuum emission has a complex structure. The main region peaks at $\alpha_{{\rm J2000}} = 17^{{\rm h}}47^{{\rm m}}19\hbox{$.\!\!^{\rm s}$ }886 \pm 0\hbox{$.\!\!^{\rm s}$ }005$ , $\delta_{{\rm J2000}} = -28^\circ22\hbox{$^\prime$ }18.4\hbox{$^{\prime\prime}$ }\pm 0.1\hbox{$^{\prime\prime}$ }$ , i.e. within 0.1 $\hbox{$^{\prime\prime}$ }$ of the position of the ultracompact H II region K2. It also shows hints of emission at the position of the ultracompact H II regions K1 and K3, although the spatial resolution is too poor to resolve them (see, e.g., Gaume et al. 1995). There are other secondary peaks. One of them coincides with the peak of the shell-like H II region K6 while another one is located close (< $2\hbox{$^{\prime\prime}$ }$ ) to the peak of the shell-like H II region K5 and traces most likely the same shell. On the other hand, we detect no 3.7 mm emission at the position of the weak ultracompact H II region K9.69 (Gaume et al. 1995).

$\begin{figure} \par\includegraphics[height=18.4cm,angle=-90,clip]{NEWFIG5/9203f5.eps}\end{figure}$

Figure 5: Integrated intensity maps (panels a) to n)) and continuum map (panel o)) obtained toward Sgr B2(N) with the Plateau de Bure interferometer at 82 GHz. Panels a) to e) show the amino acetonitrile (AAN) features F2 to F6 (see Fig. 4 and Table 3). Panel f) is a reference map integrated on the emission-free frequency range between F2 and F3. Panels g) to n) show the other molecules listed in Table 5. The lowest contour (positive in black solid line and negative in blue dotted line) and the contour step are 2 $\sigma$ for panel f), 3 $\sigma$ for panels a) to e) and panel n), 4 $\sigma$ for panels g) and h), 5 $\sigma$ for panels i) to l), and 6 $\sigma$ for panel m) (with $\sigma$ given in Col. 5 of Table 5). For panel o), the first contours are 5 $\sigma$ and 10 $\sigma$ , and the contour step is 10 $\sigma$ for the other contours (with $\sigma$ = 8.5 mJy/beam). In each panel, the 0, 0 position is $\alpha_{{\rm J2000}} = 17^{{\rm h}}47^{{\rm m}}20\hbox{$.\!\!^{\rm s}$ }00$ , $\delta _{{\rm J2000}} = -28^\circ 22\hbox {$^\prime $ }19.0\hbox {$^{\prime \prime }$ }$ , the three plus symbols mark the positions P1, P2, and P3 (labeled in panel f)), and the filled ellipse in the top right corner shows the clean beam ( HPBW = $3.35\hbox {$^{\prime \prime }$ }$ $\times$ $0.81\hbox {$^{\prime \prime }$ }$ at PA = 9.7 $^\circ$ ). The cross symbols in panel o) are the peak positions of the (ultracompact) H II regions detected by Gaume et al. (1995) at 1.3 cm (K9.69, K1, K2, K3, K5, and K6, from right to left). The spectral integration was done on the frequency ranges given in Table 5. The continuum map was computed on line-free frequency ranges. The maps are not corrected for primary beam attenuation. The amino acetonitrile features emit at the same position as the vibrationally excited state v₄ = 1 of cyanoacetylene. Feature F4 is partially blended with a transition from HCC¹³CN, v₆ = 1.

Open with DEXTER

The strong lines detected with the PdBI (Figs. 5g to n) allow us to gain insight into the distribution of molecular line emission in Sgr B2(N). The double-peaked profile of ethyl cyanide seen with the 30 m telescope (see Fig. 4g) is resolved with the PdBI into two sources P1 and P2 separated by about 5.3 $\hbox{$^{\prime\prime}$ }$ (see Figs. 4a, c, g, h, and Table 6). P1 and P2 are spatially and kinematically coincident with the quasi-thermal methanol emission cores ``i'' and ``h'' within $0.6\hbox{$^{\prime\prime}$ }$ and $0.2\hbox{$^{\prime\prime}$ }$ , respectively (Mehringer & Menten 1997). Cores ``i'' and ``h'' were both previously detected in ethyl cyanide (Jones et al. 2007; Liu & Snyder 1999; Hollis et al. 2003). Many molecules (but not amino acetonitrile within the limits of our sensitivity) actually show this double-peaked profile in our 30 m survey of Sgr B2(N) and are most likely emitted by these two sources. P1 and P2 are also detected in our PdBI data in the vibrationally excited state v₇ = 1 of cyanoacetylene (see Figs. 4b, d, and 5k, l), and there is a hint of emission toward P2 in the isotopologue HC¹³CCN while P1 is easily detected (see Figs. 4c and a). In addition, the wings of the main component of cyanoacetylene v₇ = 1 are spatially shifted: the redshifted wing peaks about $1\hbox{$^{\prime\prime}$ }$ North-West of P1 while the blueshifted wing peaks about $1\hbox{$^{\prime\prime}$ }$ East of P1 (Figs. 5k and m). This East-West velocity gradient was previously reported by several authors (e.g. de Vicente et al. 2000; Hollis et al. 2003; Lis et al. 1993). Although it could result from cloud rotation, it is most likely a sign of outflow activity (see, e.g., Liu & Snyder 1999). The transition with highest energy in our PdBI sample is a transition of cyanoacetylene in the vibrationally excited state v₄ = 1 ( $E_{\rm u}/k_{\rm B} = 1283$ K). Within the limits of our sensitivity, we detect emission only toward P1 in this highly excited transition (Fig. 5j). Finally, methylformate peaks at a position significantly offset from P1, at $1.7\hbox{$^{\prime\prime}$ }$ to the North-West (Fig. 5n). It has no counterpart in the continuum map of Fig. 5o. To sum up, our PdBI data reveal three main positions of molecular line emission (P1 and P2 corresponding to the methanol cores ``i'' and ``h'', and P3 the peak position of methylformate), and an East-West velocity gradient around P1.

