2 Observations and data reduction

2.1 Observations

  

Figure 1: Spectra observed during the 330-360 GHz survey of G5.89-0.39. The frequency scale given for each block of spectra is the lower sideband scale, the upper sideband scale is omitted as it is not continuous for the blocks of concatenated spectra. Species identifications are shown for each line and upper sideband lines are indicated by brackets

 

Figure 1: continued

The observations were made with the James Clerk Maxwell Telescope (JCMT) between the 8th and 14th March 1996. All observations were made at the coordinates $\alpha(1950)$ = $17^{\rm h}$ $57^{\rm m}$ $27^{\rm s}$ and $\delta(1950)$ = -24$^\circ$ 03$^\prime$ 57$^{\prime\prime}$, which is the estimated centre of the UC HII region (Wood & Churchwell 1989). The pointing accuracy of the telescope was checked regularly against the peak continuum position of the UC HII region G34.26+0.15 and was found to be good to within 5$^{\prime\prime}$. It was found that beam-switching (i.e. chopping the secondary mirror from on-source to off-source) was much superior to position switching for obtaining extremely flat baselines. A chop throw of 3$^\prime$ in RA was used to keep a constant reference position, with a chopping frequency of 1 Hz. 3$^\prime$ was more than sufficient to avoid contamination in the reference position for all species except CO (as can be seen in Fig. 1).

To cover the frequency range of the survey the 345 GHz SIS junction receiver B3i (RxB3i) was used in conjunction with the Dutch Autocorrelation Spectrometer (DAS). The DAS was used in 760 MHz bandwidth mode and with RxB3i as a frontend produces dual sideband spectra. Dual sideband spectra comprise two frequency bands (the upper and lower sidebands) folded over one another to produce a composite spectra. The upper and lower sidebands are separated in frequency by approximately twice the local oscillator intermediate frequency (IF), depending on the doppler correction for the source velocity. The upper sideband frequency scale is reversed relative to the lower sideband scale. The velocity of G5.89-0.39 with respect to the Local Standard of Rest ($V_{\rm LSR}$) was assumed to be +9.4 $\,{\rm km\,s}^{-1}$. For RxB3i the IF is 1.5 GHz and the upper and lower sidebands are separated by approximately 3 GHz. Each spectrum taken thus represents a total frequency range of $\sim \! 1.5$ GHz and this was used to reduce the total number of spectra needed to cover the frequency range of the survey.

The spectra were all observed with the "main band'' set to the lower sideband, which means that the other (upper) sideband covers a frequency range of the same width roughly 3 GHz higher in frequency. We took spectra with their central frequency incremented by 700 MHz (ensuring an overlap of 30 MHz between spectra) until the lower sideband had covered the first 2.8 GHz of the frequency range. The upper sidebands of these spectra cover the next 2.8 GHz of the frequency range with a 200 MHz gap in coverage. This block of 4 spectra thus covers a total frequency range of 5.6 GHz. The remaining parts of the frequency range were observed in the same manner. The 200 MHz gaps between the blocks of spectra were to be covered by additional spectra taken at the end of the observing run, however due to bad weather this was not achieved. These gaps do not contain many lines of significance; searches of spectral line catalogues (both of predicted and observed lines) indicate that few lines occur in these frequency ranges. The blocks of spectra (with individual spectra concatenated) are shown in Fig. 1.

Two problems inherent in dual sideband spectra are the allocation of features to a particular sideband (i.e. upper or lower) and the possible overlapping (blending) of lines from each sideband. To determine the sidebands (and hence line frequencies) extra spectra with a local oscillator shift of +10 MHz were taken. In the shifted spectra lines in the upper sideband will appear to shift frequency by 20 MHz relative to those in the lower sideband. Blended lines from both sidebands were separated by this technique whenever possible.

With the DAS in 760 MHz mode the spectral resolution is 0.756 MHz. Each spectrum was divided into channels of 0.625 MHz, although later in the data reduction process all spectra were binned to a channel width of 1.25 MHz to improve signal to noise. The standard chopper-wheel calibration method of Kutner & Ulich (1981) was used to obtain line temperatures on the $T_{\rm A}^{*}$ scale, i.e. corrected for the atmosphere, resistive telescope losses and rearward spillover and scattering. $T_{\rm A}^{*}$ can also be corrected for forward spillover and scattering to give the corrected receiver temperature $T_{\rm R}^{*}$ where $T_{\rm R}^{*}$ = $T_{\rm A}^{*}$/$\eta_{\rm fss}$ and $\eta_{\rm fss}$ is the forward spillover and scattering efficiency (0.7 for RxB3i at 345 GHz).

2.2 Data reduction and line identification

The data were reduced using the Starlink spectral line package SPECX. Linear baselines were subtracted from the spectra and the line parameters of peak temperature ($T_{\rm A}^{*}$), central frequency ($\nu({\rm obs})$) and line width at half maximum ($\Delta \nu_{1/2}$) were measured. Values for the noise in the spectra were evaluated using line-free channels and the typical rms noise level was found to be $\sigma \sim\! 0.1$ K at a spectral resolution of 1.25 MHz. Features below a detection limit of 5$\sigma$ were ignored to avoid inaccurate line identifications. These data are given in Table 5 for each line. Multiple independent detections of each line (i.e. including the 10 MHz shifted spectra) are listed.

The identification of molecular lines detected in this survey was achieved by comparing their central frequencies primarily with the JPL spectral line database (Poynter & Pickett 1985). Other lists used include Lovas (1992), the methanol lists of Anderson et al. (1993) and the observational lists of lines seen by Jewell et al. (1989) toward Orion-A, Schilke et al. (1997b) toward Orion-KL and Macdonald et al. (1996) toward G34.26+0.15.