Astron. Astrophys. 341, 882–895 (1999) ASTRONOMYAND

ASTROPHYSICS

Vibrationally excited HC3N toward hot cores

F. Wyrowski1,2,?, P. Schilke2, and C.M. Walmsley3

1 I. Physikalisches Institut der Universitat zu Koln, Zulpicherstrasse 77, D-50937 Koln, Germany2 Max-Planck-Institut fur Radioastronomie, Auf dem Hugel 69, D-53121 Bonn, Germany3 Osservatorio Astrofisico di Arcetri, Largo E.Fermi 5, I-50125 Firenze, Italy

Received 15 July 1998 / Accepted 22 October 1998

Abstract. We report IRAM 30-m observations of vibrationallyexcited rotational transitions ofHC3N toward a sample of sixultra–compact Hii regions. In addition, Effelsberg 100-m andPlateau de Bure Interferometer measurements were conductedtoward the region G10.47+0.03. We detected thev7 = 1 stateof cyanoacetylene toward all six regions and thev6 = 1 linetoward five of them. Toward G10.47+0.03, we detected linesfrom 11 different vibrationally excited states with excitationenergies up to 1600 K above ground. Of these, six are first de-tections. We also have detected several transitions of13C sub-stituted cyanoacetylene in thev6 = 1, v7 = 1, andv7 = 2states and provide improved rest frequencies for several of these.The population distribution of the vibrationally excitedHC3Nmolecules is analyzed and it is concluded that our results areconsistent with thermalised level populations at a temperatureof 270 K. From the interferometer results we find the source ofHC3N emission to be coincident withNH3(4,4) andCH3CNemission within the positional error of 1′′.

In an appendix, we report the first astronomical detection oflines from vibrationally excited HNCO in G10.47+0.03.

Key words: ISM: clouds – ISM: Hii regions – ISM: molecules– radio lines: ISM

1. Introduction

Massive O-B stars form in dense condensations within molec-ular clouds. The high dust extinction (AV ∼ 1000) within suchcondensations obscures their innermost parts in the near andmid-IR making it difficult to study the internal structure in thesewavelength regions. Recently it has been recognized that onetracer for O-B stars in the earliest phases of their evolution isthe presence of emission from high excitation molecular lines.Such molecular hot cores are usually found in the neighborhoodof compact Hii regions and have masses in the range 10-3000M� and temperatures of order 200 K (Churchwell 1991). Oneof the important characteristics of these objects is their internal

Send offprint requests to: F. Wyrowski, Maryland address? Present address:University of Maryland, Department of Astron-

omy, College Park, MD 20742, USACorrespondence to: wyrowski@astro.umd.edu

radiation field or (equivalently) the distribution of dust temper-ature. Since the infrared radiation in the interior is not directlyobservable, due to the extinction mentioned above, one has toresort to indirect methods. One such method is observing rota-tional transitions at radio wavelengths excited due to radiativepumping by IR dust photons.

Most of the work done so far has been concerned with the“Hot Core” in the Orion KL region and Sgr B2. Churchwellet al. (1986) analyzed HNCO as a probe of FIR radiation. Vi-brationally excitedHC3N has been studied by Goldsmith etal. (1982, 1985). A recent interferometer molecular line surveyof Orion KL by Blake et al. (1996) with OVRO gives insightinto the spatial distribution of HNCO and vibrationally excitedHC3N. Other examples of vibrationally excited molecules stud-ied in star formation regions are HCN (Ziurys & Turner 1986),NH3 (Mauersberger et al. 1988),CH3CN (Clark et al. 1976;Olmi et al. 1996a), and maser emission of SiO (Snyder & Buhl1974).

In the last few years, new hot core sources have joinedthe classical ones, among them are the particularly interestingyoung embedded objects associated with the UC Hii regionsG10.47+0.03 and G31.41+0.31. They have been studied usingthe VLA in high excitation ammonia lines (Cesaroni et al. 1994,1998) and with both the 30-m telescope and Plateau de Bureinterferometer in methyl cyanide (Olmi et al. 1996a, b). Bothsources have also been observed in recent studies of15NH3(Wyrowski & Walmsley 1996) and ofC17O (Hofner et al. inpreparation). The sources are associated with hot, high columndensity molecular gas with characteristics similar to those ofthe hot core seen toward the Orion KL region but much moremassive. From the dust emission the gas masses of the compactcores are found to be of the order of 2000 M� within a regionof about 0.05 pc. As contrasting cases, we decided to observeW3(H2O) (a hot core comparable to Orion-KL) and the coresG10.62–0.38, G29.96–0.02, and G34.26+0.15 (Churchwell etal. 1990).

In this article, we present 30-m telescope measurements ofvibrationally excitedHC3N toward this sample of hot cores.Since these single dish results do not provide direct informationon the position and size of the source responsible for the vi-brationally excited lines, we have supplemented the 30-m datawith Plateau de Bure Interferometer observations at 110 GHz

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 883

Table 1.Wavelengths of the IR pump radiation of different vibrationalmodes ofHC3N, their Einstein coefficients and critical densities.

Transition λEXC Aul ncr vibration(µm) (s−1) cm−3

v4 = 1 − 0 11 0.003 2 · 1010 C-C stretchv5 = 1 − 0 15 2.2 7 · 1012 HCC bendv6 = 1 − 0 20 0.15 3 · 1011 CCN bendv7 = 1 − 0 45 6 · 10−4 4 · 108 CCC bend

of the vibrationally excitedHC3N emission toward the spec-tacular source G10.47+0.03. It is also of interest to get infor-mation on the wavelength dependence of the line intensity andhence we have obtained Effelsberg 100-m observations of thevibrationally excitedHC3N transitions at 36 and 45 GHz forcomparison with the higher frequency data.

2. Line frequencies for vibrationally excitedHC3N

The linear pentatomic molecule cyanoacetylene has sevenmodes of vibration, four stretching modes and three bendingmodes which are doubly degenerate (Mallinson & Fayt 1976).An overview of the various fundamental vibrations is given inTable 1. The stretching modesν1, ν2, ν3, which lie more than3000 K above ground, are omitted. Besides the fundamentalmodes of vibration also overtones and combinations modes canbe observed. A level diagram of all modes lower than 1700 Kis shown in Fig. 1. Several of these modes are coupled by an-harmonic resonances (Yamada & Creswell 1986). Laboratorystudies of the infrared spectrum have been made firstly by Turrelet al. (1957) and more accurately by Mallinson & Fayt (1976)who determined band centers for all seven fundamentals withhigh precision. The first interstellar detection of vibrationallyexcited states of this molecule was by Clark et al. (1976) to-ward the Orion hot core.

Since most of the vibrationally excited rotational lines ob-served by us are first detections in interstellar space, we wouldlike to discuss briefly how the rest frequencies were obtainedwith the help of the molecular constants derived from laboratorymeasurements. These extrapolations were necessary to identifyobserved lines that lie outside the frequency ranges analyzed inthe laboratory, and to predict frequencies in other wavelengthranges for future studies.