Within the limits of our sensitivity, the amino acetonitrile features F2 to F6 detected with the PdBI show only one peak, and they all peak at the same position (Figs. 5a to e). We are confident that the emission detected in features F2 to F6 is not contaminated by the continuum since no significant signal is detected in the reference map (Fig. 5f). Their weighted-mean peak position was labeled P1 above (offset -1.5 $\pm$ $0.2\hbox{$^{\prime\prime}$ }$ , 0.2 $\pm$ $0.2\hbox{$^{\prime\prime}$ }$ , see Table 6). The fact that all features are detected at the same position is consistent with their assignment to the same molecule (see above the shifted position of methylformate for instance). The deconvolved major and minor axes of the emission detected in features F2 to F6 are in the range 0-2.2 $\hbox{$^{\prime\prime}$ }$ and 1.0-1.9 $\hbox{$^{\prime\prime}$ }$ , respectively. The amino acetonitrile emission is therefore slightly resolved and has a size of roughly 2 $\hbox{$^{\prime\prime}$ }$ FWHM, which we used for the LTE modeling. The spatially integrated fluxes of F4 and F6 agree within 20 $\%$ with the fluxes measured with the 30 m telescope (see Cols. 12 and 13 of Table 5). The emission detected with the 30 m telescope in these two features of amino acetonitrile is therefore compact (2 $\hbox{$^{\prime\prime}$ }$ ) and was not filtered out by the interferometer. The other features F2, F3, and F5 have 30 m fluxes 1.6, 1.4, and 3.8 times larger than the PdBI fluxes, respectively: the emission filtered out by the interferometer most likely corresponds to the unidentified transitions blended with these amino acetonitrile features (see Fig. 4a). In addition, the low signal-to-noise ratio of feature F5 detected with the PdBI may significantly affect the flux measurement.

We used the parameters of the 30 m model (see Table 4) to compute a model spectrum of amino acetonitrile with the spatial resolution of the PdBI (using the geometrical mean of the elliptical beam). The agreement with the peak spectrum is good, within a factor of 2 (see Figs. 4a and b). The small discrepancy may come from the somewhat uncertain source size and from our approximate modeling of the interferometric beam pattern: spherical beam and full uv coverage for the model versus elliptical beam and partially sampled uv coverage for the observations. Overall, our LTE model of amino acetonitrile is therefore well consistent with the compact emission detected with the PdBI.

3.5 Mapping amino acetonitrile with the ATCA

Table 6: Peak positions of continuum and molecular line emission detected with the PdBI and the ATCA toward Sgr B2(N).

$\begin{figure} \par\includegraphics[width=9cm,clip]{9203f6.eps}\end{figure}$

Figure 6: Spectra obtained with the Australia Telescope Compact Array (a) to j)) and the 30 m telescope (k) and l)) toward Sgr B2(N) (black histogram). The dotted lines show the frequency ranges listed in Table 7. The offset position with respect to the reference position of Fig. 7 is given in each panel. The lines identified in our 30 m survey are labeled in blue. The red spectrum shows our best-fit model for amino acetonitrile (AAN) while the green spectrum corresponds to the 30 m model including all molecules. The observed lines which have no counterpart in the green spectrum are still unidentified. a) and b) show the extended configuration H 214, c) and d) the intermediate configuration H 168, e) and f) the compact configuration H 75, g) and h) the combination of H 214 and H 168, and i) and j) the combination of all three configurations. The spectral coverage is not the same for all configurations because the sky tuning frequency for H 168 and H 75 was not corrected for the observatory velocity variations. The clean beam size ( HPBW) is given in each panel.

Open with DEXTER

The two 3 mm spectral windows of the ATCA were chosen to cover the four amino acetonitrile features F7 to F10. The spectra toward position P4 of Sgr B2(N) (offset $-2\hbox{$^{\prime\prime}$ }$ , 0.1 $\hbox{$^{\prime\prime}$ }$ ) are presented for both windows in Fig. 6. Since the spectral windows were not exactly the same in each configuration (see Sect. 2.3), we show the spectra for each configuration (Figs. 6a to f), plus the combination of the two broadest ones (Figs. 6g and h), and the combination of all three configurations (Figs. 6i and j). Nearly all the lines seen with the 30 m telescope are detected with the ATCA toward P4. In the 90.6 GHz band, we detect the blue wing of an SO₂ transition (Fig. 6a), the red wing of an HC¹³CCN ground-state transition (Fig. 6e), an unidentified line, and feature F7. In the 90.8 GHz band, we detect the low-velocity component of a $v_{{\rm t}}$ = 1 transition of methanol, the low-velocity component of a v₇ = 1 transition of HC¹³CCN (Figs. 6d and f), and the three amino acetonitrile features F8, F9, and F10. Toward the northern position P5, we detect a second velocity component of methanol $v_{{\rm t}}=1$ and HC¹³CCN v₇=1 (not shown in Fig. 6, see below).

The integrated intensity maps of the amino acetonitrile features F7 to F10 in the different configurations are presented in Figs. 7a to p, along with maps of the excited states of methanol and HC¹³CCN (Figs. 7q to v). The frequency intervals used to compute the integrated intensities are given in Cols. 3 and 4 of Table 7 and drawn in dotted lines in Fig. 6. We used the fitting routine GAUSS_2D of the GILDAS software to measure the position, size, and peak flux of each integrated emission. The results are listed in Cols. 6 to 11 of Table 7. We label P4 the weighted-mean peak position of features F7 to F10, computed using the combined configuration H 214 + H 168 (only H 214 for F8), and P5 the average northern peak position of methanol and HC¹³CCN. The mean peak position P4 is at the offset (-2.0 $\pm$ $0.3\hbox{$^{\prime\prime}$ }$ , 0.1 $\pm$ $0.2\hbox{$^{\prime\prime}$ }$ ), and the average position P5 is at (-1.9 $\pm$ $0.1\hbox{$^{\prime\prime}$ }$ , 5.2 $\pm$ $0.1\hbox{$^{\prime\prime}$ }$ ) (see coordinates in Table 6 and positions in Fig. 7x). Finally, the ATCA velocity-integrated flux spatially integrated over the emitting region is listed in Col. 12 of Table 7 and the 30 m velocity-integrated intensity is given in Col. 13.

