2.1. Object selection
The sample of 11 hot spot candidates observed in the present work (and listed in Table 1) includes most of the radio-bright sources accessible from the northern hemisphere for which deep optical imaging and high-resolution radio maps have already been obtained either by ourselves or other authors. Optical polarization data, and millimetre wavelength data also exist for some of the sources. It does not however constitute a statistically complete sample.
Table 1. Hot spots observed (coordinates refer to the peak of the radio emission).
The hot spots in 3C 33 S and 3C 111 E were both observed and detected in the infrared at UKIRT in 1986 with a single aperture photometer (as described in Paper I) but it was important to check these results with imaging, given that in both cases bright contaminating stars lie only five arcsec from the optical counterpart to the hot spot. Those earlier data were obtained with a five arcsec aperture (the smallest aperture that one may practically use with the photometer at UKIRT) and it was necessary to model and subtract the flux from the contaminating stars in each case in order to derive an accurate measure of the infrared hot spot flux. The hot spot in 3C 20 W was observed by Hiltner et al. (1994) to be polarized in the optical but was not observed in the infrared in 1986 and so was a prime candidate for imaging in the infrared. A one-dimensional K -band scan of the hot spot in 3C123E was obtained with the aperture photometer at UKIRT in 1986 resulting in a tentative detection (Paper I) and it was thus of great interest to improve on this result with imaging. Given the largely improved limiting brightness attainable with infrared imaging and the steepness of most hot spot spectra in the infrared-optical regime makes searching for infrared counterparts a reasonable proposition, even in the absence of an optical candidate. For instance, there is a hot spot in the eastern lobe of 3C 20 without an optical candidate and this was observed in the infrared along with that in the western lobe. In addition, infrared detections can provide a strong constraint on the location of any cut-offs in the spectra, even if the faintness of an optical counterpart prohibits the determination of a good optical spectral index. Thus, the remaining candidates were chosen mainly on the grounds of high radio surface brightness, the availability of good radio mapping, and supplemented in most cases by the availability of deep optical imaging which suggested the presence of a possible optical counterpart to the hot spot. The radio hot spots of Cygnus A (3C 405) have been the subject of extensive imaging and polarimetry campaigns by the present authors. Despite the proximity and high radio power of this radio galaxy, attempts to locate optical candidates for the hot spots have been difficult. In the western lobe, our imaging (and that of earlier workers) suggests possible optical counterparts to hot spots A & B (in the nomenclature of Hargrave & Ryle 1974), but in the eastern lobe (hot spot D in the same nomenclature) no optical counterpart has so far been located. The hot spot in the western lobe of the well studied source 3C 303 has a diffuse optical counterpart (Kronberg 1976), although this identification has not been confirmed by a polarization measurement yet. 3C 65 is an extremely distant radio source () but it has two hot spots which are very bright in the radio, and this makes an attempt at finding an infrared counterpart worthwhile.
2.2. Infrared imaging
The hot spots listed in Table 1 were imaged with IRCAM in the K -band at the United Kingdom Infrared Telescope (UKIRT) during two observing runs in 1989. IRCAM employed an SBRC Indium-Antimonide detector with 58x62 pixels, and for these observations a pixel scale of 0.62 arcsec per pixel was used (resulting in a field of view of about 30 arcsec on a side). Each hot spot field was imaged by obtaining a number of successive integrations of 300s duration, and offsetting the telescope by a few arcsecs in a random direction after each 300s exposure so as to aid in the construction of a flat field using a superposition of the science frames. Most of our fields were uncrowded enough to make this technique feasible, and it eliminates the necessity for separate sky exposures. It also provides much better sky subtraction than individual sky frames, possibly separated from the data frames themselves by a time interval that is much greater than the typical timescale over which significant variations in the infrared sky occur.
Table 1 shows the total integration times obtained for the observed hot spots. It is worth noting in connection with the choice of on-chip integration time used here that subsequent experience by other observers using this and other infrared cameras of similar type (e.g. Cowie et al. 1990) indicates that the optimal integration time for each individual exposure (in order to provide the best estimate of the sky background) should be a little longer than the time needed to reach the sky-limited regime, that is about 80s in the K -band for this particular camera and telescope. Despite the longer than optimum on-chip exposure time employed in the current work, our reduced images are generally flatter than one part in when measured on a pixel-to-pixel basis.