For the purpose of determining the frequencies of the transi-tions observed astronomically, we have used where possible theeffective rotational constantsBeff andDeff , which Yamada &Creswell (1986) used to fit their laboratory measurements withsufficient accuracy for all the different levels shown in Fig. 1.This accounts forl-type doubling and resonance interaction be-tween different vibrational levels. We thus used the equation:

ν = 2Beff(J + 1) − 4Deff(J + 1)3 (1)

These effective constants were only available for the mainisotopomer. We calculated the frequencies of the less abundantisotopomeric species with the help of the parametersB0, D0,

Fig. 1. Vibrational energy levels of HCCCN.v4 is the quantum numberof the lowest stretching vibration, andv5,v6,v7 are bending vibrations.All detected lines are marked and the remaining were not covered byour observing bands.

Table 2. Molecular constants ofHC3N for the observed isotopomertaken from Mallinson & de Zafra (1978).

Constant HCCCN HC13CCN HC13CCN HCC13CN

B0/MHz 4549.058 4408.438 4529.755 4530.195α5/MHz −1.813 −1.387 −1.558 −1.552α6/MHz −9.262 −8.960 −9.075 −8.983α7/MHz −14.605 −14.067 −14.020 −14.184q5/MHz 2.570 2.428 2.519 2.516q6/MHz 3.582 3.367 3.593 3.635q7/MHz 6.538 6.175 6.579 6.529D0/kHz 0.5444

αi, andqi determined from the analysis of transitions in the26.5–40.0 GHz region by Mallinson & de Zafra (1978) whichare given in Table 2:

Bv = B0 − αi (2)

ν0 = 2Bv(J + 1) − 4D0(J + 1)((J + 1)2 − 1) (3)

ν = ν0 ± qi(J + 1) (4)

The accuracy of extrapolating this fit of the low frequencytransitions to high frequency transitions in the range from 3 to1.3 mm is a priori unknown, but we will see in Sect. 4 that thisprocedure works surprisingly well and leads to the identificationof 13C substitutedHC3N in vibrationally excited states withoutambiguity.

3. Observations

The bulk of our observations were carried out in July 1996using the IRAM 30-m telescope on Pico Veleta. However, wealso carried out supplementary observations of vibrationally ex-cited cyanoacetylene toward G10.47+0.03 using the Effelsberg

884 F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores

100-m telescope operated by the Max-Planck-Institut fur Ra-dioastronomie (MPIfR) and the Plateau de Bure interferometer(PdBI) operated by IRAM. In this section, we discuss these inturn.

3.1. 30-m observations

We observed simultaneously at 3, 2, and 1.3 mm with half powerbeam widths of 22′′, 16′′, and 12′′, respectively. Lists of theobserved sources and frequencies are given in Tables 3 and 4.The main targets were the sources G10.47+0.03, G31.41+0.31,and W3(H2O), but some other sources were observed in a fewfrequency bands as well. In the following tables, the sourcesare distinguished by the numbers given in the last column ofTable 3.

We used the facility 3 mm, 2 mm, and 1.3 mm (G1 and G2)SIS receivers with system temperatures of 300–450 K, 650–900 K, and 800–3000 K, respectively, depending on weather andelevations. The alignment between the different receivers waschecked through continuum cross scans on planets and foundto be accurate to within 2′′.

Our spectrometers were an autocorrelator with 1633 chan-nels and 0.32 MHz resolution and two filter-banks with 1 MHzresolution and 512 channels. This led to typical velocity resolu-tions between 1 and 2km s−1. All lines have been observed inthe lower sideband with typical sideband rejections of 30, 10,20 dB at 3, 2, and 1.3 mm, respectively. With the filter banks, wecovered 500 MHz centered on the frequencies given in Table 4.

The wobbling secondary mirror was used with a beam throwof 120′′ and a frequency of 0.25 Hz resulting in spectra withlinear baselines. To obtain main beam brightness temperatureas intensity scale, the antenna temperature has been multipliedby the ratio of the forward and the main beam efficiency whichwere taken from Table A1 of the 30-m manual (Wild 1995).

The focus was checked at the beginning of each night onJupiter or Saturn. Pointing was checked hourly by cross scanson planets, G10.62–0.38, and W3(OH) and was found to begood within 4′′.

3.2. Effelsberg 100-m observations

Using the 100-m telescope, we observed theJ=4–3 andJ=5–4transitions ofHC3N at 36 and 45 GHz, respectively. The 36 GHzobservations were carried out in two sessions in Jul./Aug. 1997and the 45 GHz observations in Apr. 1998. The frontends werethe facility 1 cm and 7 mm HEMT receivers with receiver tem-peratures of 120 and 70 K, respectively. Our spectrometer was a8192 channel autocorrelator which we used with 8 overlappingsubunits of 80 MHz bandwidth to cover a total of 490 MHz.The resulting spectral resolution was 0.3 MHz after smoothingthe data to increase the signal-to-noise. The beams at the fre-quencies of theHC3N lines were 25′′ and 20′′. Pointing waschecked at roughly hourly intervals by means of continuumscans through G10.62–0.38. We found the pointing to be accu-rate to within 4′′. Our calibration was based upon the continuumscans through G10.62–0.38, assuming a flux density of 4.3 Jy

Table 3.Observed sources

Source α2000 δ2000 vLSR

(km s−1)

G10.47+0.03 18:08:38.28 −19:51:50.0 +67.0 1G10.62−0.38 18:10:28.67 −19:55:50.1 +67.0 2G29.96−0.02 18:46:03.78 −02:39:21.9 +98.7 3G31.41+0.31 18:47:34.32 −01:12:45.8 +97.0 4G34.26+0.15 18:53:18.61 +01:14:57.3 +59.1 5W3(H2O) 02:27:04.63 +61:52:24.7 −49.0 6

Table 4.Frequency setup

νcenter in GHz observed sources

109.41 1,4110.095 1-6

154.23 1,3,4,6154.85 1-6

218.97 1,4,6228.1 1,4

at 43 GHz (Wood et al. 1988) corresponding to a main-beambrightness temperature of 6.1 K. The relative calibration be-tween the 36 and 45 GHz data was established with the helpof the simultaneously observed H52α and H56α recombinationlines assuming their intensities to scale simply inversely withfrequency.

3.3. Plateau de Bure observations

We observed theJ =12–11 transition ofHC3N in Jul.–Oct. 1997 using the Plateau de Bure Interferometer with 4 an-tennas in the C2/C1/B1/B2 configurations (a description of theinstrument is given by Guilloteau et al. 1992). The primary beamof the antenna at a frequency of 109.25 GHz was 44′′ and thesynthesized beam was5.2×1.6′′. The correlator was split into 6units of 80 MHz width and 128 channels to cover the frequencyrange from 109.0 to 109.5 GHz in the upper sideband and from106.0 to 106.5 GHz in the lower sideband of the SIS receivers.This correlator setup was slightly changed after observationof the first configuration to cover allHC3N lines in the fre-quency range. This lead to a non-uniform UV-coverage acrossthe total frequency range. The resulting spectral resolution was1.7km s−1. To calibrate the phase, nearby point sources (1730-130, 1908-201) were observed every 5 minutes. The bandpasscalibration was carried out using 3C273, 2230+114, and 1730-130.