Table 7: Measurements obtained toward Sgr B2(N) with the Australia Telescope Compact Array at 91 GHz.

$\begin{figure} \par\includegraphics[angle=270,width=18cm,clip]{9203f7.eps}\end{figure}$

Figure 7: Integrated intensity maps (panels a) to v)) and continuum map (panel w)) obtained toward Sgr B2(N) with the Australia Telescope Compact Array at 3 mm. Panels a) to p) show the amino acetonitrile features F7 to F10 in the different configurations (see Fig. 6). Panels q) to v) show the other molecules listed in Table 7. The lower contour (positive in black solid line and negative in blue dotted line) and the contour step are 3 $\sigma$ for panels a), b), d) to h), j) to m), and o) to q), 4 $\sigma$ for panels r), t), and u), 5 $\sigma$ for panels c), i) and n), and 6 $\sigma$ for panels s) and v) (with $\sigma$ given in Col. 5 of Table 7). For panel w), the first contours are 3 and 6 $\sigma$ , and the contour step is 6 $\sigma$ for the other contours (with $\sigma = 37$ mJy/beam). In each panel, the 0, 0 position is $\alpha_{{\rm J2000}} = 17^{{\rm h}}47^{{\rm m}}20\hbox{$.\!\!^{\rm s}$ }00$ , $\delta _{{\rm J2000}} = -28^\circ 22\hbox {$^\prime $ }19.0\hbox {$^{\prime \prime }$ }$ , the two thick plus symbols mark positions P4 and P5 (labeled in panel a)), and the filled ellipse in the top right corner shows the clean beam. The cross symbols in panel w) are the peak positions of the (ultracompact) H II regions detected by Gaume et al. (1995) at 1.3 cm (K9.69, and K1 to K6, from right to left). The spectral integration was done on the frequency ranges given in Table 7. The maps are not corrected for primary beam attenuation. Panel x) displays the ATCA positions P4 and P5 with thick plus symbols and the PdBI positions P1 to P3 with thin plus symbols, for visual comparison.

Open with DEXTER

Since the line-free frequency ranges in the two spectral windows are too small to compute a reliable continuum map, we present in Fig. 7w a map combining the emission obtained at 93.2 and 97.4 GHz in the continuum mode. The contaminating lines were avoided in the integration. The continuum emission detected with the ATCA has properties very similar to the emission detected with the PdBI (see Sect. 3.4). The emission peaks at $\alpha_{{\rm J2000}} = 17^{{\rm h}}47^{{\rm m}}19\hbox{$.\!\!^{\rm s}$ }87$ $\pm$ $0\hbox{$.\!\!^{\rm s}$ }01$ , $\delta_{{\rm J2000}} = -28^\circ22\hbox{$^\prime$ }18.7\hbox{$^{\prime\prime}$ }$ $\pm$ $0.1\hbox{$^{\prime\prime}$ }$ , i.e. within 0.3 $\hbox{$^{\prime\prime}$ }$ of the ultracompact H II region K2 (Gaume et al. 1995). The continuum emission is somewhat extended around the main peak: it extends toward the other ultracompact H II regions K1, K3, and the shell-like H II region K5, but these features are not resolved with the ATCA. We also detect a shell-like emission close to K6, but its signal-to-noise ratio is lower than in the PdBI data. We do not detect any emission toward the weak ultracompact H II region K9.69 (Gaume et al. 1995).

The strong methanol $v_{{\rm t}}=1$ and HC¹³CCN v₇=1 lines show the same structure as the strong lines detected with the PdBI, namely emission from two different positions separated by about 5.1 $\hbox{$^{\prime\prime}$ }$ in declination, each with a distinct velocity (Figs. 7q, r, t, and u). These two positions P4 and P5 coincide with the PdBI positions P1 and P2, respectively, within 0.5 $\hbox{$^{\prime\prime}$ }$ which is about one fifth of the ATCA synthesized beam (see Fig. 7x for a visual comparison). Therefore we are very confident that they correspond to the same regions. The third position P3 detected with the PdBI in methylformate is not detected in the small number of transitions observed with the ATCA.

Within the limits of our sensitivity, the amino acetonitrile features F7 to F10 show only one peak, and they all peak at the same position P4 coincident with the PdBI position P1. Therefore, all amino acetonitrile features detected with both the ATCA and the PdBI peak at the same position, which is consistent with their assignment to the same molecule. The emission detected in features F7 to F10 is barely resolved with the ATCA. Given the uncertainties, it is consistent with the source size of 2 $\hbox{$^{\prime\prime}$ }$ suggested by our measurements with the PdBI. The ATCA spatially integrated flux of feature F7 in the mixed configuration H 214 + H168 agrees with the 30 m telescope flux within 40 $\%$ . Given the calibration uncertainties of both instruments, and the somewhat larger noise in the 30 m spectrum, most of the flux of feature F7 is recovered with the ATCA. On the other hand, the ATCA spatially integrated fluxes of features F8 to F10 are significantly weaker than the 30 m telescope fluxes. The main reason for this disagreement may be the contamination of the 30 m spectrum by emission from transitions of other molecules which are still unidentified in our survey (see Fig. 6l). This contaminating emission is missed by the interferometer either because it peaks at an offset position or because it is extended and filtered out.

