2 Observations and data reduction

2.1 Observations

The observations were carried out between 1986 and 1997. The 2-mm SiS mixer receiver was generally optimized for the lower sideband (LSB) with an attenuation of the upper sideband (USB) between 3 and 20 dB. Rejections $\leq 6$ dB were used at the high end of the surveyed band ($\nu > 160$ GHz) where the receiver could not be tuned single-sideband (SSB), and at a number of frequencies where the receiver was found to be less stable. The image rejection was measured by injecting lines in the upper and lower sidebands; this procedure is accurate to $\sim$ 3 dB and could lead to significant calibration errors for low rejections of the image side band. Most observations were made in the balanced wobbler-switching mode, with a wobbling period of 0.5 Hz and a beam throw of $\pm 2'$. The telescope pointing and focus were checked every two hours through azimuth-elevation cross scans on the nearby continuum source OJ 287, or directly on IRC+10216 by observing centrally peaked lines, like SiS, and SiO. The system noise temperature varied between 250 K and 1000 K, depending on the observed frequency, the weather, and the source elevation.

The spectra in Figs. 1 and 2 are in units of antenna temperature, $T_{\rm A}^*$, corrected for atmospheric absorption and spillover losses (see, e.g., Cernicharo [1985]). $T_{\rm A}^*$ is related to the main beam-averaged brightness temperatures by the relation $T_{\rm A^*}=B_{\rm eff}/F_{\rm eff} T_{\rm MB}$, where $B_{\rm eff}/F_{\rm eff}$, the antenna main beam efficiency, varies between 0.65 at the low end of the spectral window and 0.55 at its high end. In view of the various calibration uncertainties discussed below, we have rounded off $B_{\rm eff}/F_{\rm eff}$ to 0.6 throughout the entire band.

Most spectra were observed with a filterbank consisting of 512 two-pole filters, with halfpower widths and spacings equal to 1.0 MHz. Half a dozen of spectra were observed with an autocorrelator or with an AOS spectrometer with half-power channel-widths of 1.25 MHz; a few strong lines were observed with a 100 kHz resolution. A baseline of order 1-3 was subtracted from each 0.5 GHz-wide spectrum, after all noticeable spectral lines had been blanked out.

2.2 Determination of the line parameters

The 2-mm spectral survey extends almost continuously between 129 GHz and 172.5 GHz. Because of the limited USB rejection, the raw spectra include lines originating from both receiver sidebands. The IF center frequency being 3.932 GHz, overlapping LSB and USB lines have frequencies differing by 7.8 $\pm$ 0.5 GHz -- the exact value depending on their location in the IF band. Since all spectra were observed with two or more settings of the local oscillator frequency, each line appeared at least twice in the IF band, so that its frequency could be unambiguously determined. Using our own molecular rotational transition catalog, which includes the transition frequencies of most molecules of astrophysical interest, we were able to identify the vast majority of the 380 lines we observed. One third of those were assigned to astrophysical species which were unknown at the time we started our survey. One sixth of the lines remain unidentified. The detected lines, ordered by frequency, and their assigments are listed in Table 2. The unidentified lines are summed up in Table 3.

Depending on their apparent shapes, the line profiles were fitted with the cusped, flat, or parabolic profiles expected from optically thin/thick lines arising in a uniformly expanding spherical shell (see Morris [1985]). The expansion velocity, $V_{\rm exp}$ and the systemic velocity $V_{\rm sys}$, were derived through fits to the strongest lines. They were found to be the same for all lines, $V_{\rm exp}= 14.5~\pm$ 0.2 km s^-1 and $V_{\rm sys}{\rm (LSR)}= -26.5\pm 0.3$ km s^-1, except in the cases of the vibrationally excited lines of SiS and CS which arise in the hot region close to the star.

Tables 4-9 present the fitted transitions and line parameters, ordered by molecular species and isotopomer. As explained in Sect. 2.1 the integrated main beam brightness temperatures, $T_{\rm MB}$, are related to antenna temperatures (the units of Figs. 1 and 2) through the formula $\int T_{\rm MB}{\rm d}v=F_{\rm eff}/B_{\rm eff} \int T_{\rm A}^* {\rm d}v$.