The absolute flux scale was checked by observations ofMWC 349 (assuming a flux of 1.1 Jy at 109.3 GHz) and isfound to be correct within 15%. A continuum map was builtusing “line-free” spectral ranges at 109.05, 109.2, and 109.33GHz. The total continuum flux found is 0.9 Jy. As another checkwe determined the flux of the continuum of G10.47+0.03 foreach observing day and found a day to day variation of 10%.

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 885

Table 5.ObservedHC3N(17–16) intensities. Note that thev7 andv6

lines at 155035 MHz are blended and the total intensity is given.

ν Vib. state Source∫

TMB dv(MHz) (K km s−1)

154657.281 (0,0,0,0) G10.47+0.03 43.0±1.0G10.62−0.38 46.0±0.4G29.96−0.02 22.6±0.4G31.41+0.31 37.5±0.9G34.26+0.15 49.6±0.9W3(H2O) 13.1±0.2

154910.969 (0,0,1,0)1− G10.47+0.03 6.6±0.7G10.62−0.38 0±0.2G29.96−0.02 0.6±0.1G31.41+0.31 1.7±0.4G34.26+0.15 1.7±0.3W3(H2O) 0.1±0.05

155032.734 (0,0,1,0)1+ G10.47+0.03 23.2±0.5+155037.359 (0,0,0,1)1− G10.62−0.38 0.5±0.2

G29.96−0.02 6.1±0.3G31.41+0.31 11.0±0.3G34.26+0.15 5.5±0.7W3(H2O) 2.6±0.1

Due to the low elevation of the source and only modestweather conditions during summer, the phase noise lead todecorrelation on the longer baselines resulting in a “radio see-ing” of about 1.′′5.

4. Observational results

In this section, we present our observations starting with the30-m results.

We observed theJ=17–16HC3N lines near 154 GHz to-ward all six of the sources in Table 3. Within this sample, themost spectacular case is G10.47+0.03 for which we have alsomeasurements at 3 and 1.3 mm. We summarize these and discussthe line identifications. Then, we present our 100-m measure-ments of two lowJ transitions toward G10.47+0.03. Finally wediscuss the Plateau de Bure results of theJ =12–11 transitionof HC3N.

4.1. The J=17–16HC3N observations near 154 GHz

Near 154 GHz, we detectedHC3N(17–16) in its ground andlowest (v7) vibrationally excited states toward all six sourcesand thev6 state in all sources except G10.62–0.38. The ob-served line parameters are given in Table 5 and the spectra areshown in Fig. 2. To analyze the different excitation conditions inthe sources, we plotted the ratios of the integrated line intensitiesagainst each other in Fig. 3. For comparison purposes, the well-known hot core sources Orion KL and Sgr B2N were added withtheir line intensities taken from the literature (Goldsmith et al.1985 and de Vicente 1994). Since the vibrational ground stateline of HC3N is expected to be dominated by more extendedemission, the ratio between thev7 and the ground state is not a

Fig. 2.HC3N (17–16) spectra at 155 GHz toward all observed sources.

measure of the vibrational temperature but rather a rough mea-sure of the relative amount of molecules in vibrationally excitedstates in the whole region, whereas the ratio of the intensitiesIv6/Iv7 in the excited statesv6 andv7 can be converted to avibrational temperatureTvib of the hot core assuming opticallythin conditions:

I6

I7= exp

(−∆E67

kTvib

)(5)

where ∆E67 denotes the energy difference between thev6and v7 states. Ratios of 0.1, 0.2, and 0.4 then correspond toTvib =170, 250, and 430 K, respectively. This assumes opti-cally thin emission which is, as discussed below, not valid in

886 F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores

Fig. 3. Ratios of line intensities in the vibrational statesv0, v7 andv6

toward the observed hot cores and, additionally, for the hot cores OrionKL and Sgr B2N where the values of the (12–11) intensities are takenfrom Goldsmith et al. (1985) and de Vicente (1994).

G10.47+0.03 in which even13C isotopomer lines ofHC3Nwere observed, but a good approximation for the other sources,in which we don’t find any13C isotopomer lines. There is atrend for higher vibrational temperatures with a larger amountof vibration relative to the extended ground state emission. Theonly exception is G34.26+0.15. In this complex, UC Hii regionsof different morphologies are found (e.g. Gaume et al. 1994).Hence, star formation in different evolutionary states is presentwithin our 30-m beam. The high vibrational temperature couldbelong to the dense cores associated with the water masers andthe high ground state emission to the bulk of a halo forming sur-rounding gas (see the discussion of Garay & Rodriguez 1990 ondifferent ammonia structures in this source). In G10.62–0.38,the vibrationally excited emission is very weak. There is a trendfor higher excitation to be found in higher mass sources with themoderately massive cores W3(H2O) and Orion KL at the lowerend, and the very massive cores Sgr B2N and G10.47+0.03 atthe higher end. The position of G10.47+0.03 is exceptional incomparison to the other sources. In this source, many other vi-brationally excited states were detected as discussed in detail inthe next section.

4.2. Vibrationally excitedHC3N in G10.47+0.03

We detected rotational transitions in eleven vibrationally excitedstates ofHC3N (six of these are detected for the first time ininterstellar space) toward the hot core associated with the UCHii region G10.47+0.03. This allows the most detailed study ofvibrational (as opposed to rotational) excitation in a molecular

cloud so far. Since the line blending problem becomes severe at1.3 mm, theHC3N lines are most easily identified at 3 and 2 mm.We show the spectra obtained at these wavelengths together withidentifications and a formal fit withT=500 K in Figs. 4 and 5(details about the fit procedure are given by Schilke & Phillips1998, in preparation). Lines from vibrationally excited HNCOare detected as well and will be discussed in Appendix A.

We have detected transitions in the (v4, v5, v6, v7) =(0,0,0,1), (0,0,0,2), (0,0,0,3), (0,0,1,0), (0,0,2,0), (0,0,1,1),(0,0,1,2), (0,1,0,0), (0,1,0,1) bending vibrational states and alsothe (1,0,0,1) and (1,0,0,0) stretching vibrational levels. One maycompare with Orion, where, despite deep searches, only the(0,0,0,1), (0,0,0,2) and (0,0,1,0) states have been found (Clarket al. 1976; Goldsmith et al. 1982, 1985; Blake et al. 1987). Anoverview of the detected states is shown in Fig. 1 where onesees they range up to more than 1500 K above ground. Linesfrom levels marked as “not observed” lay outside our observingbands. The observed intensities of the lines are given in Table 6.