The model spectrum of amino acetonitrile computed with the same parameters as the 30 m model (see Table 4) but with the spatial resolution of the ATCA is shown in red in Figs. 6a to j. The agreement with the spectra obtained with the ATCA is good, within a factor of 2 (see comment about the interferometric modeling in Sect. 3.4). Overall, our LTE model of amino acetonitrile is therefore well consistent with the compact emission detected with the ATCA.

3.6 Abundance of amino acetonitrile in Sgr B2(N)

The continuum emission detected with the PdBI (see Fig. 5o) has a peak intensity of 0.459 $\pm$ 0.009 Jy/ $3.3\hbox{$^{\prime\prime}$ }$ $\times$ $0.8\hbox{$^{\prime\prime}$ }$ -beam. In the 2''-diameter region convolved with the PdBI beam centered on the AAN peak position P1 (convolved FWHM: $3.9\hbox{$^{\prime\prime}$ }$ $\times$ $2.2\hbox{$^{\prime\prime}$ }$ ), we measure a mean intensity of 0.257 Jy/beam and a flux of 0.556 Jy. The continuum emission from Sgr B2(N) at 3.7 mm is dominated by thermal dust emission, at most 30 $\%$ come from free-free emission (Kuan et al. 1996), probably even less toward the ultracompact H II region K2 since its flux is only 20 and 60 mJy at 2 and 1.3 cm, respectively (Gaume & Claussen 1990; Gaume et al. 1995). In addition, the thermal dust emission is optically thin (Carlstrom & Vogel 1989), so we can estimate the H₂ column density using the equation:

Using the column density derived from our modeling (see Table 4), we find an amino acetonitrile abundance relative to H₂ of about 2.2 $\times$ 10^-9 in the inner region of deconvolved FWHM diameter 2 $\hbox{$^{\prime\prime}$ }$ , with an uncertainty on the abundance of at least a factor of 2 given the uncertainties on the dust mass opacity and the dust emissivity exponent.

The continuum emission detected with the ATCA at a mean frequency of 95.3 GHz has a peak intensity of 1.00 Jy/ $3.5\hbox{$^{\prime\prime}$ }$ $\times$ $2.4\hbox{$^{\prime\prime}$ }$ -beam. This intensity is 1.8 times larger than the PdBI continuum flux measured over an area $3.9\hbox{$^{\prime\prime}$ }$ $\times$ $2.2\hbox{$^{\prime\prime}$ }$ (see above). If this flux difference comes from the frequency dependence of the continuum emission, then we derive a frequency exponent of 3.9 which is, within the calibration uncertainties, consistent with the exponent of 3.5 expected for thermal dust emission with a dust emissivity exponent $\beta = 1.5$ assumed above. We are therefore confident that the continuum emission detected toward K2 with the PdBI and the ATCA is largely dominated by thermal dust emission.

3.7 Limits on possible extended emission of cold amino acetronitrile with the VLA

$\begin{figure} \par\includegraphics[width=8.6cm,clip]{9203f8.eps}\end{figure}$

Figure 8: Continuum image of the Sgr B2 region obtained with the Very Large Array at 9.1 GHz . The lowest contour level is 4 times the rms noise level of 8 mJy beam^-1 and contours double in value until they reach 512 times that level. The dotted circle represents the FWHM of the VLA antennas' primary beam at 9.1 GHz. The image is not corrected for attenuation due to the primary beam's response. Note that it is produced from data taken over only a 1 h period and has limited dynamic range and sensitivity. It is, however, consistent with the higher sensitivity three-pointing-mosaic 4.8 GHz image presented by Mehringer et al. (1995), which has a similar resolution. The synthesized beam is represented in the lower left corner. The upper and lower shaded circles are centered at the Sgr B2(N) and (M) pointing positions of our 30 m telescope spectral line survey, respectively. Their size corresponds to the $25\hbox {$^{\prime \prime }$ }$ FWHM of the 30 m telescope at 100 GHz.

Open with DEXTER

Figure 8 shows the continuum image of the Sgr B2 region we obtained at 9.1 GHz with the VLA. At this frequency, the primary beam has a FWHM of $\sim$ $5\hbox{$^\prime$ }$ . Our uv-data should adequately sample structures with sizes of up to $1.5\hbox{$^\prime$ }$ . There are two major continuum sources, Sgr B2(N) and (M), whose emission is a blend of the contributions of many separate sources that are resolved in higher resolution images (e.g., Gaume et al. 1995). A number of secondary sources are evident. The peak brightness, $S_{\rm p}$ , toward Sgr B2(N) is 2.85 Jy beam^-1. Toward Sgr B2(M) we measure $S_{\rm p} = 4.12$ Jy beam^-1. The image has a relatively high 1 $\sigma$ rms noise level of 8.4 mJy beam^-1 that is determined by dynamic range limitations. Nowhere in the whole line datacube do we find any significant line emission or absorption above 3.4 mJy beam^-1 (5 times the 1 $\sigma$ rms noise level in the images of the spectral line emission). This corresponds to a brightness temperature upper limit of 0.35 K. With this, and assuming a width of 30 km s^-1, similar to other lines with extended emission (e.g., Hollis et al. 2004; Hunt et al. 1999) we can use the standard relation to calculate an upper limit to the column densities of the upper levels of the strongest of the J_{K_a,K_c} = 1₀₁-0₀₀ hfs components of $\sim$ 8 $\times$ 10¹² cm^-2. Hollis et al. (2004) use a two temperature component picture to explain their multitransition glycolaldehyde study, invoking components with temperatures of T = 8 and 50 K. For 8 K, our limit on the total column density of extended amino acetonitrile would be $\sim$ 3 $\times$ 10¹⁴ cm^-2 and for 50 K it would be $\sim$ 3 $\times$ 10¹⁶ cm^-2. The low-T value is on the same order as the total column density of glycolaldehyde that Hollis et al. (2004) calculate for this temperature.