2.3 Calibration and removal of USB lines

In order to properly calibrate the intensity scale, many strong lines were re-observed with a high rejection of the image side band ($\geq 17$ dB). The intensity of these "primary'' calibrators was used to calibrate the lines >0.3 K appearing in the same spectra. Those lines, in turn, were used to calibrate more distant lines. Because of the large number of overlapping spectra, it was possible to calibrate with three such steps one 2-3 GHz-wide band around each "primary'' line. The intensities of the lines at the intersection of two such bands were typically found to be consistent within 10%. We note, however, that we have assumed a constant rejection of the image signal across the IF band. This may not be true at some frequencies, and could in exceptional cases lead to larger calibration errors. Finally, because of the higher atmospheric attenuation and of the low image side band rejection above 160 GHz, the calibration uncertainty raises to 10-25% between 160-172.5 GHz.

Because each portion of our spectral scan is the average of several LSB spectra, observed with different settings of the LO, any line from the USB improperly rejected will

Figure 2: The $\lambda$ 2 mm spectral survey of IRC+10216 at a full resolution. The spectral resolution is 1 MHz, except in the intervals marked by an horizontal dotted line, where it is 2 MHz (see text). Ordinate is $T_{\rm A}^*$, the antenna temperature corrected for atmosphere absorption and spillover losses. This latter is related for to the main beam-averaged brightness temperatures by the relation $T_{\rm A^*}=B_{\rm eff}/F_{\rm eff} T_{\rm MB}$, where $B_{\rm eff}/F_{\rm eff}$ is the antenna main beam efficiency. The spectra are drawn at the scale $-0.1< T_{\rm A}^*< 0.3$ K, except for those with peak temperatures between 0.3 and 0.5 K which are drawn at a slightly smaller scale (they are marked with a black square on the top right corner). In addition, the spectra with lines > 0.5 K are redrawn at full scale (marked with a dotted background). Note that most lines exhibit cusped profiles with full width 29.0 km s-1

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

Figure 2: continued

appear at several locations in the average spectra. The resulting ghosts of USB lines would then be difficult to trace out and identify. All traces from USB lines have thus to be removed prior to averaging the original spectra, which was done in the following way:

i) USB lines were first identified from a comparison of overlapping spectra observed with different LO settings and/or by an inspection of the portion of the spectral scan 8 GHz higher in frequency. ii) In the case of weak ghosts (i.e. when the USB lines were weak or the USB rejection was high), we fitted to the USB ghosts theoretical profiles, whose shape, frequency, and width were derived whenever possible from the high frequency spectra. Then iii) we subtracted the fitted profiles from the original spectrum.

iv) In the cases of strong USB ghosts, or of severe blending between USB and LSB lines, the channels covered by the USB ghost were simply blanked out.

The spectra processed in this way were averaged and combined yielding a first version of the 42 GHz-wide spectrum. Processes i) to iv) were then repeated to yield a second version, then a third version of the complete spectrum. We performed five such iterations to make sure that all remaining spectral features are lines from the signal sideband.

We note that the re-calibration of the spectra and the elimination of USB lines was greatly eased by the complete coverage of our survey: the line intensities we previously published from the analysis of individual spectra are a priori less reliable than those reported in this work.

2.4 Time variability

IRC+10216 is known to be a Mira-type variable with a periodicity of 1.71 yr. Its 10 $\mu$m flux varies by a factor of 2 between minimum and maximum. Because of the calibration method described above, it was important to check for time-related intensity variations. Depending on the relative importance of radiative and collisional excitation, the millimeter line intensities may or may not follow the infrared flux variations. Among the 2-mm lines, the most likely to be affected are: i) those of CS, HC₃N, SiO and SiS, four species whose IR lines are known to be optically thick, as well as ii) the vibrationally excited lines of C₄H and HCN (Lucas & Cernicharo [1989]). These lines were observed at several occasions during the 10 yr-long observing period. The ground-state mm lines, observed at a resolution of $\simeq 2$ km s^-1, were found to have stable shapes and intensities (within $20\%$ which is consistent with our calibration uncertainty). The v= 1, J=3-2 line of CS and several mm lines of vibrationally excited C₄H were observed with a good signal-to-noise ratio at different IR-phase periods in the course of our survey. We saw no intensity variations $> 20\%$ which could be correlated with the IR flux phase. However, the strong $\nu_2=1, J=2-1$ line of HCN near 177 GHz shows factor of 2 intensity variation with time; this line, however, is known to be masering (Lucas & Cernicharo [1989]).

Intensity comparisons are more difficult for weaker lines. We can only quote an upper limit of 20%, which represents the scatter of the intensities recorded at different epochs for the 0.3-0.5 K lines. Most of this scatter is probably related to calibration errors, since we found no obvious relation with the IR flux variations.