An example of modes with two quanta of vibration ex-cited is shown in the upper panel of Fig. 6. The combinationmode (0,1,0,1), which lies about 1300 K above ground, is splitinto four components with|l|=0, 2. This splitting is due tol-type doubling and, additionally, due to anharmonic resonancesin the (1,0,0,0)-(0,1,0,1)-(0,0,2,0)-(0,0,0,4) system (Yamada &Creswell 1986). From this four state resonance, we also see twov6 = 2 lines about 1450 K above ground. Here, the splittingof the |l|=2 lines is too small to be observed and both linesoverlap resulting in a relatively high intensity. One should notethe excellent coincidence of line frequencies measured in thelaboratory and astronomically which gives strong confidence inthe identification of these lines. Evidence for (0,0,1,1) lines isfound in a 1.3 mm spectrum near theC18O(2–1) line (Hofneret al. in prep.). Fig. 7 shows three spectral features close to thepredicted frequencies. Small errors in the prediction are pos-sible because of large perturbation terms in the Hamiltonian(Yamada & Creswell 1986). But the detection is further con-firmed by more lines from that mode observed with the 100-mtelescope (Fig. 8). Also all lines in our observed bands with threequanta of vibration excited are clearly identified ((0,0,1,2) and(0,0,0,3)) in the lower panel of Fig. 4. The highest lying detectedlines are three lines in the (1,0,0,1) mode 1600 K above ground.Although the signal-to-noise of these detections is rather low(3–5 σ), all three lines are very well reproduced by our LTEfit (Figs. 4 and 5, 109306.7, 109469.4, 154844.8 MHz) withT=500 K. The line at 155074.9 GHz seems to be blended withan U-line. A line in the pure stretching modev4 is detected at154443.8 MHz. A possible blend with the ethanol716 − 624line at 154442.7 MHz is believed to be weak (considering theresults of Wyrowski 1997, Chapter 5), but there is still an contri-bution from an unknown, extended spectral feature in the rangefrom 154434 to 154460 MHz. Another detection of a line ofthe v4 mode is reported in Sect. 4.4 with the Plateau de Bureinterferometer.

Moreover, we have also detected for the first time thev7 =1,2 and thev6 = 1 states of13C substitutedHC3N. No DC-CCN was found at the position of a (12–11)v7 line at 110015.4

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 887

Fig. 4. 30-m spectra ofHC3N(12–11) toward G10.47+0.03. The identified vibrational excited states are marked with dots and their vibrationalquantum numbers (v4, v5, v6, v7). The applied fit toHC3N assumes one single temperature for all lines of 500 K. Vibrationally excited HNCOis marked with dashes. The strong line at 110.2 GHz is the13CO line which shows several absorption features caused by the small wobblerthrow of 2′.

Fig. 5. Same as Fig. 4 butHC3N(17–16)

888 F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores

Table 6. HC3N intensities in G10.47+0.03 observed with the 30-mtelescope

ν Trans. Vib. state Eup∫

TMB dv(MHz) (K) (K km s−1)

HC3N109182.9 12–11 (0, 1, 0, 0)1− 988 2.0109244.0 12–11 (0, 1, 0, 0)1+ 988 2.4109306.7 12–11 (1, 0, 0, 1)1− 1627 0.6109352.8 12–11 (0, 0, 1, 0)1− 751 3.7109438.7 12–11 (0, 0, 1, 0)1+ 751 3.2109442.0 12–11 (0, 0, 0, 1)1− 354 7.1109469.4 12–11 (1, 0, 0, 1)1+ 1627 0.6109522.5 12–11 (0, 0, 2, 0)0 1469 0.7109549.5 12–11 (0, 1, 0, 1)0+ 1308 1.1109552.1 12–11 (0, 1, 0, 1)2+ 1308 1.3109558.0 12–11 (0, 1, 0, 1)0− 1308 1.3109563.7 12–11 (0, 1, 0, 1)2− 1308 1.4109598.8 12–11 (0, 0, 0, 1)1+ 354 7.9109616.3 12–11 (0, 0, 2, 0)2 1469 1.1109862.8 12–11 (0, 0, 0, 2)0 674 4.5109866.1 12–11 (0, 0, 0, 2)2− 674 2.4109870.3 12–11 (0, 0, 0, 2)2+ 674 4.7109990.0 12–11 (0, 0, 1, 2)1− 1391 1.4110035.6 12–11 (0, 0, 1, 2)−1 1391 1.8110051.0 12–11 (0, 0, 0, 3)− 994 2.7110097.6 12–11 (0, 0, 1, 2)3 1391 1.3110148.8 12–11 (0, 0, 1, 2)−1− 1391 1.2110189.8 12–11 (0, 0, 1, 2)1+ 1392 1.8110211.4 12–11 (0, 0, 0, 3)3 994 2.2155037.4 17–16 (0, 0, 0, 1)1− 387 15.4219173.8 24–23 (0, 0, 0, 1)1+ 451 17.3228303.2 25–24 (0, 0, 0, 1)1− 462 18.8

HCC13CN154259.6 17–16 (0, 0, 1, 0)1− 784 1.9154365.8 17–16 (0, 0, 0, 1)1− 386 3.8154370.6 17–16 (0, 0, 1, 0)1+ 784 2.4154969.9 17–16 (0, 0, 0, 2)0 706 1.2

HC13CCN154383.2 17–16 (0, 0, 1, 0)1+ 784 2.0154387.2 17–16 (0, 0, 0, 1)1− 386 4.1154943.8 17–16 (0, 0, 0, 2)0 706 1.3154954.5 17–16 (0, 0, 0, 2)2− 706 1.2154966.8 17–16 (0, 0, 0, 2)2+ 706 1.6

MHz (rms=20 mK), and only the HCCC15N (12–11) vibra-tionally ground state line has been observed with the PdBI (seeSect. 4.4). The identifications of these lines is shown in moredetail in the middle and lower panels of Fig. 6. The assignmentswere done according to the frequencies discussed in the lastsection. Thel-type doubling of the|l|=2 state was unknown andestimated to be equal to the splitting of the main isotopomer.Predicted and measured frequencies, given in Table 7, agreewithin a few MHz, which is remarkable since the frequenciesare extrapolated from centimeter laboratory measurements. Therelatively strong line at 154965 MHz in the lower panel of thefigure is presumably due to the blending of two lines of differ-

Fig. 6.HC3N spectra toward G10.47+0.03 showing in the upper panelthe combination modev5v7 and the overtone2v6, in the center panel thefundamental modesv7 andv6 of two 13C substitutedHC3N species,and in the lower panel their2v7 overtones. The fitted line is a Gaussianfit with the line widths fixed to 8.25km s−1 (the width measured forthev5 lines) and the frequency and intensity as free parameters.

ent isotopomers (HC13CCN v7 = 2+ and HCC13CN v7 = 20)and was fitted with two Gaussian lines. In contrast to the lineswith only one quantum of vibration excited, thev7 = 2 linesobserved show a systematic offset from the predictions. Thel-type doubling prediction itself is better than 0.5 MHz.

4.3. The Effelsberg 100-m measurements of HC3N

Four different vibrationally excited states ofHC3N in theJ=4–3andJ=5–4 transitions were detected with the Effelsberg 100-mtelescope. The spectra are shown in Fig. 8 and the line intensitiesare given in Table 8. The lines from states with more than onequantum of vibration are blends of several components and inthe table only the total intensity of all components is given.

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 889

Table 7. Observed frequencies of vibrationally excitedHC3N iso-topomers at 110 GHz with the PdBI and at 154 GHz with the 30-mtelescope.