Using the ATCA, Hunt et al. (1999) imaged Sgr B2 in the J=1-0 line of HC₃N at a frequency near 9.1 GHz, very close to our amino acetonitrile frequency. In their $4.4\hbox{$^{\prime\prime}$ }$ $\times$ $9.9\hbox{$^{\prime\prime}$ }$ resolution images they found the spatial distribution of the line emission to resemble that of the continuum emission very closely. Assuming that possible extended amino acetonitrile had a similar distribution, the high continuum flux densities allow us to determine very sensitive limits on possible absorption (or weakly inverted emission) toward the continuum emission. We calculate very low $5\sigma$ limits on the absolute value of the optical depth of 8.3 $\times$ 10^-4 and 1.2 $\times$ 10^-3 toward Sgr B2(N) and (M). This implies $\sim$ 125 and 20 times lower column density limits for T = 8 and 50 K, respectively, compared to limits on possible thermal emission quoted above.

4 Discussion

4.1 Amino acetonitrile in Sgr B2(N)

We detected compact emission from amino acetonitrile in Sgr B2(N) with a source size of 2 $\hbox{$^{\prime\prime}$ }$ FWHM, a column density of 2.8 $\times$ 10¹⁶ cm^-2, an excitation temperature of 100 K, a linewidth of 7 km s^-1, and a centroid velocity of 64 km s^-1 (see Table 4). We estimated the abundance of amino acetonitrile to be 2.2 $\times$ 10^-9 in this compact region. We found no evidence for a possible colder, more extended emission. The compact emission peaks at position P1 (see Sect. 3.4 and Table 6), which is located 0.4 $\hbox{$^{\prime\prime}$ }$ South of the ultracompact H II region K2 where the 3.7 mm thermal dust continuum emission detected with the PdBI also peaks. This angular separation is at a level of 2 $\sigma$ only, so it may not be significant. Our PdBI and ATCA data show that ethyl cyanide C₂H₅CN, cyanoacetylene HC₃N in its excited states v₇ = 1 and v₄ = 1, HC¹³CCN in its excited state v₇ = 1, and methanol CH₃OH in its excited state $v_{{\rm t}}=1$ also peak at this position P1. The amino acetonitrile emission arises therefore from the hot core region called the ``Large Molecule Heimat'' (see Sect. 1.1.2). Our PdBI continuum data show that this compact region is extremely dense (1.7 $\times$ 10⁸ cm^-3) and massive (2340 $M_\odot$ ). The ultracompact H II region K2 is most likely still embedded in the dense, hot core traced by the thermal dust emission and seems therefore to be the youngest source among the numerous ultracompact H II regions populating the Sgr B2 molecular cloud. In addition, the LMH hosts the powerful Sgr B2(N) H₂O maser region. The distribution of the maser emission over $4\hbox{$^{\prime\prime}$ }$ $\times$ $2\hbox{$^{\prime\prime}$ }$ was mapped using Very Long Baseline Interferometry by Reid et al. (1988), who also fitted a kinematical model invoking expansion and rotation. The best fit center of expansion is displaced from our interferometric position P1 by ( $\Delta \alpha, \Delta \delta$ ) = ( $-0.49\hbox{$^{\prime\prime}$ },-0.66\hbox{$^{\prime\prime}$ }$ ), which is less than the combined positional uncertainty of the VLBI and PdBI data. The best fit radial component of the expansion velocity, 63 $\pm$ 3 km s^-1, is also in excellent agreement with the 64 km s^-1 that we obtain for the LMH (see Table 4). H₂O masers are associated with young stellar objects in their earliest stages when they drive powerful outflows, such as the one found in Sgr B2(N) which has a total velocity extent of $\sim$ $\pm 50$ km s^-1 (see, e.g., Reid et al. 1988). Thus, the water vapor maser provides evidence for the youth of the LMH, the very compact region where the amino acetonitrile emission originates from.

Bisschop et al. (2007) measured the abundances of various complex molecules in massive hot core regions and classified these molecules as ``cold'' (<100 K) or ``hot'' (>100 K). Based on the high abundances, the similar high rotation temperatures, and the relative constant abundance ratios of the oxygen-bearing species and two nitrogen-bearing species, they concluded that the ``hot'' molecules share a common solid state formation scheme. From an analysis of the emission of complex organic molecules in molecular clouds in the Galactic center region and a comparison to results previously obtained in hot cores, Requena-Torres et al. (2006) support also the scenario in which complex organic molecules are formed on the grain surfaces. The high temperature and abundance we measured for amino acetonitrile suggest it shares the same properties as the ``hot'' molecules found by Bisschop et al. (2007), which favors its formation on the grain surfaces, although its detection in other hot cores to check if it follows the abundance correlations found by Bisschop et al. (2007) and Requena-Torres et al. (2006) is needed to prove this conclusion.

Wirström et al. (2007) failed to detect amino acetonitrile in the hot cores Orion KL, W51 e1/e2, S140, and W3(OH) with the Onsala 20 m telescope. They found beam-averaged column density upper limits of 1.1-3.5 $\times$ 10¹³ cm^-2 for amino acetonitrile, while they detected vinylcyanide C₂H₃CN with a column density of $\sim$ 2 $\times$ 10¹⁴ cm^-2 in the first two sources. Our 30 m observations of Sgr B2(N) imply a column density of 8 $\times$ 10¹⁷ cm^-2 for vinylcyanide with a source size of 2.3 $\hbox{$^{\prime\prime}$ }$ (Belloche et al., in prep.), i.e. about 30 times our amino acetonitrile column density. If the column density ratio of these two species in Sgr B2(N) holds for other hot cores, then the observations of Wirström et al. (2007) were not sensitive enough to detect amino acetonitrile in their sources.