All of the data were obtained during photometric conditions, apart from the latter half of the night of 1989 June 10 which affected the image of the two hot spots in the western lobe of 3C405 (hot spots A and B). This non-photometric image was subsequently calibrated with a 560s calibration frame obtained during photometric weather at UKIRT on 1991 July 10 (and provided to us courtesy of the Service Observations scheme). This calibration frame was obtained with exactly the same instrumentation as our deeper imaging carried out in 1989, and was constructed in the same manner (i.e. from a series of shifted, short-exposure frames). All of our data were taken at airmasses less than 1.3, and standard stars were observed every night.
2.3. Data reduction
The dark current and bias level was subtracted from each of the individual exposures using suitably scaled dark frames obtained each night (a standard non-linearity correction had already been made to the data). Permanently bad pixels were then corrected by interpolation using a map of the known bad-pixel positions. Sky frames were constructed for each source using the individual 300s exposures (and using only the data for that source rather than data obtained earlier or later in the night). The median of the pixel values at each given pixel position was calculated and a median sky frame constructed from these values. In a second step, each individual frame is compared with this median and all pixels more discrepant than a certain threshold (i.e. the pixel illuminated by objects) are flagged. A new median is calculated by ignoring the flagged pixels. This process is iterated if necessary. Finally the un-flagged pixel values are averaged to obtain the optimum blank sky frame. The data frames were then flattened by division of a normalised version of this sky frame. Any remaining bad or hot pixels were then dealt with before stacking the individual frames. The data for each source were stacked after a rebinning process which simultaneously accounts for the different registration and variation of the seeing point-spread-function (PSF) of the individual frames. Offsets and PSF have been determined from bright stars in each field.
Fig. 1 shows grey scales of the hot spots imaged. The field of view which received the full integration time (and is shown here) is smaller than the nominal field provided by the chip because of the shifting of the individual exposures. The hot spot identifications and detections are discussed in Section 3 below.
Table 2 presents infrared photometry of the detected hot spot candidates and limiting magnitudes in the cases where no detection was made. The calibration was done with respect to standard stars observed in the same night and should be accurate to better than 10%. We give the photometric results both un-corrected and corrected for galactic extinction. Galactic extinction estimates () were taken either from Paper I or from the HI maps of Burstein & Heiles (1982) with the assumption that (Rieke & Lebofsky 1985). 3C405 (Cygnus A) lies at low Galactic latitude and the reddening is not accurately known. We have derived a mean value for from the estimates of van den Bergh (1976), Yee & Oke (1978) and Spinrad & Stauffer (1982), which are 0.39, 0.50 and 0.36 respectively. For the absolute flux calibration we assume a Vega flux Jy (as in Paper I). Our photometry generally refers to a 5 arcsec diameter aperture, except in cases where there are close contaminating objects in which case a smaller aperture is used. In the case of 3C 303 west the infrared emission appears to be extended: the value shown in Table 2 was obtained using a 4 arcsec diameter aperture centered on the brightest part of the optical emission. The optical and infrared candidate for hot spot A of 3C 405 is another problematic object in that it is merged with a nearby star. Rather than attempt to model (probably erroneously) the star's contribution to the infrared hot spot flux (e.g. by modelling the star with a Gaussian) we show instead in Table 2 the infrared flux obtained with a 1.75 arcsec diameter aperture centred on the hot spot's radio position. Since we pursued the same procedure on the optical images in order to get the flux values given in Table 3, the NIR-optical spectrum should be correct while the error of the absolute flux may easily exceed 20%.
Table 2. K -band photometry of hotspots.
Table 3. Optical photometry of hot spots not contained in Paper I.
Errors are calculated from the error associated with the estimate of the background noise. Limiting fluxes are detection limits evaluated for a 2 arcsec aperture. The wide variation in limiting magnitudes is largely a result of the variable success in deriving a good sky-flat for each source. For example, the median sky technique used here was less successful for the source 3C123E where a large fraction of all frames is covered by the radio galaxy, thus a brighter limiting magnitude was obtained.
© European Southern Observatory (ESO) 1997
Online publication: May 5, 1998