Transition νobs ∆νobs O–C(MHz) (MHz) (MHz)

H13CCCNv7 = 1− 106063.0 0.6 0.7H13CCCNv7 = 1+ 106211.0 0.5 0.5H13CCCNv7 = 20 106473.1 0.4 –1.0H13CCCNv7 = 22− 106470.2 0.2 –0.5H13CCCNv7 = 22+ 106476.8 0.3 –1.5HC13CCNv7 = 1− 109125.7 0.1 –0.1HC13CCNv7 = 20 109382.1 0.3 –1.3HC13CCNv7 = 22− 109378.4 0.4 –1.7HC13CCNv7 = 22+ 109386.3 0.3 –1.4HCC13CN v7 = 20 109399.7 0.4 –2.1HCC13CN v7 = 22− 109396.5 0.4 –2.1HCC13CN v7 = 22+ 109404.8 0.3 –1.3

HCC13CN v6 = 1− 154259.1 1.6 –0.5HC13CCNv7 = 1− 154365.7 0.6 –0.1HC13CCNv6 = 1+ 154371.8 1.5 1.2HCC13CN v6 = 1+ 154382.5 1.6 –0.7HCC13CN v7 = 1− 154387.6 1.0 0.4HC13CCNv7 = 20 154941.0 0.4 –2.8HC13CCNv7 = 22− 154952.0 0.5 –2.5HC13CCNv7 = 22+ 154964.5 0.9 –2.3HCC13CN v7 = 20 154967.5 1.3 –2.4

Fig. 7. HC3N (0,0,1,1) lines detected close toC18O(2–1) inG10.47+0.03.

4.4. Plateau de BureHC3N J=12–11 results

With the Plateau de Bure Interferometer, we detect a large vari-ety of vibrationally excited states in ourJ=12–11 observationstoward G10.47+0.03. To produce spectra over the whole ob-served spectral range, we fit a point source model directly tothe uv-data. The results are shown in Fig. 9 and a summary ofthe detected lines is presented in Table 9 which gives the totalobserved flux density estimated from fitting elliptical Gaussianto the uv-data of the lines to avoid any artifacts introduced dur-ing the image processing. Since large phase noise on the longerbaselines led to smearing of the emission between 1 and 2 arcsec(“radio seeing”), the total fluxes are more reliable than the peakvalues. Of particular interest in Fig. 9 is the confirmation of the

Table 8.HC3N (4–3) and (5–4) intensities in G10.47+0.03 observedwith the 100-m

ν Trans. Vib. state Eup∫

TMB dv(MHz) (K) (K km s−1)

36392.33 (4–3) (0, 0, 0, 0) 4 14.736452.06 (4–3) (0, 0, 1, 0)1− 722 .836480.72 (4–3) (0, 0, 1, 0)1+ 722

+36481.82 (4–3) (0, 0, 0, 1)1− 324 3.236534.13 (4–3) (0, 0, 0, 1)1+ 324 2.436581.92 (4–3) (0, 0, 1, 1)0 1042

+36582.15 (4–3) (0, 0, 1, 1)0 1042+36583.53 (4–3) (0, 0, 1, 1)2 1042+36583.54 (4–3) (0, 0, 1, 1)2 1042 0.6

36623.24 (4–3) (0, 0, 0, 2)2 644+36623.38 (4–3) (0, 0, 0, 2)2 644+36623.41 (4–3) (0, 0, 0, 2)0 644 1.6

45490.31 (5–4) (0, 0, 0, 0) 7 30.945564.98 (5–4) (0, 0, 1, 0)1− 724 1.145600.80 (5–4) (0, 0, 1, 0)1+ 724

+45602.18 (5–4) (0, 0, 0, 1)1− 327 7.145667.56 (5–4) (0, 0, 0, 1)1+ 327 5.745727.13 (5–4) (0, 0, 1, 1)0 1044

+45727.49 (5–4) (0, 0, 1, 1)0 1044+45729.40 (5–4) (0, 0, 1, 1)2− 1044+45729.48 (5–4) (0, 0, 1, 1)2+ 1044 4.7

45778.95 (5–4) (0, 0, 0, 2)2− 647+45779.04 (5–4) (0, 0, 0, 2)2+ 647+45779.23 (5–4) (0, 0, 0, 2)0 647 4.5

Fig. 8.HC3N (4–3) and (5–4) lines detected with the Effelsberg 100-mtelescope in G10.47+0.03.

890 F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores

detection of thev4=1 stretching level at 1306 K above groundas well as the detection of the (1,0,0,1) transitions at 1621 K.Another significant result is the detection of several lines of thev7 = 1, 2 levels of13C isotopomers ofHC3N and the detectionof the HCCC15N vibrational ground state.

Predicted and measured frequencies of13C isotopomers aregiven in Table 7, and agree, as already the case for the 30-mobservations ofJ =17–16, within a few MHz. Again thev7 = 2lines observed show a systematic offset from the predictions tolower frequencies.

Our interferometer measurements allow us to determine theposition of the source of vibrationally excitedHC3N emis-sion. The position measured by us in the (1,0,0,0) transition at109023.3 Ghz isα(2000)=18:08:38.25,δ(2000)=–19:51:50.5,which is in agreement with the methyl cyanide position (Olmiet al. 1996b), taking the seeing effects of more than 1 arcsecinto account. Fig. 10 shows our observation superimposed ontheNH3(4,4) map of Cesaroni et al. (1998). Within our accu-racy of our 1′′, HC3N andNH3 have the same positional origin.To decide, whether the vibrational emission originates from aninterface region between hot core and UC Hii or is due to aseparate, embedded exciting star in the hot core, higher angularresolution observations are needed.

5. Cyanoacetylene level population distribution towardG10.47+0.03

5.1. Optical depths in the vibrationally excited transitions

A first estimate of the vibrational temperature in G10.47+0.03can be obtained by assuming optically thin conditions and plot-ting the observed intensities in a Boltzmann plot against exci-tation for all detected lines in the observed 109 GHz band withthe 30-m. Since the lines are observed simultaneously with thesame front- and backends, the relative calibration is excellent.The resulting plot is shown in Fig. 11 where one sees that oneobtains a reasonable fit with a formal excitation temperatureof 485 K (as compared to only 150 K in Orion). We omittedthe ground state which forms in a more extended cooler region(see Sect. 4.4) and is deviant by a factor 4. The analysis of the13C substitutedHC3N, detected at 155 GHz, leads to a similartemperature estimate, but the accuracy of the Boltzmann fit isconsiderably lower due to the lower signal-to-noise ratio of thedetections and due to possible contamination with U-lines (seeagain Fig. 11).

The ratios of intensities of the main isotopomer and of13Csubstitutes can be used to estimate the optical depths of thecorresponding transitions. This is done, assuming an12C/13Cabundance ratioX and equal excitation and filling factors forthe lines, with the following formula:

I12

I13=

1 − exp(−τ)1 − exp(−τ/X)

(6)

The ratioX is a function of galactocentric distance (Dahmen etal. 1995) and is, for a galactocentric distance of 3.0 kpc, 40±15.The resulting optical depths for thev7 lines ofHC3N(17–16) isthen 19±5 and for thev6 lines 20±10. Since the optical depth is

estimated for the 155 GHz range (orJ=17) and the vibrationaltemperature is determined at 110 GHz (orJ=12), we have tocalculate the expected optical depth at the latter frequency. Onecan show that for a linear molecule withTex � hν/k the ratioof τ at differentJk(k = i, j) is aboutJ2

i /J2j (see for example

Genzel 1992) in LTE and therefore, the optical depth at 110 GHzis approximately 10, consistent with the PdBI results. Hence,the assumption of optically thin lines, which we made to derive avibrational temperature at 110 GHz, is clearly violated. A betterguess can be obtained by considering our results of the (4–3)transition at 36 GHz; here the optical depths in thev7 line isexpected to be of order unity. A Boltzmann fit of the 36 GHzresults alone leads to an excitation temperature of270 ± 35 K.