Our PdBI and ATCA data show that the double peak structure seen in many transitions detected with the 30 m telescope are produced by two sources separated by about 5.3 $\hbox{$^{\prime\prime}$ }$ in the North-South direction (positions P1 and P2, see Sect. 3.4). The centroid velocity difference between these two positions is about 9 km s^-1. The northern and more redshifted source (P2) is about twice weaker in the molecular emission detected in our 30 m data (Belloche et al., in prep.). These two sources were already detected in ethylcyanide with high-resolution observations (see, e.g., Jones et al. 2007; Liu & Snyder 1999; Hollis et al. 2003). Our interferometric data show that cyanoacetylene HC₃N and its isotopologue HC¹³CCN in their excited state v₇ = 1, and methanol in its excited state $v_{{\rm t}}=1$ are also detected toward both sources. In the PdBI spectra shown in Figs. 4c and d, we do not find a clear evidence for amino acetonitrile at a velocity of $\sim$ 73 km s^-1. There may be a hint of emission at a level about a factor of 2 lower than the emission toward P1 (see also Feature F3 in the 30 m spectrum in Fig. 4g), but it is below our 3 $\sigma$ detection limit. Therefore we cannot rule out that amino acetonitrile shares the same property as, e.g., ethylcyanide, cyanoacetylene, and methanol and is also present in the northern source P2 at a level about twice lower than in P1.

The molecular source P2 is not detected at a 3 $\sigma$ level of 26 mJy/ $3.4\hbox{$^{\prime\prime}$ }$ $\times$ $0.81\hbox {$^{\prime \prime }$ }$ -beam in the continuum map we obtained with the PdBI at 82.0 GHz (see Fig. 5o). On the other hand, we detect some emission with an intensity of 280 mJy/ $3.5\hbox{$^{\prime\prime}$ }$ $\times$ $2.4\hbox{$^{\prime\prime}$ }$ -beam in the continuum map obtained with the ATCA at a mean frequency of 95.3 GHz (see Fig. 7w). If this emission is more extended than the PdBI beam, then the PdBI upper limit translates into 79 mJy in the ATCA beam, which yields an unphysical frequency exponent of $\sim$ 8 for the continuum emission toward P2 (see Sect. 3.6). We suspect that the ATCA continuum toward P2 is contaminated by a low-level line emission. In any case, since the continuum emission toward P2 is much weaker than toward P1 at 3 mm, P2 must be less dense and/or less hot than the hot core P1. P2 coincides with a weak blob of emission in the 1.3 cm VLA map of Gaume et al. (1995) (see their Figs. 6 and 7). However, this blob is located within the shell-like, weak, extended emission associated with K5 and it is difficult to know if it is compact or not from the 1.3 cm map published by these authors. Therefore the hot core P2 traced by the molecular emission may also be associated with an ultracompact H II region, weaker than K2. Alternatively, if it is not directely associated with a compact source of free-free emission, it may have been formed by the interaction of the shell-like structure K5 with the ambient medium, and could be in an earlier stage of evolution than the hot core P1 (LMH) associated with the ultracompact H II region K2.

4.2 Amino acetonitrile in Sgr B2(M)

We do not detect amino acetonitrile in our 30 m survey toward Sgr B2(M). Using the same parameters as for Sgr B2(N) (100 K and a FWHM source size of $2\hbox{$^{\prime\prime}$ }$ ), we find a $\sim$ $3\sigma$ column density upper limit of 6 $\times$ 10¹⁵ cm^-2 in the LTE approximation. The column density of amino acetonitrile is thus at least a factor $\sim$ 5 weaker toward Sgr B2(M) than toward Sgr B2(N). This is not surprising since, e.g., Nummelin et al. (2000) found that hot-core-type molecules are more abundant in Sgr B2(N) by factors 3-8 as compared to Sgr B2(M).

4.3 Amino acetonitrile, a precursor of glycine?

Amino acids, building blocks of proteins and therefore key ingredients to explain the origin of life, have been found in meteorites on Earth. Their deuterium isotopic composition suggests that they, or at least their direct precursors, were formed in the cold interstellar medium (e.g. Pizzarello & Huang 2005). Looking for amino acids in the interstellar medium is therefore appealing. However, the simplest amino acid glycine has been intensively searched for in the past 30 years, but has unfortunately not been discovered yet (e.g. Snyder et al. 2005; Brown et al. 1979; Cunningham et al. 2007).

Amino acetonitrile was proposed early on as a possible direct precursor of glycine in the interstellar medium (e.g. Brown et al. 1977). The formation of glycine via a Strecker-cyanohydrin synthesis has long been favored (Peltzer et al. 1984; Ehrenfreund et al. 2001; Bernstein et al. 2004). This pathway involves a carbonyl compound (such as an aldehyde or a ketone), hydrogen cyanide, and ammonia, and produces the amino nitrile which, after hydrolysis, yields the amino acid. However, the Strecker synthesis cannot explain the higher deuterium fractionation of amino acids compared to hydroxy acids which was measured in meteorites (see Elsila et al. 2007, and references therein).

Amino acids were successfully produced in the laboratory by UV-photolysis of ice mixtures mimicking the mantles of insterstellar grains (Bernstein et al. 2002; Muñoz Caro et al. 2002). For an ice mixture composed of H₂O, CH₃OH, NH₃, and HCN, Woon (2002) proposed theoretically a pathway of radical-radical reactions involving the radicals t-HOCO and CH₂NH₂ produced by UV irradiation. This hypothesis was tested and verified experimentally by Holtom et al. (2005) with an ice mixture of CH₃NH₂ and CO₂ bombarded by energetic electrons mimicking the impact of cosmic rays in the interstellar medium. This pathway leading to glycine does not involve the formation of amino acetonitrile. On the other hand, Elsila et al. (2007) experimented the UV-photolysis of an ice mixture of H₂O, CH₃OH, HCN, and NH₃, and found with isotopic labeling techniques multiple pathways leading to the formation of amino acids. The main pathway involves the formation of the amino nitrile and they proposed ``a modified radical-radical mechanism that takes into account the formation of nitriles as amino acid precursor molecules''. They also noticed that a Strecker-type synthesis may be at most a minor contributor to the formation of glycine.

