The long slit observations were conducted during two runs in April and October 1994 at the equivalent 4.5 meter Multi-Mirror Telescope on Mt Hopkins, Arizona. The "Blue Channel" spectrograph was used with a Loral 3k1k CCD. A grating with 500 grooves/mm and a blaze at 5410 Å was employed; the spectral range covered in the reduced spectra is about 3500 Å (3500 7000) with a dispersion of 1.17 Å/pixel. An order sorting filter was used (UV-36) and all the data were obtained with the CCD binned by a factor of two in the spatial direction (0.6"/pixel). All these observations were carried out with an unvignetted 2"150" slit positioned close to the parallactic angle to avoid any light loss. Three exposures of 15 or 20 minutes were obtained for each slit position in the bar. Between exposures, the alignment of the MMT six mirrors was verified and corrected, if necessary. Numerous standard stars were observed with a larger slit (5") during the night for the flux calibration procedure.
Table 1 presents the journal of observations for the H ii regions of ten of the eleven objects studied by MF97 (NGC 5068 was not observed). For some objects, a few slit positions were required to optimize the number of H ii regions observed in the bar and the disc. Average seeing during these observations was about 1.2" and conditions were photometric. Note that the only galaxy in the sample considered as an AGN (LINER) is NGC 7479.
Table 1. Journal of observations.
2.2. Data reduction and analysis
The long slit spectra were reduced following standard procedures available in the LONGSLIT package in IRAF 1. First, a bias and flat-field correction was applied. The illumination pattern along the slit was corrected using a set of sky flat-fields taken in the same optical configuration. Wavelength calibration was done using a Helium-Neon-Argon exposure taken immediately after the science observation. Geometric distortion and alignment of the spectra were corrected by 2D-mapping of the spectral lines from the calibration sources using the FITCOORDS and TRANSFORM algorithms.
The next step, sky subtraction, was done by extracting a sky background using sections along the long slit outside the galaxy. To make sure that the signal from the galaxy was minimized, the average sky backgrounds were compared between the different slit positions since most of the slits used for the disc H ii regions were less contaminated by the disc emission. The spectra were then flux calibrated using a sensitivity function obtained from a set of standard stars observed through the nights. Extraction of the spectrum for each H ii region was done using the spatial profile at H seen along slit. In general, no continuum trace was found for these H ii regions. The extraction trace was forced across the spectral range using the positions of the main nebular lines along the spectral domain. Finally, the individual spectra were combined.
The integrated fluxes of the main nebular lines [O ii] 3727, H, [O iii] 4959, 5007, [O i] 6300, [N ii] 6584, H and [S ii] 6717, 6731 were measured using a Gaussian fitting algorithm available with SPLOT in IRAF . A background continuum estimated from each side of the lines was automatically subtracted.
Many H ii regions in bars show a strong underlying Balmer absorption at H. McCall et al. (1985) have shown that adding about 2 Å of equivalent width for normal disc regions constitutes an appropriate correction. This correction was applied for all the H fluxes for the H ii regions in our control sample and the bar regions with shallow underlying absorption. However, for about 20% of the bar sample, this correction was not sufficient. For these regions, mostly located in the bars with a strong continuum (e.g. NGC 3504, NGC 5921, NGC 7479), the amplitude of the underlying absorption was estimated from a Gaussian fit. In general, we found that about 5 Å of equivalent width were necessary to assure a good correction. This value was applied for all bar regions with absorption higher than normal.
All of the line fluxes were corrected for interstellar reddening by comparing the H/H ratio to the theoretical Balmer decrement (2.86) for Case B recombination (that is, for nebulae with large optical depths for the H I resonance lines) at K. In reality, the temperature of bar H ii regions is probably around 7000-8000 K due to their high O/H abundances (see below) so that our extinction based on the Balmer decrement might be overestimated by about 0.1 mag. Since the Balmer ratio represents only an approximation of the real extinction, we did not take this difference into account. The reddening law formulated by Savage & Mathis (1979) was assumed for reddening correction.
It is generally difficult to evaluate the accuracy of the absolute spectrophotometric fluxes for individual objects. The main uncertainties include the contamination by the bright galactic continuum, the underlying absorption at H, the accuracy of the photometric calibration, the spectrum extraction in a crowded field of H ii regions, the positioning of the slit, the correction for the interstellar extinction and, of course, the intensity of the lines. We evaluate the accuracy of the fluxes to be at about 20-30%. Since we will only compare the global behavior of H ii regions, these high uncertainties should not influence our conclusions.
© European Southern Observatory (ESO) 1999
Online publication: June 17, 1999