Besides the population of levels of different vibrational ex-citation but constantJ , we can also study the level populationfor a constant state of vibration. We detected thev7 vibrationalstate at 8, 3, 2, and 1.3 mm and can therefore cover a wide rangeof rotational excitation within this vibrational mode. The result-ing rotational Boltzmann plot for all observedv7 lines is shownin Fig. 11. In contrast to the very high temperature found for thedifferent vibrationally excited states, the temperature which fitsthe rotational population of thev7 mode is only 45 K. This isagain misleading since, as discussed above, the optical depth ofthe lines is considerable and, due to the dependence of opticaldepth onJ , high J lines are depressed and a small rotationaltemperature is simulated.

Thus we have to interpret the observed intensities takingaccount of these high optical depths which can be done by in-cluding the optical depth into a LTE analysis of the intensities(e.g. Olmi et al. 1996a):∫

T dv =8π3νµ2S

3k

Ntot

ZXexp

(− E

kT

)G(τ) R(ν) η (7)

Every transition is characterized by its frequencyν, line strengthS andµ, the relevant component of the dipole moment of themolecule. The population of the upper level with energyE ofthe transition is described by the total column densityNtot ofthe molecule, its partition function as the product of rotationaland vibrational partZ = ZvZr, the isotopomeric ratioX, andthe temperatureT . The functionG(τ) corrects for the opticaldepth of the line (Stutzki et al. 1989):

G(τ) =∫

1 − exp(−τg(v))τ∆v

dv (8)

≈ 1 − exp(−τ ′)τ ′ , τ ′ = 0.679 τ0.911 (9)

The approximation of the integral with this expression is validfor a Gaussian line profile and leads to errors of 15% (4%) forτ in the range of 0.01 to 100 (0.01 to 10). In contrast to thesimpler formula using just the line center optical depth (givenby settingτ ′ = τ ) suggested by e.g. Turner (1991) and Olmi etal. (1996a), the correction also accounts for the line broadeningas function of optical depth. However, this one component ap-proach would produce rectangular line shapes for lines of highoptical depth, which are not observed. Indeed, even the highoptical depth lines are Gaussian suggesting a more complicated

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 891

Fig. 9.HC3N (12–11) lines observed with the PdBI in G10.47+0.03.

line forming process at work, e.g. macroturbulence or large ve-locity gradients. Having optically thin and thick lines available,the filling factorη can be estimated as well. The optical depthitself, which enters intoG(τ), is given by

τ =8π3νµ2S

3k

Ntot

ZXexp

(− E

kT

)1

T∆v(10)

The functionR(ν) in Eq. 7 corrects for the weakening of thespectral line emission by dust absorption due to the very hightotal column densities of about1025 cm−2 involved. We take:

R(ν) = exp[−τ1.3

(ν

ν1.3

)α](11)

Here the dust absorption optical depth at 1.3 mm isτ1.3 andthe frequency dependence of the dust opacity is assumed to beα = 2 in the millimeter wavelength range (see Wyrowski et al.1997 for an observational estimate of the dust opacity index).We include this effect here, since it could also cause the lowrotational temperature betweenv7 lines with differentJ andhence frequency (Fig. 11).

Using these formulae, one can search for the minimum ofthe corresponding reduced least square sumχ2 of this LTE fitto thend observed intensitiesIobs =

∫T dv:

χ2 =1

nd − np − 1

nd∑i=1

(Iobsi − Imod

∆Ii

)2

(12)

Thenp free parameters of the model areT , Ntot, θS , τ1.3, dv,and (because of the relatively large errors in the assumed iso-topomeric ratio)X. The sum is normalized with the number of

Fig. 10.Overlay of the integrated emission of theHC3N v4 line (thincontours: 50, 70, 90% of peak) with the VLA results of Cesaroni et al.(1998) showing the 1.3 cm continuum of the UC Hii regions as dashedcontours and the main line ammonia (4,4) emission as greyscale. Watermasers observed by Hofner & Churchwell (1996) are shown as crosses.

degrees of freedomnd − np − 1. The best fit values of severalparameter combinations are given in Table 10. The optical depthof the dust at 1.3 mmτ1.3 is only important for low values ofX.

892 F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores

Table 9.HC3N (12–11) intensities in G10.47+0.03 observed with thePdBI. The given flux is estimated from Gaussian fits to the uv data.

ν Vib. state Eup∫

Sν dv(MHz) (K) (Jy km s−1)

HC3N109023.3 (1, 0, 0, 0) 1306 5.0109173.6 (0, 0, 0, 0) 34 102.0109182.9 (0, 1, 0, 0)1− 988 9.3109244.0 (0, 1, 0, 0)1+ 988 11.1109306.7 (1, 0, 0, 1)1− 1627 1.9109352.8 (0, 0, 1, 0)1− 751 17.5109438.7 (0, 0, 1, 0)1+ 751

+109442.0 (0, 0, 0, 1)1− 354 49.2109469.4 (1, 0, 0, 1)1+ 1627 3.0

HCC13CN109401.9 (0, 0, 0, 2) 674 6.3

HC13CCN109125.8 (0, 0, 0, 1)1+ 354 6.3109383.4 (0, 0, 0, 2) 674 5.7

H13CCCN106062.3 (0, 0, 0, 1)1− 353 5.3106210.5 (0, 0, 0, 1)1+ 353 3.8106474.1 (0, 0, 0, 2) 673 9.5

HCCC15N105998.3 (0, 0, 0, 0) 33 2.8

Table 10. Results of LTE model fits including optical depth to theHC3N data. Theτcor? flag indicates, whether we used Eq. 8 or thesimpler expression withτ ′ = τ (indicated by no).

τcor? X ∆v χ2min T θS

(km s−1) (K) (arcsec)

no 20 8 1.9 325 1.130 2.3 295 1.040 2.9 275 0.9530 7 2.3 295 1.05

yes 20 8 1.7 320 0.830 2.0 290 0.7540 2.3 270 0.730 7 2.0 290 0.8

ForX > 25, one can obtain a good fit withR(ν) = 1. A Boltz-mann plot in which the intensities of all observed lines are cor-rected for optical depth, beam filling, and differentX, is shownin Fig. 12 and an examples of the dependence of theχ2−sumon the model parameters is given in the panels of Fig. 13: withfixed∆v andX, we varied the source size (upper panel), columndensity (lower panel) and temperature and determined then therest of the parameters to minimize theχ2−sum applying also acorrection for line broadening. The contours give the confidencelevels of 1, 2 and 3σ for all six degrees of freedom which confinethe source size to a range of 0.65′′ to 0.85′′ and the temperature

Fig. 11. Boltzmann plot of all detected lines of vibrationally excitedHC3N(12–11) in the observed band at 109 GHz (dots), allv7 lines(squares) with differentJ , and all lines of13C isotopomer at 155 GHz(triangles). All intensities were corrected to match the 109 GHz beamby assuming a source size much smaller than the beams.