The formation of glycine in the gas phase was also investigated. Blagojevic et al. (2003) synthesized ionized glycine via the reaction of the hydroxylamine ion NH₂OH⁺ with acetic acid CH₃COOH. They proposed the formation of the precursor hydroxylamine NH₂OH in the grain mantles and the formation of acetic acid via ion-molecule reactions in the gas phase. Based on quantum chemical calculations, Maeda & Ohno (2006) found another pathway to form glycine in the gas phase involving barrierless reactions between closed-shell species. Their pathway starts from CO₂, NH₃, and CH₂, and leads to glycine via the reaction of CO₂ with the closed-shell molecule CH₂NH₃, a higher energy isomer of methylamine CH₃NH₂. However, they mentioned that CH₂NH₃ should be efficiently destroyed by H₂O, so this pathway may be unlikely in the interstellar gas phase where water can be very abundant. Both gas phase formation routes do not involve amino acetonitrile as a direct precursor of glycine.

This brief overview of the experimental and theoretical work on the formation of amino acids in the interstellar medium shows that there is no consensus about the chemical precursors of amino acids. It is however important to note that the amino acids produced in the ice experiments mentioned above (except Holtom et al. 2005) are found experimentally after the hydrolysis of the ice residues. It is possible that only their precursors (e.g. amino nitriles) are synthesized by the ice photochemistry, and that the amino acids are formed only later, e.g. on the comet/asteroids surfaces, by hydrolysis (Elsila et al. 2007). Therefore amino acetonitrile may well be a direct precursor of glycine.

The formation of amino acetonitrile itself was also investigated theoretically by Koch et al. (2008). They found that water can efficiently catalyze a reaction between methylenimine CH₂NH and hydrogen isocyanide HNC to form amino acetonitrile in the grain mantles at a temperature of 50 K. Methylenimine was detected in the gas phase toward Sgr B2(N) by, e.g., Nummelin et al. (2000). They found evidence for both hot, compact and cold, extended components and derived a column density of 3.3 $\times$ 10¹⁷ cm^-2 for the compact component, with a source size of 2.7 $\hbox{$^{\prime\prime}$ }$ and a temperature of 210⁺⁴⁰⁰_-80 K, which is consistent with our own analysis (Belloche et al., in prep.). This column density is an order of magnitude larger than the column density we derived for amino acetonitrile, which does not rule out methylenimine as a precursor of amino acetonitrile.

4.4 Glycine in Sgr B2(N)

The frequency coverage of our 30 m survey of Sgr B2(N) includes many transitions of glycine as listed in the CDMS catalog (entries 75 511 and 75 512), but we do not detect this molecule within the limits of our LTE analysis. Using the same parameters as for amino acetonitrile (100 K and a source size of 2 $\hbox{$^{\prime\prime}$ }$ , see Table 4), we derive a column density upper limit of 2.0 $\times$ 10¹⁷ cm^-2 for conformer I and 5.0 $\times$ 10¹⁵ cm^-2 for conformer II. Alternatively, the upper limit on emission from glycine more extended than the 30 m beam at 3 mm ( $\sim$ $26\hbox{$^{\prime\prime}$ }$ ) is 1.2 $\times$ 10¹⁵ cm^-2 for conformer I and 3.0 $\times$ 10¹³ cm^-2 for conformer II. For a temperature of 75 K, we find column density upper limits for conformer I of 1.5 $\times$ 10¹⁷ cm^-2 for a source size of 2 $\hbox{$^{\prime\prime}$ }$ and 8.9 $\times$ 10¹⁴ cm^-2for emission more extended than the 30 m beam, and for conformer II 3.7 $\times$ 10¹⁵ cm^-2 and 2.2 $\times$ 10¹³ cm^-2, respectively.

Jones et al. (2007) did not detect glycine conformer I in Sgr B2(N) with the ATCA and derived a 3 $\sigma$ upper limit of 1.4 $\times$ 10¹⁵ cm^-2 for the beam-averaged column density at 75 K, which translates into an upper limit of 2.0 $\times$ 10¹⁶ cm^-2 for a source size of 2 $\hbox{$^{\prime\prime}$ }$ after correction for beam dilution ( $17.0\hbox{$^{\prime\prime}$ }$ $\times$ $3.4\hbox{$^{\prime\prime}$ }$ ). This upper limit on any compact emission from glycine is at a level 8 times lower than the one we derive with the 30 m telescope. As Jones et al. (2007) mentioned, the tentative detection of Kuan et al. (2003) is inconsistent with this upper limit in the case of compact emission. On the other hand, the ATCA non-detection does not exclude extended emission at the level found by Kuan et al. (2003) who reported a beam-averaged column density of 4.2 $\times$ 10¹⁴ cm^-2 with the NRAO 12 m telescope and a rotational temperature of 75 K. However Cunningham et al. (2007) did not detect glycine conformer I in Sgr B2(N) with the 22 m Mopra telescope and derived a 3 $\sigma$ upper limit of 3.7 $\times$ 10¹⁴ cm^-2 for the beam-averaged column density at 75 K. This upper limit is at a level 2.4 times lower than our beam-averaged upper limit. However, their upper limit was derived for a position offset by 26 $\hbox{$^{\prime\prime}$ }$ from the hot core position, so it only rules out emission from Sgr B2(N) more extended than $\sim$ $50\hbox{$^{\prime\prime}$ }$ in diameter. If glycine's emission were centered on the hot core position with a diameter of $\sim$ $25\hbox {$^{\prime \prime }$ }$ , Mopra would have missed $\sim$ $80\%$ of the flux, and their upper limit would be 1.9 $\times$ 10¹⁵ cm^-2. In that case, our upper limit is more significant, but not low enough to rule out the tentative detection reported by Kuan et al. (2003) if glycine is confined to a source size of $\sim$ $30\hbox{$^{\prime\prime}$ }$ . However, the arguments presented by Snyder et al. (2005) do rule out this case. As a conclusion, the upper limits of Cunningham et al. (2007), Jones et al. (2007), and Snyder et al. (2005) rule out emission of glycine conformer I at the level reported by Kuan et al. (2003) in Sgr B2(N) for any source size.