Fig. 12. Boltzmann plot of all vibrationally excitedHC3N lines cor-rected for optical depth (without taking line broadening into account)with a source size of 0.95′′, an isotopomeric ratio of 40, and a line widthof 8km s−1. Different Symbols are used for different telescopes. Filledmarkers denote lines from the main isotopomer, whereas empty mark-ers denote lines from the13C and15N substituted isotopomer.

to a range of 240 K to 310 K for dust opacity solutions close to0. TheHC3N column density is about4 × 1018 cm−2.

Smaller values ofX lower the optical depths and accord-ingly the resulting column density, but increase the temperature.Without correcting for the line broadening due to optical deptheffects, larger source sizes are needed to account for the intensityof the optical thick lines.

5.2. The excitation of the lines

All of these estimates rely on the assumption of LTE which isprobably quite good in the light of the LTE fit which yieldedto Fig. 12. But for a full understanding of the optical depth ef-

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 893

Fig. 13.Results of LTE analysis of the observed vibrationally excitedHC3N lines assuming a [12C]/[13C] ratio of 40, a linewidth of 8km s−1

and using Eq. 8 to correct for optical depth line broadening. The con-tours give the confidence levels of 1, 2 and 3σ for six degrees of free-dom. In the upper panel, the curve represents the loci ofL = 1.5 LIRAS

(see Sect. 6).

fects, a detailed modeling of the level population by statisticalequilibrium calculation together with radiative transfer througha model cloud, is needed but would go beyond the objectivesof the present analysis. Here, we just want to estimate whetherthe lines are populated mainly by collision or radiation. At theso-called critical densityncrit = βAul/γul, whereAul is thetransition probability of spontaneous radiation,γul the colli-sional rate coefficient andβ the escape probability, collisionalexcitation can compensate for the radiative decay. Therefore, ifthe density is considerably lower than the critical density, onlyexcitation by the radiation field can lead to population of thelevels.

The radiative transition probability between vibrationalstates can be obtained by far-infrared absorption measurements.In the case ofHC3N this has been done by Uyemura & Maeda

(1974) and Uyemura et al. (1982) for all the fundamental stretch-ing and bending modes, respectively. The measured infraredintensities of the modes can then be converted to Einstein coef-ficients (see appendix of Deguchi et al. 1979) as has been doneto produce Table 1.

Since no measurements or calculations of the collisionalcross sections which change the vibrationally state ofHC3N areavailable, approximations have to be found. We have followedthe suggestion of Goldsmith et al. (1982), who used the semi-empirical formula for vibrational relaxation times as a functionof temperature, reduced mass, and energy of vibration in di-atomic molecules given by Millikan & White (1963). With thisrelation, we calculated deexcitation cross sections. This is donefor a temperature of 300 K and the resulting critical densitiesare shown in the last column of Table 1. The density in the hotcore associated with G10.47+0.03 is of the order 107 cm−3 (Ce-saroni et al. 1994; Wyrowski & Walmsley 1996) and thereforethe vibrationally excited states cannot be populated collision-ally, but require excitation by a strong IR radiation field unlessstrong radiative trapping in the infrared reduces the radiationprobability to an effective value ofAul/τIR where the opticaldepth in the IR could exceed the radio optical depth by a factor10 to 100 (Genzel 1992). This is likely for theHC3N ν7 statein extreme cases but not for more highly excited states.

6. Characteristics of the IR sources

Most of the critical densities of the IR transitions are much toohigh for them to be populated by collisions (cf. 5.2). Hence, theymust be radiatively excited and therefore probe the IR radiationfield of the observed sources at the pumping frequencies of therotational lines. The total column densities involved (1024 −1025 cm−2) lead to high optical depths in the infrared and theobtained vibrational temperatures of the IR levels are expectedto measure the dust temperature.

One important property of an IR source is its total bolometricluminosity which can be derived from the IRAS fluxes. Thisluminosity gives us a constraint on the possible temperatureand size of the source by applying the Stefan–Boltzmann law:

LIRAS ≥ πθ2Sd2σT 4 (13)

θS is the source size,d the distance of the source,T the tem-perature andσ the Stefan–Boltzmann constant. With the IRASluminosity LIRAS = 5 × 105 L� for G10.47+0.03 (Cesaroniet al. 1994), this constrains the temperature for assumed sourcesizes of 2′′ and 0.′′5 to 140 and 390 K, respectively. It should benoted that this assumes spherical symmetry and that the lumi-nosity for a disk-like geometry could be significantly smaller asdiscussed in Cesaroni et al. (1998).

Since the detection of13C species ofHC3N revealed thatthev7 lines of the main isotopomer are optically thick, we canuse the brightness temperature of these lines to relate sourcesize and temperature in another way:

T = TLθ2

B

θ2S

(14)

894 F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores

HereθB denotes the beam size. Combining both formulae, wecan calculate for each observedv7 line a unique source sizeand temperature (assuming equality in Eq. 13). The 3 and 2 mmlines lead to a size of 1.′′25 and 175 K whereas the 1.3 mm linesresult in a size of 0.′′7 and 230 K.

Additionally, we used the luminosity constraint of Eq. 13 forour LTE fit. In Fig. 13 the range of forbidden source sizes andtemperatures is marked with a curve, allowing for a 50% errorin the IRAS luminosity. The source size is then constrained tovalues between 0.′′65 and 0.′′85.

However, we do not expect this combination of IR-sourcesize and temperature to explainall observed vibrationally ex-cited lines. For the excitation of the very high lying modes highertemperatures are necessary extending over smaller regions. Themost likely scenario is an increase of temperature toward theinterior of the source (see also Wyrowski & Walmsley 1996)in the high lying states ofHC3N. This can only be revealed byfuture sub arcsecond interferometer studies of these lines.

7. Conclusions

A sample of six hot core sources has been observed invibrationally excited lines ofHC3N. Among the sources,G10.47+0.03 is unique in terms of high excitation molecularlines and revealed a panoply of new identifications ofHC3N invibrationally excited states, even from13C isotopomeric formsof HC3N. We derive an excitation temperatures of 250–330 Kfor the vibrationally excitedHC3N lines, depending on the as-sumed12C/13C abundance ratio, though still higher tempera-tures may excite the highest lying modes of vibration. This vi-brational temperature is consistent with the temperature foundusing15NH3 (Wyrowski & Walmsley 1996) and with the coretemperature of theCH3CN emission observed by Olmi et al.(1996b) but much higher than the peak brightness temperatureof 120 K from the optically thickNH3(4,4) line observed byCesaroni et al. (1998). The latter inconsistency could be due toan unresolved clumped source structure which would reduce theobserved brightness temperature of the ammonia observations.

The size of the IR source is found to be between 0.′′65 and1′′. This is about half of the size determined fromNH3(4,4) andground stateCH3CN emission but within the errors equal tothe size of vibrationally excitedCH3CN emission (Cesaroni etal. 1998 and Olmi et al. 1996b). From the interferometer resultswe found the source ofHC3N emission to be coincident withNH3(4,4) andCH3CN within the positional error of 1′′.

In this study many vibrationally excited states ofHC3Nhave been detected for the first time in interstellar space showingthe valuable interrelation between laboratory spectroscopy andastrophysics and the particular usefulness of this molecule intracing a wide range of rotationalandvibrational excitation.

Acknowledgements.We would like to thank Carsten Kramer for helpwith the 30-m, Helmut Wiesemeyer for help with the PdBI obser-vations and Frederic Gueth and Karl Menten for a critical readingof the manuscript. FW has been partially supported by the DeutscheForschungsgemeinschaft with grant number SFB 301. MW acknowl-edges travel support from CNR grant 97.00018.CT02 and ASI grant

Table 1.HNCO intensities in G10.47+0.03 observed with the 30-m

ν Trans. Vib. state Eup∫

TMB dv(MHz) (K) (K km s−1)

109871.952 524 − 423 v0 188.9 8.5109904.922 505 − 404 v0 15.8 14.3109495.200 515 − 414 v0 59.0 7.7110297.516 514 − 413 v0 59.2 9.0154414.181 716 − 615 v0 72.9 20.1

110066.104 54 − 44 v5 1539.8 0.3110084.368 505 − 404 v5 847.4 0.8110089.690 53 − 43 v5 1236.9 0.5110104.112 524 − 423 v5 1020.5 0.5110105.356 523 − 422 v5 1020.5 0.7154090.197 74 − 64 v5 1553.5 0.9154114.411 707 − 606 v5 861.1 3.7154123.323 73 − 63 v5 1250.6 2.9154142.560 726 − 625 v5 1034.2 3.3154146.055 725 − 624 v5 1034.2 3.8

110080.464 53 − 43 v6 1355.4 0.4110086.440 505 − 404 v6 965.9 0.5110164.245 514 − 413 v6 1009.3 0.6154110.119 73 − 63 v6 1369.1 2.1154119.019 707 − 606 v6 979.6 2.1154227.515 716 − 615 v6 1023.0 4.0

109870.278 515 − 414 v4 1178.0 blend109961.242 505 − 404 v4 1134.8 blend154828.282 716 − 615 v4 1191.9 0.7

Fig. A1. Boltzmann plot of all detected lines of vibrationally excitedHNCO. All intensities were corrected to match the 109 GHz beam byassuming a source size much smaller than the beams.

ARS-96-66. He also thanks the MPIfR for their hospitality during thetime this article was in preparation.

Appendix A: vibrationally excited HNCO

Toward G10.47+0.03, we report the first detection of vibra-tionally excited HNCO. Using the frequencies from labora-

F. Wyrowski et al.: Vibrationally excitedHC3N toward hot cores 895

tory measurements of Yamada (1977), a group of spectral linefeatures in our 3 and 2 mm measurements can be attributed toHNCO in itsv4, v5 andv6 vibrational states which lie between800 and 1100 K above ground (see Figs. 4 and 5). We checkedline catalogs for possible blends with lines from other speciesbut no blending lines were found. Integrated line intensitieswere derived from simultaneous Gaussian fits to all lines withthe sameJ and are given in Table 1. To analyze the excita-tion of the lines, we used the line strengths of the ground staterotational transitions for the vibrational excited states as well.The resulting Boltzmann plot is shown in Fig. A1 and yields atemperature of 380 K, which should be regarded as an upperlimit due to possible high optical depths of the lines with lowerenergies above ground and possible blends of high excitationlines with U-lines.

References

Blake G.A., Sutton E.C., Masson C.R., Phillips T.G., 1987, ApJ 315,621

Blake G.A., Mundy L.G., Carlstrom J.E., et al., 1996, ApJ 472, L49Cesaroni R., Churchwell E., Hofner P., Walmsley C.M., Kurtz S., 1994,

A&A 288, 903Cesaroni R., Hofner P., Walmsley C.M., Churchwell E., 1998, A&A

331, 709Churchwell E., Wood D., Myers P.C., Myers R.V., 1986, ApJ 305, 405Churchwell E., Walmsley C.M., Cesaroni R., 1990, A&AS 83, 119Churchwell E., 1991, In: Lada C.J., Kylafis N. (eds.) The Physics of

Star Formation and Early Stellar Evolution. NATO AsiClark F.O., Brown R.D., Godfrey P.D., Storey J.W.V., Johnson D.,

1976, ApJ 210, L139Dahmen G., Wilson T., Matteucci F., 1995, A&A 295, 194

Deguchi S., Nakada Y., Onaka T., 1979, Publ. Astron. Soc. Japan 31,105

Garay G., Rodriguez L.F., 1990, ApJ 362, 191Gaume R.A, Fey A.L, Claussen M.J, 1994, ApJ 432, 648Genzel R., 1992, In: Burton W.B., Elmegreen B.G., Genzel R. (eds.)

The Galactic Interstellar Medium, Springer, p. 275,Goldsmith P.F, Krotkov R., Snell R.L., 1982, ApJ 260, 147Goldsmith P.F, Krotkov R., Snell R.L., 1985, ApJ 299, 405Guilloteau S., et al., 1992, A&A 262, 624Hofner P., Churchwell E., 1996, A&AS 120, 283Hofner P., Wyrowski F., et al., 1998 (in prep.)Mallinson P.D., Fayt A., 1976, Molec. Phys. 32, 473Mallinson P.D., de Zafra R.L, 1978, Molec. Phys. 36, 827Mauersberger R., Henkel C., Wilson T.L., 1988, A&A 205, 235Millikan R.C., White D.R., 1963, J. Chem. Phys. 39, 3209Olmi L., Cesaroni R., Walmsley C.M., 1996a, A&A 307, 599Olmi L., Cesaroni R., Neri R., Walmsley C.M., 1996b, A&A 315, 565Snyder L.E., Buhl D., 1974, ApJ 189, L31Stutzki J., Genzel R., Graf U.U., Harris A.I., 1989, ApJ 340, L37Turner B.E., 1991, ApJS 76, 617Turrel G.C., Jones W.D., Maki A., 1957, J. Chem. Phys. 26, 1544Uyemura M., Maeda S., 1974, Bull. Chem. Soc. Japan 47, 2930Uyemura M., Deguchi S., Nakada Y., Onaka T., 1982, Bull. Chem.

Soc. Japan 55, 384de Vicente P., 1994, Ph.D. Thesis, Universidad Complutense de MadridWild 1995: The 30m Manual, IRAM Tech. Report 377/95Wood D.O.S., Handa T., Fukui Y., et al., 1988, ApJ 326, 884Wyrowski F., Walmsley C.M., 1996, A&A 314, 265Wyrowski F., Hofner P., Schilke P., et al., 1997, A&A 320, L17Wyrowski F., 1997, Ph.D. Thesis, University of CologneYamada K.M.T., 1977, J. Mol. Spectr. 68, 423Yamada K.M.T., Creswell R.A., 1986, J. Mol. Spectr. 116, 384Ziurys L.M., Turner B.E., 1986, ApJ 300, L19