Cunningham et al. (2007) found an upper limit of 7.7 $\times$ 10¹² cm^-2for the beam-averaged column density of glycine conformer II with Mopra toward the central position of Sgr B2(N). This upper limit is nearly a factor of 3 lower than the upper limit we derived above with the 30 m telescope for extended emission. With the ATCA, Jones et al. (2007) found an upper limit of 8.6 $\times$ 10¹³ cm^-2 for the beam-averaged column density, which translates into 1.2 $\times$ 10¹⁵ cm^-2 for a source size of 2 $\hbox{$^{\prime\prime}$ }$ . This upper limit is again a factor of 3 lower than the upper limit we derived above with the 30 m telescope for compact emission.

Bernstein et al. (2004) found experimentally that organic acids are less stable than organic nitriles against UV photodestruction but they concluded that in dense molecular clouds, the ratio nitrile to acid should be affected by less than a factor of 2 over the lifetime of the cloud. Therefore it could be instructive to compare the pairs methylcyanide/acetic acid (CH₃CN/CH₃COOH) and amino acetonitrile/glycine (NH₂CH₂CN/NH₂CH₂COOH). In our 30 m line survey (Belloche et al., in prep.), we derive a column density ratio on the order of 200 for CH₃CN/CH₃COOH toward Sgr B2(N). If the two pairs are produced by similar chemical pathways yielding similar column density ratios, then we expect the glycine column density to be two orders of magnitude smaller than the amino acetonitrile column density, i.e. about 2 $\times$ 10¹⁴ cm^-2 for a compact source of 2 $\hbox{$^{\prime\prime}$ }$ diameter, which is nearly two orders of magnitude smaller than the upper limit derived above for glycine conformer I from the ATCA measurements of Jones et al. (2007), and a factor 5 smaller for conformer II. Therefore glycine emission may be well below the confusion limit in Sgr B2(N).

5 Conclusions

We used the complete 3 mm and partial 2 and 1.3 mm line surveys obtained with the IRAM 30 m telescope toward the hot cores Sgr B2(N) and (M) to search for emission from the complex molecule amino acetonitrile. We carried out follow-up observations with the IRAM Plateau de Bure and ATCA interferometers at selected frequencies. We also looked for extended emission from cold amino acetonitrile with the VLA. We report the detection of amino acetonitrile toward the hot core Sgr B2(N)-LMH, which is the first detection of this molecule in the interstellar medium. Our main results and conclusions are the following:

References

Online Material

Table 2: Transitions of amino acetonitrile observed with the IRAM 30 m telescope toward Sgr B2(N). The horizontal lines mark discontinuities in the observed frequency coverage. All lines with $S\mu ^2$ < 20 D² are not listed, since they are expected to be much too weak to be detectable with our sensitivity.

$\begin{figure} \par\includegraphics[angle=270,width=7.5cm,clip]{9203f101.ps}\hsp... ...ce*{0.5cm} \includegraphics[angle=270,width=7.3cm,clip]{9203f108.ps}\end{figure}$

Figure 1: Transitions of amino acetonitrile (AAN) detected with the IRAM 30 m telescope. Each panel consists of two plots and is labeled in black in the upper right corner. The lower plot shows in black the spectrum obtained toward Sgr B2(N) in main-beam temperature scale (K), while the upper plot shows the spectrum toward Sgr B2(M). The rest frequency axis is labeled in GHz. The systemic velocities assumed for Sgr B2(N) and (M) are 64 and 62 km s^-1, respectively. The lines identified in the Sgr B2(N) spectrum are labeled in blue. The top red label indicates the AAN transition centered in each plot (numbered like in Col. 1 of Table 3), along with the energy of its lower level in K ( $E_{\rm l}/k_{{\rm B}}$ ). The other AAN lines are labeled in blue only. The bottom red label is the feature number (see Col. 8 of Table 3). The green spectrum shows our LTE model containing all identified molecules, including AAN. The LTE synthetic spectrum of AAN alone is overlaid in red, and its opacity in dashed violet. All observed lines which have no counterpart in the green spectrum are still unidentified in Sgr B2(N).

Open with DEXTER

$\begin{figure} \par\includegraphics[angle=270,width=7.5cm,clip]{9203f109.ps}\hsp... ...e*{0.5cm} \includegraphics[angle=270,width=7.3cm,clip]{9203f116.ps} \end{figure}$	Figure 1: continued.
Open with DEXTER

Detection of amino acetonitrile in Sgr B2(N),,

1 Introduction - general methods

1.1 The target: Sagittarius B2

1.1.1 Sgr B2 as part of the Central Molecular Zone

1.1.2 The Large Molecule Heimat

1.2 The complex spectra of complex molecules

1.3 Amino acetonitrile

2 Observations and data reduction

2.1 Single-dish observations and data reduction

2.2 Interferometric observations with the PdBI

2.3 Interferometric observations with the ATCA

2.4 Interferometric observations with the VLA

3 Identification of amino acetonitrile

3.1 Amino acetonitrile frequencies

3.2 Modeling of the 30 m line survey

3.3 Detection of amino acetonitrile with the 30 m telescope

3.4 Mapping amino acetonitrile with the PdBI

3.5 Mapping amino acetonitrile with the ATCA

3.6 Abundance of amino acetonitrile in Sgr B2(N)

3.7 Limits on possible extended emission of cold amino acetronitrile with the VLA

4 Discussion

4.1 Amino acetonitrile in Sgr B2(N)

4.2 Amino acetonitrile in Sgr B2(M)

4.3 Amino acetonitrile, a precursor of glycine?

4.4 Glycine in Sgr B2(N)

5 Conclusions

References

Online Material

Detection of amino acetonitrile in Sgr B2(N)^,^,